(Behram N. Kursunoglu (Auth.), Behram N. Kursunogl

Neutrino Mass, Dark Matter,
Gravitational Waves,
Monopole Condensation,
and Light Cone Quantization
Neutrino Mass, Dark Matter,
Gravitational Waves,
Monopole Condensation,
and Light Cone Quantization
Edited by
Behram N. Kursunoglu
Global Foundations, Inc.
Coral Gables, Florida
Stephan L. Mintz
Florida International University
Miami, Florida
and
Arnold Perlmutter
University of Miami
Springer Science+Business Media, LLC

L i b r a r y of Congress C a t a l o g 1 n g - 1 n - P u b l t c a t l o n Data
Neutrino mass, dark natter, gravitational waves, monopole

condensation, and light cone quantization / edited by B e h r a m N.
Kursunoglu., Stephan L. Mtntz and A r n o l d Perlmutter.
p. cm.
"Proceedings of the International Conference on O r b l s Scientlae
1996 f o c u s i n g on n e u t r i n o nass, dark matter, gravlatlonal waves,
condensation of atoms and monopoles, light cone quantization, held
January 25-28, 1996, 1n M i a m i Beach, Florida." T.p. verso.
Includes bibliographical references and index.
ISBN 978-1-4899-1566-5
1. P h y s i c s C o n g r e s s e s . 2. A s t r o p h y s i c s C o n g r e s s e s .
3. P a r t i c l e s (Nuclear p h y s i c s ) C o n g r e s s e s . 4 . Dark m a t t e r
(Astronomy)Congresses. 5. N e u t r i n o s M a s s C o n g r e s s e s .
I. K u r s u n o g l u , B e h r a m , 1922- . II. M 1 n t z , S t e p h a n L.
III. P e r l m u t t e r , A r n o l d , 1928- . I V . I n t e r n a t i o n a l C o n f e r e n c e on
O r b i s S c i e n t l a e (1996 : Miami B e a c h . Fla.)
QC1.N487 1996
530dc20 96-43713
CIP
Proceedings of the International Conference on Orbis Scientiae 1996, focusing on Neutrino Mass, Dark
Matter, Gravitational Waves, Condensation of Atoms and Monopoles, Light Cone Quantization, held
January 2 5 - 2 8 , 1996, in Miami Beach, Florida
This volume was taken from a series of conferences sponsored by Global Foundation, Inc.,
ISBN 978-1-4899-1566-5 ISBN 978-1-4899-1564-1 (eBook)

DOI 10.1007/978-1-4899-1564-1
1996 Springer Science+Business Media New York
Originally published by Plenum Press, New York in 1996
Softcover reprint of the hardcover 1st edition 1996
All rights reserved
10 9 8 7 6 5 4 3 2 1
No part of this book may be reproduced, stored in a retrieval system, or transmitted in any form or by any
means, electronic, mechanical, photocopying, microfilming, recording, or otherwise, without written
permission from the Publisher
PREFACE
The International Conference, Orbis Scientiae 1996, focused on the topics: The
Neutrino Mass, Light Cone Quantization, Monopole Condensation, Dark Matter, and
Gravitational Waves which we have adopted as the title of these proceedings. Was there any
exciting news at the conference? Maybe, it depends on who answers the question. There
was an almost unanimous agreement on the overall success of the conference as was
evidenced by the fact that in the after-dinner remarks by one of us (BNK) the suggestion of
organizing the conference on a biannual basis was presented but not accepted: the
participants wanted the continuation of the tradition to convene annually. We shall, of
course, comply.
The expected observation of gravitational waves will constitute the most exciting
vindication of Einstein's general relativity. This subject is attracting the attention of the
experimentalists and theorists alike. We hope that by the first decade of the third millennium
or earlier, gravitational waves will be detected, opening the way for a search for gravitons
somewhere in the universe, presumably through the observations in the CMBR. The
theoretical basis of the graviton search will take us to quantum gravity and eventually to the
modification of general relativity to include the Planck scale behavior of gravity - at energies
of the order of 1019GeV.
We were very pleased to welcome the 1995 Nobel Laureate Frederick Reines to the
Orbis Scientiae 1996, who moderated the conference session on neutrino masses. Professor
Reines has been an enthusiastic participant of the Coral Gables Conferences, and in 1980 was
awarded the J. Robert Oppenheimer Memorial Prize. We preceded the Nobel Committee!
The Trustees and Chairman of the Global Foundation wish to extend special thanks
to Edward Bacinich of Alpha Omega Research Foundation for his generous support of the
1996 Orbis Scientiae.
Behram Kursunoglu
Stephan L. Mintz
Arnold Perlmutter
v
About The Global Foundation, Inc.
The Global Foundation, Inc., utilizes the world's most important resource ... people.
The Foundation consists of distinguished men and women of science and learning, and of
outstanding achievers and entrepreneurs from industry, governments, and international
organizations, along with promising and enthusiastic young people. These people convene to
form a unique and distinguished interdisciplinary entity to address global issues requiring
global solutions and to work on the frontier problems of science.
Global Foundation Board of Trustees

Behram N. Kursunoglu, Global Foundation, Inc., Chairman of the Board, Coral Gables
M. Jean Couture, Former Secretary of Energy of France, Paris
Manfred Eigen*, Max-Planck-Institut, Gottingen
Robert Herman, University of Texas at Austin
Willis E. Lamb*, Jr., University of Arizona
Walter Charles Marshall, Lord Marshall of Goring, London
Louis Neel*, Universite de Gronoble, France
Frederick Reines *, University of California at Irvine
Abdus Salam*, International Centre for Theoretical Physics, Trieste
Glenn T. Seaborg*, Lawrence Berkeley Laboratory
Henry King Stanford, President Emeritus, Universities of Miami and Georgia
*Nobel Laureate
vi
Global Foundation's Recent Conference Proceedings
Making the Market Right for the Efficient Use of Energy
Edited by: Behram N. Kursunoglu
Nova Science Publishers, Inc., New York, 1992
Unified Symmetry in the Small and in the Large

Edited by: Behram N. Kursunoglu and Arnold Perlmutter
Nova Science Publishers, Inc., New York, 1993
Unified Symmetry in the Small and in the Large 1

Edited by: Behram N. Kursunoglu, Stephen Mintz, and Arnold Perlmutter
Plenum Press, 1994
Unified Symmetry in the Small and in the Large 2

Plenum Press, 1995
Global Energy Demand in Transition: The New Role of Electricity

Plenum Press, 1996
Economics and Politics of Energy

Plenum Press, 1996
Neutrino Mass, Dark Matter, Gravitational Waves, Condensation Of

Atoms And Monopoles, Light Cone Quantization
Plenum Press, 1996
Contributing Co-Sponsors of the Global Foundation

Conferences
Gas Research Institute, Washington, DC

General Electric Company, San Jose, California
Electric Power Research Institute, Palo Alto, California
Northrop Grumman Aerospace Company, Bethpage, New York
Martin Marietta Astronautics Group, Denver, Colorado
Black and Veatch Company, Kansas City, Missouri
Bechtel Power Corporation, Gaithersburg, Maryland
ABB Combustion Engineering, Windsor, Connecticut
BellSouth Corporation, Atlanta, Georgia
National Science Foundation
United States Department of Energy
vii
CONFERENCE PROGRAM
INTERNATIONAL CONFERENCE
ON
ORBIS SCIENTIAE 1996
(NEUTRINO MASS, DARK MATTER, GRAVITATIONAL WAVES,

CONDENSATION OF ATOMS AND MONOPOLES, LIGHT CONE QUANTIZATION )
(24TH IN A SERIES OF CORAL GABLES CONFERENCES ON

ELEMENTARY PARTICLE PHYSICS AND COSMOLOGY SINCE 1964)
JANUARY 25.28, 1996
EDEN Roc RESORT AND SPA
PROGRAM
THURSDAY.January 25.1996 (Cotillion Ballroom)

8:00 AM Noon REGISTRATION at the entrance of the Cotillion Ballroom
1:30 PM SESSION I: PROLOGUE

Moderator: BEHRAM N. KURSUNOGLU, Global Foundation, Inc.,
Coral Gables
Dissertators: BEHRAM N. KURSUNOGLU
"Creation of Matter via Condensation at Absolute Zero and
Planck-Scale Temperatures"
C.W. KIM, Johns Hopkins University
"Scale Dependent Cosmology for an Inhomogeneous
Universe"
KAZUHIKO NISHUIMA, Chuo University, Tokyo
"Unbroken Non-Abelian Gauge Symmetry and Confinement"
Annotators: ALAN KRISCH, University of Michigan
JOSEPH LANNUTTI, Florida State University
LARRY RATNER, University of Michigan
Session Organizer: BEHRAM N. KURSUNOGLU
3:30 PM Coffee Break
ix
3:45 PM SESSION II: INSPIRATIONS FROM COSMOLOGY AND
ELEMENTARY PARTICLE PHYSICS
Moderators: KATHERINE FREESE, University of Michigan
ROBERT HERMAN, University of Texas
Dissertators: KATHERINE FREESE
"Inflationary Cosmology: From Theory to Observation and
Back"
HARRISON PROSPER, Florida State University
"Bayesian Analysis of Solar Neutrino Data"
EDWIN L. TURNER, Princeton University Observatory
"Do the Cosmological Parameters have Natural Values?"
Annotators: ANDREW HECKLER, Fermilab
Session Organizer: KATHERINE FREESE
5:00 PM SESSION III: PROGRESS ON SOME NEW AND OLD IDEAS I

Moderator: KAZUHIKO NISHUIMA, Chuo University, Tokyo
GEORGE SUDARSHAN, Center for Particle Physics,
University of Texas
Dissertators: VERNON BARGER, University of Wisconsin
"Fixed Points in Supersymmetry: R-Parity-Violating Yukawa
Couplings"
GERALD B. CLEAVER, Ohio State University, Columbus
"Grand Unified Theories from Superstrings"
VASKEN HAGOPIAN, Florida State University
"Capability of Future CMS Detector at the LHC Searching for
Dark Matter"
FREYDOON MANSOURI, University of Cincinnati
"Supersymmetric Wilson Loops and their Stringy Extensions"
KATSUMI TANAKA, Ohio State University, Columbus
"Comments on the Symmetry Breaking Terms in the Quark
Mass Matrix"
YUN WANG, Fermilab, Batavia,illinois
"Statistics of Extreme Gravitational Lensing Events"
Annotators: RICHARD ARNOWITT, Texas A & M University
Session Organizer: Dissertators
7:30 PM Orbis Scientiae adjourns for the day
FRIDAY, .January 26 1995 (Mona Lisa)

8:30 AM SESSION IV: GRAVITATIONAL WAVES
Moderator: SYDNEY MESHKOV, Cal. Tech.
Dissertators: BARRY BARISH, Cal. Tech.
"Status of LIGO"
x
SAMUEL FINN, Northwestern University
"Binary Inspiral, Gravitational Radiation, and Cosmology"
PETER FRITSCHEL, M.I.T.
"Interferometry for Gravity Wave Detection"
Annotators: RICHARD P. WOODARD, University of Florida
EDWARD KOLB, Fermilab
Session Organizer: SYDNEY MESHKOV, CALTECH
10:15 AM Coffee Break
10:30 AM ROUND TABLE DISCUSSION of Gravitational Waves

Moderator: BARRY BARISH, Cal. Tech.
Round Table
Dissertators: SAMUEL FINN, PETER FRITSCHEL, PETER
SAULSON
Noon Lunch Break
1:30 PM SESSION V: NEUTRINO MASSES

Moderator: FREDERICK REINES, University of California, Irvine
Dissertators: MAURY GOODMAN, Argonne National Laboratory,
Argonne
"Oscillation Searches Using Atmospheric Neutrinos and Long
Baseline Neutrino Experiments"
C. W. KIM, Johns Hopkins Univ., Baltimore
"The Role of the Third Generation in the Analysis of
Oscillation Experiments"
WILLIAM LOUIS, Los Alamos National Laboratory
"Ongoing Neutrino Oscillation Searches at Accelerators"
RABIADREA MOHAPATRA, Univ. of Maryland, College
Park
"Neutrino Mass Textures and New Physics Implied by Present
Neutrino Data"
NEVILLE REAY, Kansas State Univ., Manhattan
"Fermilab MUNI Project"
JOHN WILKERSON, Univ. of Washington, Seattle
"Solar Neutrino Measurements: Current Status and Future
Experiments"
LINCOLN WOLFENSTEIN, Carnegie-Mellon Univ.,
Pittsburgh
"Theoretical Ideas About Neutrino Masses"
Annotators: JEREMY MARGULIES, Los Alamos National Laboratory
HARRISON PROSPER
Session Organizer: STEPHAN MINTZ, Florida International University
3:30 PM Coffee Break
xi
3:45 PM ROUND TABLE DISCUSSION of Neutrino Masses by the Above
Dissertators
Moderator: WILLIAM LOUIS, Los Alamos National Laboratory
5:30 PM SESSION VI: STRINGS AND FIELD THEORY

Moderator: LOUISE DOLAN, University of North Carolina
Dissertators: LOUISE DOLAN, Department of Physics, University of
North Carolina
"BPS States and Type II Superstrings"
BRIAN GREENE, Department of Physics, Cornell University
"Changing the Topology of the Universe"
RENAT A KALLOSH, Department of Physics, Stanford
University
"F and H Monopoles"
JEFFREY MANDULA, DOE, Washington D.C.
Annotators: GERALD B. CLEAVER, Ohio State University, Columbus
Session Organizer: LOUISE DOLAN
SATURDAY, January 27,1996 (Mona Lisa)

8:30 AM SESSION VII: DIRAC'S LEGACY: LIGHT- CONE QUANTIZATION
Moderator: STANLEY BRODSKY, SLAC
Dissertators: STEPHEN PINSKY, Ohio State Univ., Columbus
"Introduction: Dirac's Legacy"
STAN BRODSKY, SLAC

"Applications of Light-Cone Quantization"
HANS-CHRISTIAN PAULI, Max Plank Institute,
Heidelberg
"Discrete Light-Cone Quantization"
10:00 AM Coffee Break
10: 15 AM Dissertator Presentations continue

Dissertators: ALEX KALLONIA TIS, University of Erlangen-Nurnberg,
Erlangen
"2-D Non-Perturbative Light-Cone Results"
DAVE ROBERTSON, Ohio State University
"The Light-Cone Gauge and Zero Modes"
BRETT VAN DE SANDE, Max-Planck Institute, Heidelberg
"Tube Model Solutions of QCD"
Annotators: ZACHARY GURALNIK, Princeton University
Session Organizer: STEPHEN PINSKY
xii
12:00 PM Lunch Break
1:00 PM SESSION VIII: THE MATTER OF DARK MATTER

Moderator: EDW ARD KOLB, FNAL, Chicago
Dissertators: RICHARD ARNOWITT, Texas A&M University
"SUSY Dark Matter with Non-Universal Soft Breaking
Masses"
SHARON HAGOPIAN, Florida State University
"Search for SUSY in the DO Collider Experiment"
ANDREW HECKLER, Fennilab
"On the Fonnation of a Hawking-Radiation Photoshpere: The
Cloak Around Microscopic Black Holes"
EDW ARD KOLB, Fennilab
"Light Photinos as Dark Matter"
IGOR TKACHEV, Ohio State Univ.
"Primordial Axions Appearing as Dark Matter and Other
Astrophysical Objects"
Annotators: VASKEN HAGOPIAN
Session Organizer: EDW ARD KOLB
3:30PM Coffee Break
3:45 PM SESSION IX: PROGRESS ON SOME NEW AND OLD IDEAS II

Moderators: FRED ZACHARIASEN, CALTECH
Dissertators: PRAN NATH, Institute for Theoretical Physics, Santa
Barbara, CA
"Superunification and Planck Scale Interactions"
MARK SAMUEL, Oklahoma State University
"Going to Higher Order - The Hard Way and the Easy Way:
The Agony and the Ecstasy"
INA SARCEVIC, University of Arizona
"Domain Structure of a Disoriented Chiral Condensate from a
Wavelet Perspective"
RICHARD P. WOODARD, University of Florida
"Quantum Gravity Slows Inflation"
Annotators: GERALD GURALNIK, Brown University
Session Organizer: SESSION DISSERTATORS
7:30 PM Conference Banquet - MONA LISA ROOM
SUNDAY. January 28. 1996 (Key Biscayne Room)

8:30 AM SESSION X: PROGRESS ON SOME NEW AND OLD IDEAS - III
Moderator: DON B. LICHTENBERG, Indiana University
xiii
Dissertators: DON B. LICHTENBERG, Indiana University
"Superflavor Symmetry and Relations Between Meson and
Baryon Masses"
PAUL H. FRAMPTON, University of North Carolina at
Chapel Hill
"Constraining a(Ma) From the Hidden Sector"
LUCA MEZINCESCU & RAFAEL NEPOMECHIE,
University of Miami
"Integrable Systems with Boundaries"
GREGORY TARLE, University of Michigan
"Cosmic Ray Signatures for Neutralinos: New Measurements
and their Implications"
Annotators: G. BHAMA THI, University of Texas at Austin
Session Organizer: DISSERTATORS
10: 15 AM Coffee Break
10:30 AM SESSION XI: EXACTLY SOLVABLE QUANTUM MODELS

Moderator: ANDRE LeCLAIR, Newman Laboratory, Ithaca, New York
Dissertators: PAUL FENDLEY, University of Southern California, Los
Angeles
"Two-Dimensional Field Theory Meets Experiment"
SERGEI LUKY ANOV, Newman Laboratory, Cornell
University
"Yang-Baxter Equation and Baxter's Q-Operators in CFT"
GIUSEPPE MUSSARDO, Scuola Internationale Superiore
di Studi Avanzati, Trieste, Italy
"Form Factor Approach to Integrable Quantum Field Theory:
The spin-spin correlation function of 2-d Ising Model in a
magnetic field"
LUC VINET, Laboratorie de Physique Nuc1eaire et Centre de
Recherches Mathematiques, Montreal, Canada
"Exact Operator Solution of the COLOGERO
SUTHERLAND Model"
Annotators: ZACHARY GURALNIK, Princeton University
Session Organizer: ANDRE LeCLAIR, Newman Laboratory, Ithaca, New York
12:30 AM Lunch Break
1:30 PM SESSION XII: EPILOGUE
Moderator: GEOFFREY WEST, Los Alamos National Laboratory
Dissertators: ALAN CHODOS, Yale University
"Sonoluminescence and the Heimlich Effect"
ZACHARY GURALNIK, Princeton University
"Critical Phenomena and the Boundary Conditions for
Schwinger-Dyson Equations"
GERALD GURALNIK, Brown University
xiv
"Using Symmetry to Numerically Solve Quantum Field
Theory"
GEOFFREY WEST
"Glueballs, the Essence of Non-Perturbative QeD"
DONALD WEINGARTEN, IBM, New York
"Evidence for the Observation of a Glueball"
Annotators: SESSION MODERATORS
Session Organizer: GEOFFREY WEST
3:30 PM THE 1996 ORBIS SCIENTIAE ADJOURNS
xv
CONTENTS
SECTION I - PROLOGUE
Innennost Structure of Matter ...................................................................... 3
Unbroken Non-Abelian Gauge Symmetry and Confmement ....................... 13
K. Nishijima
SECTION II - PROGRESS ON NEW AND OLD IDEAS - A

R-parity-violating Yukawa Couplings ........................................................ 19
V. Barger, M.S. Berger, RJN. Phillips, and
T. Wohrmann
Grand Unified Theories from Superstrings ................................................. 31
Gerald B. Cleaver
Searching for Dark Matter with the Future LHC Accelerator
at CERN Using the CMS Detector ............................................................ 43
Vasken Hagopian and Howard Baer
A Scale Invariant Superstring Theory with Dimensionless Coupling
to Supersymmetric Gauge Theories ............................................................ 49
M. Awada and F. Mansouri
Superstring Solitons and Conformal Field Theory ...................................... 57
L. Dolan
Comments on Symmetry Breaking Tenns in Quark Mass Matrices ............ 65
K. Tanaka
SECTION III - GRAVIT ATIONAL WAVES

LIGO: An Overview ................................................................................ 73
Barry C. Barish
Cosmology and LIGO ................................................................................ 79
Lee Samuel Finn
Interferometry for Gravity Wave Detection ................................................ 95
Peter Fritschel
xvii
SECTION IV - NEUTRINO MASSES
LSND Neutrino Oscillation Results ......................................................... 103
W.C.Louis
Theoretical Ideas about Neutrino Mass .................................................... 111
Lincoln W olJenstein
A Bayesian Analysis of Solar Neutrino Data ............................................ 115
Harrison B. Prosper
Ultrahigh-Energy Neutrino Interactions and Neutrino Telescope
Event Rates ............................................................................................. 121
Raj Gandhi, Chris Quigg, MH. Reno, and Ina Sarcevic
SECTION V - DIRAC'S LEGACY: LIGHT-CONE QUANTIZATION

Dirac's Legacy: Light-Cone Quantization ............................................... 133
Stephen S. Pinsky
Light-Cone Quantization and Hadron Structure ........................................ 153
Stanley 1. Brodsky
Discretized Light-Cone Quantization ....................................................... 183
Hans-Christian Pauli
Possible Mechanism for Vacuum Degeneracy in YM 2 In DLCQ .............. 205
Alex C. Kalloniatis
The Vacuum in Light-Cone Field Theory ................................................. 223
David G. Robertson
The Transverse Lattice in 2+ 1 Dimensions .............................................. 241
Brett van de Sande and Simon Dalley
SECTION VI - THE MATTER OF DARK MATTER

SUSY Dark Matter with Universal and Non-Universal Soft
Breaking Masses ..................................................................................... 253
R. Arnowitt and Pran Nath
Search for SUSY in the D0 Experiment.. ................................................ 265
Sharon Hagopian
Formulation of a Photosphere around Microscopic Black Holes ............... 273
Andrew F. Heckler
A Supersymmetric Model for Mixed Dark Matter. ................................... 283
Antonio Riotto
Light Photinos and Supersymmetric Dark Matter ..................................... 287
Edward W. Kolb
xviii
SECTION VII - PROGRESS ON NEW AND OLD IDEAS - B
Non-Universality and Post-GUT Physics in Supergravity Unification ...... 301
Pran Nath and R. Arnowitt
Pade Approximants, Borel Transform and Renormalons:
The Bjorken Sum Rule as a Case Study ................................................... 309
John Ellis, Einan Gardi, Marck Kanliner, and Mark A. Samuel
Hadron Supersymmetry and Relations between Meson and
Baryon Masses ........................................................................................ 319
DB. Lichtenberg
Constraining the QCD Coupling from the Superstring Hidden Sector ....... 323
Paul H. Frampton
Weak Interactions with Electron Machines: A Survey of
Possible Processes ......................... ;......................................................... 331
SL Mintz, M.A. Barnett, G.M. Gerstner, and M. Pourkaviani
SECTION VIII - EXACTLY SOLUBLE QUANTUM MODELS

Matrix Elements of Local Fields in Integrable QFr .................................. 349
G. Delfino and G. Mussardo
Boundary S Matrix for the Boundary Sine-Gordon Model
from Fractional-Spin Integrals of Motion ................................................. 359
Luca Mezincescu and Rafaell. Nepomechie
SECTION IX - EPILOGUE
Sonoluminescence and the Heimlich Effect.. ............................................ 371

Alan Chodos
Boundary Conditions for Schwinger-Dyson Equations and Vacuum
Selection .................................................................................................. 377
Zachary Guralnik
Numerical Quantum Field Theory Using the Source Galerkin Method ..... 385
G. S. Guralnik
Index ....................................................................................................... 395
xix
COLD VERSUS HOT CONDENSATION
TO CREATE MATTER
The process of condensation of the monopoles carrying magnetic charges gn with n

ranging from zero to infinity, to create an orbiton was first obtained twenty years ago in my
paper in Physical Review D Vol. 13, Number 6,15 March 1976, (see especially the pages
1539 and 1551). In contrast to the Bose-Einstein condensation of a dense gas of atoms near
absolute zero termperature in the recent experiments (July 1995) by Eric A. Cornell and his
colleague Carl Wieman to create a condensate as a "large atom", condensation of the magnetic
charges at the dawn of the universe was taking place in an inferno at Planck-scale
temperatures (- 10 30 degrees Kelvin) to create an orbiton (a quark with structure). In both
instances of the resulting condensates the distribution of atoms and monopoles, respectively,
range from packed to sparse. The color picture above for orbiton represents a layered
structure of magnetic charges with alternating signs and decreasing magnitudes. The painting
was commissioned in 1980 to Ms. Sheila Rose of Miami. It appeared in black and white on
page 1539 of the referred Physical Review paper. The confined magnetic charges, resulting
from the condensation prior to the Big-Bang creation of the universe, do also confine the
electric charges. For more scientific discussions, see my paper "After Einstein and
SchrOdinger: A New Unified Field Theory ," Journal of Physics Essays, Vol. 4, No.4, pp 439-
518, 1991 and the references there in pages 517 and 518. See also the 1994, 1995 and 1996
proceedings of the Coral Gables Conferences sponsored by the Global Foundation and
published by Plenum Publishing Company, New York.
The similarities of the Bose-Einstein condensation of a dense gas of atoms at a

temperature of absolute zero and that of magnetic charges at Planck -scale temperatures are
most striking. Do recent experiments pertaining to Bose-Einstein cold condensation also
vindicate the hot condensation pertaining to magnetic charges to create matter?
kursungf@netrunner.net
xx
SECTION I - PROLOGUE
INNERMOST STRUCTURE OF MATTER
Behram N Kllrsllnoglu
Global Foundation, Inc., Coral Gah/es, Florida
(kursungf@netrunner.net)
INTRODUCTION
I would like to present some new results regarding the ongm of mass and
distributions of electric and magnetic charges in an elementary particle. The discreteness
of the electric charge distribution and its confinement results from the layered
distribution, with alternating signs, of the magnetic charge. It is found that magnetic
charge layers thin out towards the surface of the particle while the electric charge
increases so that most of it resides on the particle' s surface. The fact of the electric and
magnetic charges lying on a circle [i.e., ro2 = (2G/c 4) (e 2+g 2 )] implies that the
discreteness of the magnetic charge distribution will result in the discreteness of electric
charge distribution.
In my January 1995 Coral Gables conference presentation (II "Exact Solutions for
Confinement of Electric Charges via Condensation of a Spectrum of Magnetic Charges"
it was pointed out that layered magnetic charge constituency with alternating signs
corresponds to the structure ot: for example. a quark of spin angular momentum K It
was clear that a point like quark could not possibly carry a spin and have mass. On this
occasion I shall provide, based on recent experiments. more theoretical evidence on the.
however small. ultimate structure of matter. In what follows I would like to discuss some
remarkable similarities of the condensation of gas of atoms near absolute zero
temperature with the condensation of a gas of monopoles at a temperature of the order of
1032 degrees kelvin that may have prevailed during the Big Bang creation of the universe.
The July 14, 1995 issue of the New York Times contained a spread announcing
the experimental results on the theoretical prediction of the Bose-Einstein condensation
phenomenon. The Eric A. Cornell et al experiment with rubidium gas cooled near
absolute zero revealed the creation of a Bose-Einstein condensate. The same type of
condensate was. after a month. obtained in an experiment with lithium gas at Rice
University. The details of the first experiment appearcd also in the July 14, 1995 issue of
the AAAS Science magazine in full color to illustrate the distribution of the atoms in the
condensate. The color picture was also included in page 19 of the August 1995 issue of
Scientific American. The recent 1996 calendar received from the American Physical
Society is graced by the same color picture of the condensate. Here I would like to
discuss this and another kind of condensation during the early universe referring to
monopoles and the corresponding condensate, the elementary particle. The latter type of
condensation was for the first time introduced in my paper in Physical Review D, Volume
13, Number 6, 15 March 1976.
3
BOSE-EINSTEIN CONDENSATION AT T~O
The most remarkable aspect of the condensate' s picture was the distribution of
atoms at 35 nanokelvin -- 35 billionths of a degree above zero -- across 100 microns from
packed (red rimmed portion in the picture) to .Iparse (yellow rimmed portion in the
picture) i.e. decreasing density qjatoms with the distance ji-om the origin. The Colorado
group saw the condensate formed at around 20 nanokelvins, the lowest temperature ever
achieved and included around 2000 atoms. The Rice University group achieved the
condensation of some 100.000 atoms at a temperature between 100 and 400 nanokelvins.
The Bose-Einstein condensation, compared to other phase transitions governed by

the forces between atoms and molecules, is driven by the quantum mechanical concepts.
In accordance with the uncertainty principle the position of the atoms are spread
proportional to their wave-lengths A related to their momentum p by the relation Ap=h.
When coiled near the zero temperature. atoms are barely moving. their positions become
uncertain. Correspondingly the wave function of the atoms spreads out and merge
leading to a quantum state occupied by a large number of atoms. As the temperature
dropped so did the size of the condensate atoms. CorneIrs group while scanning the
cloud of rubidium atoms with a laser found a sharp increase in density toward the middle.
The properties oj the condensate ineludes a survival lime of one minute bej(n'e .freezing
into rubidium-R7 ice.
In all these. bosons lose their individual identities. condensing into a part of a
superboson or a superatom. It is this loss of identity ncar absolute zero temperature that
the quantum mechanical wave function of neighboring atoms overlap and lead to the
formation of a condensate. Thus. for a condensate to emerge the experiment must
overcome the fact that the long-lived atoms are composite products and can stick together
not allowing the formation of a condensate. However, with the lowering of the
temperature. the atoms' wave lengths become longer and they can be packed close
enough together to merge to create a condensate.
The Rice University group used Iithium-7 gas which. unlike rubidium-87 atoms
repelling each other weakly (residual forces arising from their orbiting electrons).
consists of atoms that at/ract each other. This meant that they would form a liquid and
drain away long before the formation of a condensate. However, the Rice group seems to
have achieved the condensation of 100.000 atoms of lithium. The rather brief discussions
of the experimental findings on the Bose-Einstein condensation occurring at absolute zero
temperature will now be compared with the monopole condensation during the early
universe and creation of matter or quarks as condensates of monopoles at Planck-scale
temperatures (~I 01 ' degrees kelvin).
A BRIEF OUTLINE OF THE GENERALIZED THEORY OF GRA VITA nON
The idea of monopole condensation was inferred from the spherically symmetric
form of the generalized theory of gravitation (121). The theory was originated from the
nonsymmetric structure of general relativity in the presence of an electromagnetic field
where electric charges were not present. The basic nonsymmetric field variables in
general relativity expressed in their contravariant form are given by
(1)
4
which can be obtained. to order qo-l from the inverse of the covariant nonsymmetic
tensor(2)
(2)
where the constant qo has the dimensions of an electric field and the tensors 9J..!v and
<l>J..!v represent the generalized gravitational and generalized electromagnetic fields.
respectively. The addition of the anti symmetric tensor <l>J..!v to' 9J..!v as in (2) is
equivalent to turning on the electric and magnetic charges.
The Largrangian of general relativity can be expressed in terms of the

nonsymmetric contravariant tensor (I) provided the constant qo is restricted by the
fundamental relation (3)
ro2q 0 2= c4 /2G , (3)
where both real r0 (fundamental length). qo and their purely imaginary forms ir0 iqo
are allowed since in both cases the field equations of general relativity are unchanged.
This kind of invariance is referred to here as a super,l)lfl1mefry degeneracy of general
relativity where electric and magnetic charges are not included and the concept of spin
angular momentum does not come in. The use of the word supersymmetry here is not
related to its use in the conventional elementary particle physics. General relativity
predicts the existence of gravitational waves which carry energy and momentum but.
because of supersymmetry degeneracy or because of the symmetric field variables. they
do not can')' mass. However. for nonsymmetricfield variables arising from turning on
electric and magnetic charges the resulting theory, besides massless gravitational waves,
predicts the existence of massive waves carrying spin O. 1. and 2. Thus, in place of Higgs
bosons. as in the conventional theory. as the origin of mass. we find that the nomymmetTy
of the field variahles g~lV defined by definition (2) is the fill1dwnental basis for the
genera/ion ofll1([ss. It must also be understood that the nonhermitian and the hermitian
field variables 9 J..!V are SO(2) and lJ( I) gauge invariant. respectively. The two
supersymmetric real and complex field variables describe fermi-like and bose-like
paliicles. respectively.
However. if the nonsymmetric tensor g~lV as given by the definition (2) is used as
the basis of the generalized theory of gravitation then the supersymmetry degeneracy is
removed (2) and we obtain a theory which includes electric and magnetic charges along
with particles of half integral and integral spin angular momenta. In this case the
..,
fundamental relation (3) between qo- (energy density). and the fundamental length ro
can be interpreted as an eqllation o!sliJte and is most versatile in its cosmological and
elementary particle physics implications. For example. if ro is taken as large as the size
of the universe and qo "'~1 C-. , represents the average mass density in the universe then we
find that the equation (3) yields the results in the ballpark. The equation of state (3) can
also be written as
(4)
5
where Eo = q02 = energy density and Po = C2/2Gr02 = mass density. The r o-2
can be interpreted as the average curvature of space. If r 0 is of the order of the size of
the universe then the curvature of space is very small and the field equations of the
generalized theory of gravitation yield flat space-time solutions and therefore the universe
is approximately flat where the mass density is, as the universe keeps expanding,
constantly decreasing. The mass of a particle or the universe itself can be defined by
(5)
which can also be obtained by integrating the equation (4) over the r0 -space. Here
again, by substituting the value of ro as the size of the universe we obtain the ballpark
value for the mass of the universe (~ 1022M o ' Mo = total solar mass).
The mass relation (5) \vhen written in the f0l111 '
(6)
is reminiscent of the Schwarzschild singularity in general relativity but not refening "0
to coordinates, it is not related to that singularity. However. the relation (6) is
reminiscent of gravitational col/apse yielding a particle where ro is its gravitational size
lying inside the particle and M is its corresponding mass. The relation (6) is. of course.
independent of the coordinate system. From (5) we can write the relation
(7)
Hence, if we consider the special case of Planck particle we obtain
ropo = Yzft , (8)

where we choose
(9)
which is the famous Planck length with
(10)
representing halfofthe Planck mass.
CREATION OF MATIER VIA MONOPOLE CONDENSATION
At the instant of creation of the universe from a region of the vacuum of size 0 r
at the prevailing Planck-scale temperatures (~l 0" kelvin) the monopoles of positive and
negative magnetic charges lost their individual identity and with a mass small compared
to their energy Cp began to condense. The monopoles' wave-lengths were of the order
of Planck length and they were closely packed to merge and to form a monopole
condensate. The monopole condensate could have lasted only a Planck-time duration
(10,]1 sec) to change phase and could have '"frozen" into an orbiton (quark with structure)
6
or an antiorbiton (antiquark with structure) as illustrated in the figures 1,2. arid 3. If an
orbiton represents a quark with sfructure then it can constitute with additional quarks (or
antiquarks) elementary particles like. for example, protons, neutrons and the variety of
bosons (figures 3. 4). In the approximate solutions of the spherically symmetric field
equations where an angular (or hyperbolic) function <l> is a constant then the
fundamental length r0 is obtained as
(11 )
where e and 9 represent fundamental units of electric and magnetic charges,

respectively. The numbers N and :Jvl can be expressed as
(12)
where nand n' range over (0. I. 2....... 60 ....... ). The minus signs in the definitions
(12) refer to elementary particles while the plus signs are related to the size and the
expansion of the universe (creation of electric and magnetic charges from the vacuum)
where, for example. for n = n' = 60 one obtains ro ~ 10 28 cm .. the size of the universe.
For example, to obtain the proton mass from (5) we must choose the gravitational size of
proton relative to gravitational size of the universe (i.e. r 0 ~ 10'8 cm.) to be of the order
of 10.52 cm. so as to yield the ratio of the mass of the universe to proton mass to be of the
order of 10 so , the number of particles in the universe. Thus. in accordance with Mach' s
principle the inertia of a mass is due to the distribution of the rest of the mass in the
ulllverse.
Figure I. Figure 2.
Figure l. Confinement of the magnetic charge. Layered distribution of the magnetic charges with
alternating signs and decreasing amounts generates short-range forces to confine all the layers.
Figure 2. Confinement of the electric charge. Layered distribution of the electric charges of the same signs,
within the magnetic charge layers, with increasing amounts leads to the confinement of the electric chatges
residing mostly in the outer magnetic charge layers or on the "surface" of the elementary particles.
7
Figure 3. Magnetic charge dipole. Represents orbiton-antiorbiton synthesis to create spin zero or spin one
particles. It could also correspond to quark-antiquark combinations. The arrows represent directions of
spins. Addition of two spin angular momenta yields 0 or 1, -1 units represented by antiparallel and parallel
spin directions, respectively.
For the special case where n = n' = 0 we have the simple relation
(13)
where the discrete values gl' (l' = I. 2. 3, .... ) of the magnetic charge implies discrete
distribution of the electric charge itself which is also related to magnetic charge by
(14)
whereJll2 [= g2/(e 2 + g2] is the eigen-value which appears in the field equations
for regions of zero magnetic charge density (i.e. the interface between positive and
negative magnetic charges in an orbiton.) We note that for a "free" monopole, as shown
by P.A.M. Dirac a long time ago. the electric and magnetic charges are related by
eg = (Yl)nhc , (15)
where n is an integer. However. in our theory. as seen from relation (11) or relation (13).
there exists no free monopoles since both the electric and magnetic charges are confined
to constitute the elementary particles i.e .. the monopoles are hidden. Such a distribution
8
,I
Figure 4. Proton's constituents consist of three orbitons (or quarks with structure). Addition of the three
spin angular momenta of 1/2 units where the latter refers to a particle different from the proton.
of magnetic charge. where Ig

n = 0 represents the vanishing of the infinite sum of
magnetic charges which generate short-range force. We also observe that if all matter is
made of confined positive and negative magnetic charges then the ratio of dark matter to
luminous matter can be represented by g/e = eg/e 2 = (Y2)nfzc/e 2 ~ 68n. Hence.
depending on the choice of the integer n(= L 2. ... ) the universe may consist
predominately of dark matter.
The solutions of the tield equations where thc angle (or hyperbolic) function <D is
not a constant may have linear dependence on the positive and negative electric and
magnetic changes. Such solutions are expected to remain unchanged under the
interchange of the positive and negative electric charges and should lead to the existence
of electric and magnetic dipole moments. In fact the theory yields four sets of
generalized Dirac wave equations ,1 with the mass defined by (5). and with diicrete
space-time symmetries. In the meantime. aside from various electric and magnetic
moments. the discussion of the relation (13). for a given r 0- constitutes a definite proof
for the confinement of the electric charge. because of the discrete nature of the magnetic
charge distribution. most of it resides on the surface of the elementary particle and as seen
in figures 1 and 2. it remains stable.
The distribution of electric and magnetic charges in an elementary particle as
predicted by the generalized theory of gravitation yields a dual running coupling of the
fields. Thus. at very high energy (like. for example. in the canceled sse accelerator of
9
the order of 40 Tev). scattering of charged particles (or preferably proton-antiproton
collisions) while the electromagnetic coupling decreases with increasing energy the
strong coupling. experienced during interparticle penetration. increases. The running
coupling constants which appear in the field equations for the spherically symmetric
fields are given by
where the electric charge e increases towards the surface of the particle whilst the
magnetic charge 9 increases towards the origin of the particle. In the same way e and 9
decrease towards the origin and towards the surface. respectively. Both e and 9 assume
zero values at the origin. Experimentally very high energy proton-antiproton scattering
may provide, hopefully. some clues with respect to the nature of the above mentioned
structural properties of the elementary particles.
DISCRETENESS OF CONFINED ELECTRIC CHARGE DISTRIBUTION
If in the fundamental relation (13) we represent the fundamental length r0 in

discrete units of the Planck length by substituting
(17)
in (13) we obtain
(18)
where
N ~ 10n , n = 0, 1, 2, .... , (19)

and where e and 9 represent electric and magnetic charges. respectively. as restricted by
the relation (18). and they lie on a circle of radius (N/(",J2))(-V(hc)). The
discreteness of the magnetic charge 9 (= 9" 92, 93, ..... 9n, ..... ) implies.
because of (18). a discrete spectrum or the quantization of the electric charge e (=e 1,
e2, e3, ..... en, ..... ) where

~9n =0, 9n=(-lt 19n1, 19n1 > 19n11, Lim. 9n =0, (20)
(21)
The distribution of the confined magnetic charges as quantified by the relation

(18) and portrayed in the figure 1 determines the confined electric charge distribution as
described by the relations (21) and portrayed in figure 2. It is clear that most of the
electric charge. as confined by the magnetic charge distribution, resides on the surface of
the elementary patticle.
10
REFERENCES
1. Behram N. Kursunoglu. Unified Symmeli)' In the Small and In the Large, 1995,
Volumes 1 and 2, Plenum Press, New York. edited by Behram N. Kursunoglu et al.
2. Behram N. Kursunoglu, .Journal of Physics Essays. Vol. 1. No.4. pp. 439-518, 1991,
University of Toronto Press.
3. Behram N. Kursunoglu, Physical Review, 88. 1369 (\ 952).
4. Behram N. Kursunoglu. Physical Review D. Volume 12. Number 6, 15 March 1976.
5. See the July 14. 1995 issue of the New York Times. AAAS Science Magazine. and
the August 1995 issue of Scientific American (page 19).
11
Unbroken Non-Abelian Gauge Symmetry and Confinement
K. Nishijima
Department of Physics, Chuo University
Bunkyo-ku, Tokyo 112, Japan
It is shown that color confinement is an inevitable consequence of unbroken color

symmetry and asymptotic freedom of QCD.
1. Interpretation of Color Confinement
The quark model of hadrons has been so successful that we can no longer think of
any other substitute for it. All the experimental evidences for this model have been
indirect, however, since no isolated quarks have been observed to date. Thus the hypothesis
of quark confinement emerged implying that isolated quarks are in principle unobservable.
Later, this was promoted to the hypothesis of color confinement that implies the
unobservability of all the isolated colored particles including gluons.
Then a natural question is raised of whether we can account for this hypothesis
within the framework of the conventional QCD or we need a new additional principle. It is
the purpose of the present paper to stress that color confinement is an inevitable
consequence of the conventional QCD provided that color symmetry is not spontaneously
broken and that asymptotic freedom is valid. Since the mathematical details of its proof
have been published elsewhere,03>, we shall give here the basic ideas underlying this
approach.
The solution of this problem is decomposed into two steps. First, we have to find a
proper interpretation or definition of color confinement, and then we have to prove it. In
fact, there is a variety of interpretations of confinement. To quote a few, Wilson's area law4 )
in the lattice gauge theory leads to the linear potential between a pair of a quark and an
antiquark that holds the system to be always in bound states. Another example is the recent
supersymmetric theory of Seiberg and Witten 5>,6l in which the duality between electric and
magnetic fields holds, and confinement is then a consequence of the condensation of
magnetic monopolies. Therefore, speaking of confinement we have to specify what it
means.
We start looking for a known example of confinement within the framework of
known field theories. Then it occurs to us that we have a prototype example of confinement
in QED. let us quantize the electromagnetic field in a covariant gauge, say, in the Fermi
gauge, and we recognize that there are three types of photons, namely, transverse,
longitudinal and scalar photons. Of these three types only the transverse photons are subject
to observation, and the latter two escape detection. This is indeed a typical example of
confinement, and we shall recapitulate the underlying implication.))
Quantization of the electromagnetic field introduces the indefinite metric that was
inherited from the Minkowski metric. In order to adopt the probabilistic interpretation of
quantum mechanics to QED, it is necessary to confine ourselves to physical states which
13
are free of negative probability. 'In fact, the Lorentz condition selects such states, and in
particular those states that involve only transverse photons and changed particles belong to
the physical subspace of the whole state vector space. The S matrix, then, transforms a
physical state into another physical state.
Let us consider the unitarity condition of the S matrix between two transverse
photon states, then the intermediate states are saturated by physical states. In fact, both
longitudinal and scalar photons show up in the intermediate states, but their contributions
cancel themselves leaving only those of the transverse photon states. As a result,
longitudinal and scalar photons are not observable, implying confinement of these
unphysical photons. This mechanism of confinement may be referred to as metric
cancellation since it is due to the indefinite metric.
We are now concerned with how we should extend this interpretation of
continement to QCD which is a typical non-abelian gauge theory. In QCD we introduce a
pair of so-called Faddeev-Popov ghost fields in order to keep the S matrix unitary. They are
anticommuting hermitian scalar fields denoted by c and c respectively. Since they violate
Pauli's theorem on the connection between spin and statistics we are obliged to introduce
indefinite metric again.
For the gauge fields as well as quark fields we can introduce local gauge
transformations. Let us consider an infinitesimal local gauge transformation and replace the
intinitesimal gauge function by either c or c, and we obtain the BRS or anti-BRS
transformation of the respective fields. For the ghost fields local gauge transformations
cannot be defined, but their BRS or anti-BRS transformations can be defined so as to keep
the total Lagrangian density invariant. Then Noether's theorem leads to the conserved
charges corresponding to the BRS and anti-BRS invariances, respectively. They are called
BRS charges.
In QCD the subsidiary condition corresponding to the Lorentz condition is the
Kugo-Ojima condition.7), 8) Namely, physical states are defined as those states that are
annihilated by applying the BRS charge. The collection of physical states forms the
physical subspace of the whole state vector space, and it is an invariant subspace of the S
matrix in QCD just as in QED. Therefore, when color multiplet states, such as isolated
quark or gluon states, do not belong to the physical subspace, they escape detection in the
same sense as the longitudinal and scalar photons do in. QED. We may interpret this as
color confinement, and we now know what we should prove for color confinement.
2. A Sufficient Condition for Color Confinement
Now that we have introduced an interpretation of confinement in the preceding

section we have to investigate the condition under which confinement is realized.
Because of the presence of the gauge-fixing term in the Lagrangian density the
equation for the gauge field deviates from the standard Maxwell equation. The current
corresponding to this deviation can be obtained can be obtained by applying the BRS and
anti-BRS transformations successively to the gauge field. Then let us introduce three-point
Green functions involving this current and a pair of quark fields or gluon fields. By taking
the four-divergence of these Green functions with respect to the space-time coordinates of
this current we tind that they can be equated to two-point functions of the quark fields or
gauge fields, respectively, by making use of the modified Maxwell equation. They are the
so-called Ward-Takahashi identities and we can write them down for any pair of colored
fields provided that their transformation properties under the color SU(3) group are known.
When isolated quark states as well as gluon states are not annihilated by applying
the BRS charge, then they escape detection since they are not physical states. In order to
relate Green functions to quark or gluon states for the purpose of checking the above
14
condition we have to refer to the LSZ reduction formula. 8) Therefore, we shall assume it~
validity in what follows. Then we can prove that the asymptotic gluon field, either
incoming or outgoing, consists of two terms, one corresponding to the gluon and the other
to the gradient of a massless spin zero (ghost) particle. By combining the Ward-Takahashi
identities with the LSZ reduction formula we can show that color confinement is realized
provided that successive applications of the BRS and anti-BRS transformations annihilate
the second term in the asymptotic gluon field. This is indeed a sufficient condition to be
referred to as the condition A. Since the asymptotic field is a rather complicated object, we
shall express this condition in terms of Heisenberg operators.
For this purpose let us consider a two-point function defined as the vacuum
expectation value (VEV) of the time-ordered product of the gauge field and the current
corresponding to the deviation from the Maxwell equation. When the residue of the
massless spin zero pole of this two-point function vanishes, the condition A is satisfied.
Let us denote this residue as C, then it cannot vanish when color symmetry is
spontaneously broken. Indeed, the non-vanishing residue is a signature of the emergence of
the Nambu-Golstone boson. Therefore, color symmetry should not be broken for the
realization of color confinement.
In general this constant C is gauge-dependent and satisfies a simple renormalization
group (RG) equation. This equation alone cannot determine C, however, unless a proper
boundary condition or a normalization condition is given. Thus we shall stand on a slightly
different point of view. We realize that this constant C can be expressed as the VEV of an
equal-time commutator (ETC) between two Heisenberg operators in the two-point function
mentioned above. This ETC is given as a sum of two terms. The first term denoted by a is
equal to the inverse of the renormalization constant of the gluon field Z3' and the second
term is the so-called Goto-Imamura-Schwinger (GIS) term 10), II) that was first discovered in
the evaluation of the ETC between the space and time components of the charge current
density in QED. For the GIS term both the RG equation and the boundary condition are
known, and it can be expressed uniquely in terms of a vanishes, so that the evaluation of a
is now the central issue.
The renormalization constant Z3 is gauge-dependent in QCD, but the concept of
confinement extends to other gauges in which a does not vanish.
Fortunately it is possible to evaluate a exactly with the help of RG, and we can
prove that we can always find gauges in which a vanishes exactly provided that asymptotic
freedom is valid.
Thus we conclude that confinement is an inevitable consequence of an unbroken
nonabelian gauge symmetry and asymptotic freedom. The electro weak interactions are not
related to confinement since the original non-abelian gauge symmetry SU(2) x U(l) is
spontaneously broken and reduces to the abelian gauge symmetry U(1).
Reference
1) K. Nishijima, Int. 1. Mod. Phys. A9 (1994) 3799.

2) K. Nishijima, Int. 1. Mod. Phys. AI0 (1995) 3155.
3) K. Nishijima, Czech. J. Phys. 46 (1996) 1.
4) K. Wilson, Phys. Rev. D14 (1974) 2455.
5) N. Seiberg and E. Witten, Nucl. Phys. B426 (1994) 19.
6) N. Seiberg and E. Witten, Nucl. Phys. B431 (1994) 484.
7) T. Kugo and L Ojima, Phys. Lett B73 (1953) 255.
8) T. Kugo and L Ojima, Prog. Theor. Phys. SuppL No.66 (1979) 1.
9) H. Lehmann, K. Symanzik and W. Zimmermann, Nuovo Cim. 1 (1955)205.
lO) T. Goto and T. Imamura, Prog. Theor. Phys. 14 (1955) 396.
11) 1. Schwinga, Phys. Rev. Lett. 3 (1959) 296.
15
SECTION II - PROGRESS ON NEW AND OLD
IDEAS- A
R-parity-violating Yukawa couplings
V. Barger(a), M.S. Berger(b), R.J.N. Phillips(c), and T. Wohrmann(a)
(a) Physics Department, University of Wisconsin,

Madison, WI 53706, USA
(b)Physics Department, Indiana University,
Bloomington, IN 47405, USA
(C) Rutherford Appleton Laboratory,
Chilton, Didcot, Oxon OXIl OQX, UK
ABSTRACT
We discuss the evolution of R-parity-violating (RPV) couplings in the minimum super-

symmetric standard model, assuming a hierarchy for coupling strengths and empha-
sising solutions where R-conserving and R-violating top quark Yukawa couplings both
approach infrared fixed points. We show that fixed points offer a new source of bounds
on RPV couplings at the electroweak scale, and that lower limits on the top quark mass
lead to RPV constraints at the GUT scale. We show how the evolution of CKM matrix
elements is affected. Fixed-point behaviour is compatible with present constraints, but
for top-quark couplings would require participating sleptons or squarks to have masses
2:: mt to avoid unacceptable top decays to sparticles.
1. INTRODUCTION
Supersymmetry is a very attractive extension of the Standard Model (SM), so its low-
energy implications are being vigorously pursued. 1,2 In the minimal supersymmetric
standard model (MSSM), with minimum new particle content, a discrete symmetry
(R-parity) is assumed to forbid rapid proton decay. The R-parity of a particle is
R == (_1)3B+L+2s, where B, Land S are baryon number, lepton number and spin;
thus R = +1 for particles and R = -1 for sparticles. An advantage of R-conservation
is that the lightest sparticle is stable and hence provides a candidate for cold dark
matter. However, since R-conservation is motivated empirically and not by any known
Talk presented by V. Barger
19
principle, the possibility of R-nonconservation also deserves serious consideration. In
addition to the Yukawa superpotential in the MSSM
(1)
there are two classes of R-violating couplings in the MSSM superpotential, allowed by
supersymmetryand renormalizability.3
The first class of superpotential terms violates L,
W= ~AabcLLLtER + A~bcLLQtlJR + JliH2Li' (2)
while the second class violates B,
W I,,, D-aD-bU-c (3)

= 2"abc R R R
Here L, Q, E, lJ, {j denote the doublet lepton, doublet quark, singlet antilepton, singlet
d-type antiquark, singlet u-type antiquark superfields, respectively, and a, b, c are gen-
eration indices. (V)ab, (D)ab and (E)ab in Eq. (1) are the Yukawa coupling matrices. In
our notation, the superfields above are the weak interaction eigenstates, which might
be expected as the natural choice at the grand unified scale, rather than the mass
eigenstates.
The term JliLiH2 in the superpotential can be rotated away into the R-parity con-
serving term JlHi H2 via a SU( 4) rotation between the superfields Hi and Li . However
this operation must be performed at some energy scale, and the mixing is regenerated
at other scales through the renormalization group equations.
To forbid fast proton decay, it is sufficient to forbid either L-violating couplings or
B-violating couplings, while retaining the other class of RPV interaction. We follow
this course.
The Yukawa couplings Aabc and A~bc are antisymmetric in their first two indices
because of superfield anti symmetry, so there are 9 independent couplings of each kind.
There are also 27 independent A~bc couplings, making 45 altogether. These superpo-
tential terms lead to the interaction lagrangians
(4)
' A: bc {vaLdcRdbL + dbLdcRVaL + (dcR)*(iJaL)CdbL
-eaLdcRubL - ubLdcReaL - (dcR)*(eaL)CUbd + h.c. , (5)
a b + Uc a b + Uc a b} + h .c.
" = 21 A"abc {Uccdcd* cd*d c -*dcd C (6)
There are phenomenological upper limits on the various couplings Aabc, A:bc, A~bc from
colliders and low-energy data,3-8 from proton decay9 and from cosmology,lO but con-
siderable latitude remains for RPV. These limits are generally stronger for couplings
with lower generation indices.
There are far too many RPV parameters for comfort. However, we know that the
dominant Higgs couplings are the third generation, At, Ab, An and there may plausibly
exist a similar generational hierarchy among the RPV couplings. We shall therefore
retain only A233, A;33, A~33' which have the maximum of third-generation indices and
are also the least constrained phenomenologically.
The renormalization group evolution equations (RGE), relating couplings at the
electroweak scale to their values at the grand unification (GUT) scale, have given new
20
insights and constraints on the observable low-energy parameters in the R-conserving
scenario. Let us see what can be learned from RGE in RPV scenarios. An initial study
of A~33 and A~33 evolution8 was later extended to all baryon-violating couplings A;jkY
Our present work is a somewhat more general study of the RGE for RPV interactions,
emphasising solutions where R-conserving and R-violating top Yukawa couplings both
simultaneously approach infrared fixed points. 12 Such fixed-point behaviour requires
a coupling A, )..', or A" to be of order unity at the electroweak scale. We implicitly
assume that RPV couplings do not have unification constraints at the GUT scale,9
which would forbid this behaviour. After our study was completed, two related works
on RGE for RPV couplings appeared,!3,14 which however have a different focus and
are largely complementary to the present work. In Ref. 14, de Carlos and White have
studied the evolution of the soft supersymmetry-breaking terms and find strong limits
can be placed on R-parity violating couplings by imposing neutrino mass limits and
bounds on lepton flavor violation.
2. RENORMALIZATION GROUP EQUATIONS AND FIXED POINTS
The evolution of the couplings d abc with the scale p, for any trilinear term in the
superpotential dabcq>aq>bq>c, is given by the RGE
p ~ d abe = ,: debe + Ib d aec + I~ d abe , (7)
where the I~ are elements of the anomalous dimension matrix.

With the simplifying assumption that only third-generation Higgs and our selected
RPV couplings contribute in the Yukawa sector, the one-loop RGE become
dQi
dt
L 7,
bi Q bi = {33/5, 1, -3} (8)
dIt ~
dt 27r It (6It + y" + Y' + 2Y" - !QQ3
3 - 3Q2 - liQI)
15 (9)
dYb
dt
2~ Yb (It + 6Yb + Y,. + 6Y' + 2Y" - .1fQ3 - 3Q2 - tsQI) (10)
dY,.
dt
1 Y,. ( 3y"
27r + 4Y,. + 4Y + 3Y'- 9
3Q2 - 5QI ) (11)
dY
dt
1 Y ( 4Y,.
27r + 4Y + 3Y'- 9
3Q2 - 5QI ) (12)
dY'
dt
~Y'
27r (It + 6y" + Y;,. + Y + 6Y' - !QQ3
3 - 3Q2 - lQI)
15 (13)
dY"
dt
2~ Y" (2It + 2Yb + 6Y" - 8Q3 - ~QI) . (14)
Here Qi = 4~g; , the variable is t = In(p/ Ma) where p is the running mass scale and
Ma is the GUT unification mass, and we define
1
Ii = ~A7 (i = t,b,r), 2
Y = -A 233 .
47r 47l'
It is understood that we take either Y = Y' = 0 or Y" = o.
2.1. At fixed point in the MSSM
An extremely interesting possibility is that It is large at the GUT scale and conse-
quently driven toward a fixed point at the electroweak scale. 15 ,16 In the pure MSSM
21
(RPV neglected), the fixed-point condition dY,./dt ~ 0 at J-t ~ mt gives
(15)
Now At and Ab at J-t = mt are related to running masses
(16)
)-1/2
In
where v = ( v2 GF = 246 GeV and tan/1 = V2/VI is the ratio of the Higgs vevs.
Here 1Jb gives the QeD/QED running of mb(J-t) between J-t = mb and J-t = mt; T/b ~ 1.5
for o:.(mt) ~ 0.10.16 Then
(17)
taking mb(mb) = 4.25 GeV, mt(mt) = 167 GeV, and hence
1/,(mt) ~ 3 X 10- 4 tan 2 ,B yt(mt). (18)

For moderate values tan /1 ~ 20, we can neglect 1/" and the the approximate values
0:3= 1/10, 0:2 = 1/30, 0:1 = 1/58 at J-t = mt then give
(19)
A more precise numerical analysis shows that At --+ 1.06 as J-t --+ mt. Since At(mt) =
V2mt( mt)/( v sin /1), this leads to the relation 16
mt(pole) = (200 GeV) sin/1, (20)
where mt(pole) is the mass at the t-propagator pole. It is interesting to examine the
impact of RPV couplings on this result.
2.2. A", At simultaneous fixed points

In the B-violating scenario with Y = Y' = 0 and Y" non-zero, the possibility that both
Yt and Y" approach fixed-point limits was found numerically in Ref. 8 (note that these
authors use a different definition of A~bc). The corresponding conditions dYt/dt ~ 0
and dY"/dt ~ 0 at J-t ~ mt give
6yt+ 1/, + 2Y" - !f0:3 - 30:2 - R0:1 0, (21)

2Yt + 21/, + 6Y" - 80:3 - ~O:l o. (22)
Solving for Yt and Y" we obtain (if 1/, Yt)
At ~ 0.94, A~33 ~ 1.18, (23)
with At displaced downward due to A~33.

This large fixed-point value of A~33 would give strong t --+ bs, sb decay, if kinemati-
cally allowed.
With both At and A~33 at fixed points as above, the predicted top quark mass
becomes
mt(pole) ~ (150 GeV) sin/1. (24)
22
Even for moderate values of tan,8 (tan,8 > 5) one has sin,8 ~ 1 (sin,8 > 0.98). This
prediction is therefore at the lower end of the present data: 17,18
mt = 176 8 10 GeV (CDF) , mt = 199!~~ 22 GeV (DO). (25)
More precise data could eventually exclude the fixed-point possibility for A~33 .
In the case of large tan,8, the coupling Yi, is non-negligible and may even be near
its own fixed point given by dYi,jdt ~ 0; then
(26)
Here Y,. can be related to Yi, since A'T (mt) = V2m'T (mt) j ('rJ'T V cos ,8), and hence
(27)
by arguments similar to those above relating Ab(mt) to At(mt). Then we have three
simultaneous equations in three unknowns, with the solutions
(28)
2.3. N or A, At simultaneous fixed points

If instead fixed points should occur simultaneously for yt and Y' (with Y" = 0), the
conditions dytj dt ~ 0 and dY' j dt ~ 0 at J1 ~ mt give
yt is [llfa3 + 15a2 + Ral + Y'T + Y] , (29)

Y' is [llfa3 + 15a2 + ~al - 35Yi, - 6Y,. - 6Y] (30)
If Y is small and we also neglect Yi, and Y'T (assuming small tan,8), then yt and Y'
approach almost the same fixed-point value
(31)
In this case At(mt) is only slightly displaced below the MSSM value, while A;33 has quite
a large value. The latter would imply substantial t -+ bT, fb decays, if kinematically
allowed; the t -+ bT mode is more likely, since T is usually expected to be lighter than
b, and we discuss its implications later. Alternatively, if Y' is negligible, yt and Y can
approach fixed points simultaneously; in this case the two conditions dytjdt ~ 0 and
dYjdt ~ 0 essentially decouple, giving the MSSM result for yt. Neglecting Yi, and Y,.,
the solution is
(32)
but if Yi, too is large and approaches its fixed point, the three corresponding conditions
give
(33)
while the A233 fixed point is very small and never truly reached in numerical studies. It
is also not possible for Y, Y' and yt to have simultaneous fixed points; the conditions
dY j dt = dY' j dt = dytj dt. = 0 cannot be satisfied with all three couplings positive.
23
2.4. CKM evolution
The presence of non-zero RPV couplings can also change the evolution of CKM mixing
angles. Assuming, as we do, that only the RPV couplings A233, A;33 or A~33 are non-
zero, it turns out 12 that the one-loop RGE for mixing angles and the C P-violation
parameter J = Im(v"d v". v,,*. v,,'d) have the same forms as in the MSSM, namely19
dW
dt = -
W
811"2
(2 2)
At + Ab , (34)
where W = lv"bI 2, lv"bI 2, IVtdI 2 , IVt.1 2 or J. Nevertheless the evolution of CKM angles
differs from the MSSM because the evolution of the Yukawa couplings on the right
hand side is altered by the RPV couplings.
3. NUMERICAL RGE STUDIES
It is instructive to supplement our algebraic arguments above with explicit numerical

solutions of the RGE. Figure 1 shows the fixed-point behaviour of the three RPV
couplings considered in this paper, (A~33' A;33, A233) along with the corresponding fixed
point behaviour for At, assuming that tan,8 is small so that Ab and AT are negligible.
We see that for all A ;::: 1 at the GUT scale, the respective Yukawa coupling approaches
its fixed point at the electroweak scale. These infrared fixed points provide theoretical
upper limits for the RPV-Yukawa couplings at the electroweak scale, summarized in
Table 1. The numerical evolution of the fixed points approaches but does not exactly
reproduce the approximate analytical values Eqs. (28), (31) and (32).
Table 1: Fixed points for the different Yukawa couplings A in different models for
i) tan,8 ,$ 30 and ii) tan,8 '" mt/mb. In the case oflarge tan,8, Ab also reaches a fixed
point.
Model At Ab A233 A;33 A~33
i) MSSM 1.06
Lepton # Violation (A N) 1.09 0.90
Lepton # Violation (N A) 1.03 1.01
Baryon # Violation 0.90 1.02
ii) MSSM 1.00 0.92
Lepton # Violation (N A) 1.01 0.72 0.71
Baryon # Violation 0.87 0.85 0.92
We remark in passing that RPV couplings must be well above their fixed-point
values to explain5 the apparent discrepancy between theory and experiment for Rb =
r(Z -+ bb)/r(Z -+ hadrons).
We obtain additional limits on the RPV couplings from the experimental lower
bound on mt (that we take to be mt > 150 GeV 17,18). These are shown in Fig. 2; the
dark shaded region is excluded in all types of models only by assuming this lower bound
on the top mass.
Finally we examine RPV effects on the evolution of off-diagonal terms in the CKM
matrix. When the CKM masses and mixings satisfy a hierarchy, the evolution from
electroweak to GUT scales is given by
W(GUT) = W(p)S(p),
24
nI, M(GUT) nI, M(GUT)
5
a) Baryon # violation b) Baryon # violation

4 -- A,(t) A,(G T)= 4.0 4 -- A'233(t) A'2J3(GUT) = 4.0
for A'~m(GUT) = 2.0 for A,(GUT) = 2.
3.0 3.0
A "-
2 2.0 2 2.0
1.0
0.2 0.2
0 0
-30 -20 -10 0 -30 -20 -10 0
5 5
c) Lepton # violation d) Lepton # violation

4 - - A,(t) A,(GUT) = 4.0 4 -- Am(t)
for Am(GUT) = 2.0. for A,(G T) = 2.8,
A:I33(GUT) = 0.2 A~\J3(G T)=0.2
3 3.0 3 3.0
A "-
2 2.0 2 2.0
-- 1.0 1.0
0.2 0.2
0 0
-30 -20 -10 0 -30 -20 -10 0
c) Lepton # violation I) Lepton # violation

4 - - A, (t) A,(G T) = 4.0 4
for Am(GUT) =0.2. ~ r A,(G T)=2.8,

A~l33(GUT) =2.0 Am(G T) = 0.2
3.0 3.0
"- "-
2.0 2 2.0
---------0.2 ----------0.2
0'---'----"'----'----'----'----'---'----'
-30 -20 -10 o -30 -20 -10 o
Fig. 1. Couplings>. as a function of the energy scale t for >'t in (a) baryon number RPV, (c) lepton
number RPV with >'233 >';33 and (e) lepton number RPV with >';33 >'233 for different starting
points at the GUT scale (t = 0). Panels (b), (d) and (f) show the same for >'~33' >'233 (>'233 >'~33)
and >'b3 (>'~33 >'233) respectively. Here t ~ -33 represents the electroweak scale, where these
couplings reach their fixed points.
25
4 r---.-.--r--~----~--~---,
b)
o ~~w---~--~----~--~--~
o 2
A,(GUT) A,(GUT)
Fig. 2. Excluded regions in the (a) At(GUT), A~33(GUT) plane and (b) At(GUT), A233(GUT)
(A233(GUT) = A;33(GUT)) plane obtained from mt > 150 GeV.
where W is a CKM matrix element connecting the third generation to a lighter gen-
eration and S is a scaling factor19 found by integrating Eq. (34) with the other RGE.
The remaining CKM elements do not evolve to leading order in the hierarchy. Figure 3
shows how S depends on the GUT-scale RPV couplings '\233, '\;33 and .\~33'
3 3
P
;:J P
;:J
Q
~
M
M
2
%
~
2
oL-J.-LL....L---LL__...L..J._ _' - - - - - "_ _- ' -_ _-'------' o L-J.~~~ __ ~~ __- L__ ~ __ ~~
o 2 4 o 2 4
,",(GUT) ,",(GUT)
Fig. 3. Contours of constant Sl/Z for different values of (a) A~33(GUT) and At{GUT) (baryon number
violation) and (b) AZ33(GUT) = A~33(GUT) and At(GUT) (lepton number violation).
4. RPV DECAYS OF THE TOP QUARK
The RPV couplings .\~33 and .\~33 would give rise to new decay modes of the top quark,20
if the final-state squark or slepton masses are small enough. L-violating .\~33 leads to
tR -t bRTR, bRfR decays, with partial widths 20
r(t-tbT) (.\~33)2 m (1 _ m~/m2)2 (35)

3271" t T t ,
r(t-tbf) (.\;33)2 m (1 _ m~/m2)2 (36)

321T t b t ,
26
neglecting mb and mT' The former mode is more likely to be accessible, since sleptons
are expected to be lighter than squarks. Since the SM top decay has partial width
(37)
the ratio of RPV to SM decays would be typically
It is natural to assume that T would decay mostly to T plus the lightest neutralino X~
followed by the RPV decay X~ --t bbvT(VT), with a short lifetime 21
giving altogether
(40)
This mode could in principle be identified experimentally, e.g. via the many taggable
b-jets and the presence of a tau. However, it would not be mistaken as the SM decay
modes t --t bW+ --t bqij',bfv, (f = e,p), that form the basis of the presently detected
pp --t tEX signals in the (W --t fv) + 4jet and dileptonchannels (neglecting leptons
from T --t fvv that suffer from a small branching fraction and a soft spectrum). On the
contrary, the RPV mode would deplete the SM signals by competition. With m T rv M w ,
fixed-point values >'~33 ~ 0.9 (Fig.l) would suppress the SM signal rate by a factor
(1 + o. 70( >'~33)2t2 ~ 0.4, in contradiction to experiment where pp --t {EX --t bbWWX
signals tend if anything to exceed SM expectations. 17,18 We conclude that either the
fixed-point value is not approached or the T mass is higher and reduces the RPV effect
(e.g. m T = 150 GeV with >'~33 = 0.9 would suppress the SM signal rate by 0.88 instead).
Note that our discussion hinges on the fact that the RPV decays of present interest
would not contribute to SM top signals; it is quite different from the approach of Ref. 7,
which considers RPV couplings that would give hard electrons or muons and contribute
in conventional top searches. _
Similarly, the B-violating coupling >'~33 leads to tR --t bRsR, bRsR decays, with
partial widths
r(t --t bs) = r(t --t bs) = (~~~2 mt (1 - m~/m;)2 , (41)
neglecting mb and m. and assuming a common squark mass mb = m. = m q. If the

squarks were no heavier than 150 GeV, say, the ratio of RPV to SM decays would be
r(t --t bs,bs)/r(t --t bW+) ~ 0.16 (>'~33? (for mq = 150 GeV) . (42)
These RPV decays would plausibly be followed by ij --t qX~ and X~ --t cbs, cbs (via the
same >'~33 coupling with a short lifetime analogous to Eq.(39)), giving altogether
t --t (bs,sb) --t bsX~ --t (cbbbs,cbbbs). (43)
This all-hadronic mode could in principle be identified experimentally, through the

multiple b-jets plus the t --t 5-jet and X~ --t 3-jet invariant mass constraints. However,
it would not be readily mistaken for the SM hadronic mode t --t bW --t 3-jet, and would
simply reduce all the SM top signal rates. If the coupling approached the fixed-point
value >'~33 ~ 1.0, while mq ~ 150 GeVas assumed in Eq.( 42), the SM top signals would
27
be suppressed by a factor (1 + 0.I6(A~3J2)-2 '::::' 0.75, which is strongly disfavored by
the present data l7 ,18 but perhaps not yet firmly excluded.
If indeed the s- and b-squarks were lighter than t to allow the B-violating modes
above, it is quite likely that the R-conserving decay t --t ix~ would also be allowed,
followed by i --t ex~ (via a loop) and B-violating decays for both neutralinos, with net
effect
- 0 0 0 -- ----
t --t tx 1 --t ex IX 1 --t (eecbbbb, ccbbchb, cccbbbb). (44)
This seven-quark mode would look quite unlike the usual SM modes and would further
suppress the SM signal rates. Depending on details of the sparticle spectrum, however,
other decays such as i --t bW X~ might take part too, leading to different final states;
no general statement can be made except that they too would dilute the SM signals
and therefore cannot be very important.
5. CONCLUSIONS
We have shown how the RGE for SM Yukawa couplings and CKM elements would
be affected by RPV, assuming hierarchical couplings.
We have identified fixed points in the RPV couplings and At simultaneously.
These give upper bounds on RPV couplings at the electroweak scale [Fig.I].
There are large tan f3 scenarios where Ab too has a fixed point.
The fixed point values [Table 1] are compatible with present constraints.
However, fixed-point values of A~33 or A~33 would require the corresponding slep-
tons or squarks to have mass 2: mt, to avoid strong top decays to sparticles.
The fixed points give constraints, correlating the RPV couplings with At at the
GUT scale, from lower bounds on mt [Fig.2).
RPV couplings affect the evolution of CKM mixing angles [Fig.3].
Acknowledgements
VB thanks Herbi Dreiner for a discussion and the Institute for Theoretical Physics at the
University of California, Santa Barbara for hospitality during part of this work. RJNP
thanks the University of Wisconsin for hospitality at the start of this study. We thank
B. de Carlos and P. White for pointing out our omission of the Higgs-lepton mixing
anomalous dimension in an earlier version of our RGE. This research was supported
in part by the U.S. Department of Energy under Grant Nos. DE-FG02-95ER40896
and DE-FG02-9IER4066I, in part by the National Science Foundation under Grant
No. PHY94-07194, and in part by the University of Wisconsin Research Committee
with funds granted by the Wisconsin Alumni Research Foundation and support by
NSF. TW is supported by the Deutsche Forschungsgemeinschaft (DFG).
28
References
1. For an introduction to supersymmetry, see J. Wess and J. Bagger, Supersymmetry and Supergravity
(Princeton University Press, 1983); P. Fayet and S. Ferrara, Phys. Rep. 32, 249 (1977); P. West,
Introduction to Supersymmetry and Supergravity (World Scientific, 1986); R.N. Mohapatra, Unifi-
cation and Supersymmetry (Springer-Verlag, 1986).
2. For phenomenological reviews of SUSY, see H.P. Nilles, Phys. Rep. 110, 1 (1984); G.G. Ross,
Grand Unified Theories (Benjamin Cummings, 1985); R. Arnowitt, A. Chamseddine and P. Nath,
Applied N = 1 Supergravity (World Scientific, 1984); H. Haber and G. Kane, Phys. Rep. 117,75
(1985); X. Tata, lectures at TASI 1995 (to be published); J. Bagger, ibid.
3. C.S. Aulakh and R.N. Mohapatra, Phys. Lett. B119, 316 (1982); L.J. Hall and M. Suzuki, Nuc!.
Phys. B231, 419 (1984); F. Zwirner, Phys. Lett. B132, 103 (1983); S. Dawson, Nuc!. Phys.
B261, 297 (1985); R. Barbieri and A. Masiero, Nuc!. Phys. B267, 679 (1986); S. Dimopoulos and
L.J. Hall, Phys. Lett. B207, 210 (1987); L. Hall, Mod. Phys. Lett. A5, 467 (1990); KS. Babu and
R.N. Mohapatra, Phys. Rev. D42, 3778 (1990).
4. V. Barger, G.F. Giudice, T. Han, Phys. Rev. D40, 2987 (1989).
5. K. Enqvist, A. Masiero and A. Riotto, Nuc!. Phys. B373, 95 (1992); G. Bhattacharyya, J. Ellis and
K. Sridhar, Mod. Phys. Lett. A10, 1583 (1995); G. Bhattacharyya, D. Choudhury and K Sridhar,
Phys. Lett. B355, 193 (1995).
6. R. Mohapatra, Phys. Rev. D34, 3457 (1986); H. Dreiner and G.G. Ross, Nucl. Phys. B365, 597
(1991); R. Godbole, P. Roy and X. Tata, Nuc!. Phys. B401, 67 (1993); K.S. Babu and R.N. Mo-
hapatra, Phys. Rev. Lett. 75, 2276 (1995).
7. K Agashe and M. Graesser, LBL-37823, UCB-PTH-95/33, hep-ph/9510439.
8. B. Brahmachari and P. Roy, Phys. Rev. D50, R39 (1994); Erratum D51, 3974 (1989); see also
J. McCurry, Oxford thesis 1993 (unpublished).
9. A. Smirnov and F. Vissani, Nuc!. Phys. B460, 37 (1996).
10. A. Bouquet and P. Salati, Nuc!. Phys. B284, 557 (1987); B.A. Campbell et a!., Astropart. Phys.
1, 77 (1992); H. Dreiner and G.G. Ross, Nuc!. Phys. B410, 188 (1993).
l1. J. L. Goity and M. Sher, Phys. Lett. B346, 69 (1995).
12. A fuller version appears in V. Barger, M.S. Berger, R.J.N. Phillips and T. Woehrmann, Madison
preprint MADPH-95-910, hep-ph 95l1473, to be published in Phys. Rev. D.
13. H. Dreiner and H. Pois, Zurich preprint ETH-TH/95-30, hep-ph/95l1444.
14. B. de Carlos and P. White, Sussex preprint SUSX-TH/96-003, hep-ph/9602381.
15. N. Cabibbo, 1. Maiani, G. Parisi, and R. Petronzio, Nuc!. Phys. B158, 295 (1979); B. Pendleton
and G.G. Ross, Phys. Lett. 98B, 291 (1981); C.T. Hill, Phys. Rev. D24, 691 (1981); E.A. Paschos,
Z. Phys. C26, 235 (1984); J.W. Halley, E.A. Paschos, and J. Usler, Phys. Lett. 155B, 107 (1985);
J. Bagger, S. Dimopoulos, and E. Masso, Nuc!. Phys. B253, 397 (1985); W. Zimmerman, Commun.
Math. Phys. 97, 211 (1985); M. Tanimoto, T. Hayashi, R. Najima, and S. Wakaizumi, Prog.
Theor. Phys 76, 1098 (1986); K.S. Babu and E. Ma, Europhys. Lett. 3,437 (1987); M. Tanimoto,
Y. Suetake, and K. Seuba, Phys. Rev. D36, 2119 (1987); C.H. Albright and M. Lindner, Phys.
Lett. B213, 347 (1988); Z. Phys. C44, 673 (1989); J. Kubo, K Sibold, and W. Zimmerman, Phys.
Lett. 200, 191 (1989); C.D. Froggatt, I.G. Knowles, and R.G. Moorhouse, Phys. Lett. B249, 273
(1990); B298, 356 (1993); E.M. Fieire, G. Lazarides, and Q. Shafi, Mod. Phys. Lett. A5, 2453
(1990); W.A. Bardeen, M. Carena, T.E. Clark, K. Sasaki, and C.E.M. Wagner, Nue!. Phys. B369,
33 (1992); M. Carena, S. Pokorski, and C.E.M. Wagner, Nue!. Phys. B406, 59 (1993); W. Bardeen,
M. Carena, S. Pokorski, and C.E.M. Wagner, Phys. Lett. B320, 110 (1994); E.G. Floratos and
G.K. Leontaris, Phys. Lett. B336 , 194 (1994).
16. V. Barger, M.S. Berger, and P. Ohmann, Phys. Rev. D47, 1093 (1993); V. Barger, M.S. Berger,
P. Ohmann, and R.J.N. Phillips, Phys. Lett. B314, 351 (1993).
17. CDF collaboration: F. Abe et aI., Phys. Rev. Lett. 73,225 (1994), Phys. Rev. D50, 2966 (1994)
and D51, 4623 (1995).
18. DO collaboration: S. Abachi et a!', Phys. Rev. Lett. 72, 2138 (1994) and 74, 2422 (1995).
19. V. Barger, M.S. Berger, and P. Ohmann, Phys. Rev. D47, 2038 (1993).
20. H. Dreiner and R.J.N. Phillips, Nue!. Phys. B367, 591 (1991).
21. S. Dawson, Nucl. Phys. B261, 297 (1985).
29
Grand Unified Theories From Superstrings
Gerald B. Cleaver
Department of Physics
The Ohio State University
Columbus, OH 43210
ABSTRACT
I discuss how traditional grand unified theories, which require adjoint (or higher
representation) Higgs fields for breaking to the standard model, can be contained within
string theory. The status of stringy free fermionic three generation SO(10) SUSY-GUT
models is reviewed. Progress in classification of both SO(lOh charged and uncharged
embed dings and in N = 1 spacetime solutions is discussed.
SUSY-GUTs and Strings
Elementary particle physics has achieved phenomenal success in recent decades,

resulting in the Standard Model (SM), SU(3)c XSU(2)L xU(l)y, and verification to high
precision of many SM predictions. However, many aspects of the SM point to a more
fUlldamental, underlying theory:
the SM is very complicated, requiring measurement of some 19 free parameters,
the SM has a complicated gauge structure,

there is a naturalness problem regarding the scale of electroweak breaking,
fine-tuning is required for the strong CP problem, and
the expected cosmological constant resulting from electroweak breaking is many,

many orders of magnitude higher than the experimental limit.
Since the early 1980's, these issues have motivated investigation of Grand Unified
Theories (GUTs) that would unite SM physics through a single force at higher tem-
peratures. Superstring research has attempted to proceed one step further and even
merge SM physics with gravity into a "Theory of Everything."
Perhaps the most striking evidence for a symmetry beyond the SM is the predicted
coupling unification not for the SM, but for the minimal supersymmetric standard
31
model (MSSM) containing two Higgs doublets. [1] Renormalization group equations
applied to the SM couplings measured around the Mzo scale predict MSSM unification
at Munif ~ 2.5 X 10 16 GeV. However, this naively poses a problem for string theory,
since the string unification scale has been computed, at tree level, to be one order of
magnitude higher. That is, Mstring ~ gs X 5.5 X 10 17 GeV, where the string coupling
gs ~ 0.7.[2] In recent years, three classes of solutions have been proposed to resolve the
potential inconsistency between Munif and Mstring:
The unification of the MSSM couplings at 2.5 x 10 16 GeV should be regarded as
a coincidence. Munif could actually be higher as a result of
1. SUSY ~breaking thresholds,

2. non~MSSM states between 1TeV and Munif'
3. non~standard hypercharge normalization (a stringy effect), or

4. non-perturbative effects.
Mstring could be lowered by string threshold effects, or
Munif and Mstring remain distinct: there is an effective GUT theory between the
two scales. MSSM couplings unify around 10 16 GeV and run with a common
value to the string scale.
I have been investigating this third possibility. The rationale for this research has
been further strengthened recently by findings suggesting that stringy GUTs and/or
non~MSSM states between 1Te V and Munif are the only truly feasible solutions on the
list (except perhaps for unknown non~perturbative effects). Shifts upward in Munif
from SUSY ~breaking and/or non~standard hypercharges appear too small to resolve
the conflict and string threshold effects in quasi~realistic models consistently increase
Mstring rather than lower it. [3]
The "birth" of string GUTs occurred in 1990, initiated in a paper by D. Lewellen.[4]
wherein Lewellen constructed a four~generation SO(10) SUSY~GUT built from the free
fermionic[5, 6] string. This quickly inspired analysis of constraints on and properties
of generic string GUTs.[7, 8] Following this string GUT research laid dormant until
searches for more phenomenologically viable GUTs commenced in 1993 and 1994. Ini-
tial results during this "infancy" stage of string GUTs seemed to suggest that three
generation string~derived GUTs were fairly simple to build and were numerous in
number. [9, 10] However, eventually subtle inconsistencies became evident in all these
models. The methods used to supposedly yield exactly three chiral generations were
inconsistent with worldsheet supersymmetry (SUSY) and, relatedly, unexpected tachy-
onic fermions were found in the models. Understanding how to produce three gen-
erations consistent with world sheet SUSY spurred the current "maturation stage" of
string GUT research.[ll, 12, 13]
String GUTs and Kac-Moody Algebras
Besides being the possible answer to the Munir/ Mstring inconsistency, string GUTs
possess several distinct traits not found in non~string~derived GUTs. First, string~
derived models can explain the origin of the extra (local) U(I), R, and discrete symme-
tries often invoked ad hoc. in non-string GUTs to significantly restrict superpotential
32
terms.[14J. The extra symmetries in string models tend to suppress proton decay and
provide for a generic natural mass hierarchy, with usually no more than one generation
obtaining mass from cubic terms in the superpotential. All string GUTs have upper
limits to the dimensions of massless gauge group representations that can appear in a
given model. Further, the number of copies of each allowed representation is also con-
strained; there are relationships between the numbers of varying reps that can appear.
These features suggest the opportunity for much interplay between string and GUT
model builders.
At the heart of string GUTs are Kat-Moody (KM) algebras, the infinite dimen-
sional extensions of Lie algebras. [15J (See Table 1.) A KM algebra may be generated
from a Lie algebra by the addition of two new elements to the Lie algebra's Cartan sub-
algebra (CSA), {Hi}. These new components are referred to as the "level" K and the
"scaling operator" La. K forms the center of the algebra, i. e. it commutes with all other
members. Therefore, K is fixed for a given algebra in a given string model and is nor-
malized to a carry a positive, integer value when the related Lie algebra is non-abelian.
La appears automatically in a string model as the zero-mode of the energy-momentum
Table 1. Kac-Moody Algebras -vs- Lie Algebras
LIE ALGEBRA with rank l:
FINITE dimensional algebra
[Hi, Hj] 0; i, j E {I, 2, ... I}

[Hi, E"] a(Hi)E"
E( a, (3) E,,+{3, if a + {3 is a root;
{ ..1...
,,2 a H , if a + {3 = 0; (18)
0, otherwise.
AFFINE KAC-MOODY ALGEBRA with rank l + 2:

New elements in CSA are "LEVEL" K (center of group) and "scaling/energy
operator" La
INFINITE dimensional algebra: m, n E 7L
[H:n,H~] Kmoijom,_n; i, j E {O, 2, ... l + I}

[H~, E~] a(H~)E~+n
E(a,{3)E~1n, if a + {3 is a root;
[E~, E~] { ;2 [a . Hm+n + K mOm,-n], if a + {3 = 0;
0, otherwise.
[K,H;;'J [K,E~J = 0
[Lo,H;;'J
[Lo,E~J
33
operator. These new elements transform the finite dimensional Lie algebra of CSA
and non-zero roots {Hi, E"'} into an infinite dimensional algebra, {K, La, H~, E~J,
by adding a new indice m E 7L to the old elements. A KM algebra is essentially an
infinite tower of Lie algebras, each distinguished by its m-value.
These KM algebras conspire with conformal and modular invariance ( i. e. the string
self-consistency requirements) to produce tight constraints on string GUTs. There are
three generic string-based constraints on gauge groups and gauge group reps. The
first specifies the highest allowed level K; for the ith KM algebra in a consistent string
theory. The total internal central charge, c, from matter in the non-supersymmetric
sector of a heterotic string must be 22. The contribution, Ci, to this from a given KM
algebra is a function of the level Ki of the algebra,
K;dimC i
c = L Ci = LT - ~ 22 . (1)
i i Ki + hi
dim Ci and hi are, respectively, the dimension and dual Coxeter of the associated Lie
algebra, Ci . Eq. (1) places upper bounds of 55, 7, and 4, respectively, on permitted
levels of SU(5), SO(10), and E6 KM algebras.[7, 8]
Once an acceptable level K for a given KM algebra has been chosen, the next
constraint specifies what Lie algebra reps could potentially appear. Unitarity requires
that if a rep, R, is to be a primary field, the dot product between its highest weight,
),R, and the highest root of the KM algebra, W, must be less than or equal to K.
(2)
For example only the 1, 10, 16, and 16 reps can appear for SO(10) at levell. (See table
2.) For this reason adjoint Higgs require K :2: 2 for SO(10) or any other KM algebra.
Table 2. Potentially Massless Unitary Gauge Group Reps

k=l k=2 k=3 k=4
SU(5) c=4 C = 48/7 c=9 c = 32/3
rep h rep h rep h rep h
5 2/5 5 12/35 5 3/10 5 4/15
10 3/5 10 18/35 10 9/20 10 2/5
15 4/5 15 7/10 15 28/45
24 5/7 24 5/8 24 5/9
40 33/35 40 33/40 40 11/15
45 32/35 45 4/5 45 32/45
75 1 50 14/15
70 14/15
75 8/9
SO(10) c=5 c=9 c = 135/11 c = 15

rep h rep h rep h rep h
10 1/2 10 9/20 10 9/22 10 3/8
16 5/8 16 9/16 16 45/88 16 15/32
45 4/5 45 8/11 45 2/3
54 1 54 10/11 54 5/6
120 21/22 120 7/8
144 85/88 144 85/96
210 1
34
Masslessness of a heterotic string state requires that the total conformal dimension,
h, of the non-supersymmetric sector of the state equal one. Hence the contribution hR
coming from rep R of the KM algebra can be no greater than one. For a fixed level K,
hR is a function of the quadratic Casimir, CR , of the rep,
hR = Cr/iJ!~ . (3)
K+h
Requiring hR ~ 1 presents a stronger constraint than does unitarity. For instance,
although all SO(10) rep primary fields from the singlet up through the 210 are allowed
a.t level 2, only the singlet up through the 54 can be massless. In particular, the 126
('annot be massless unless K :::: 5.
Free fermionic string models impose one additional constraint.[12] Increasing the
level K decreases the length-squared, Q;ooo of a non-zero root of the KM algebra by
a factor of K. In free fermionic strings Q;oot at level 1 is normalized to 2 for the long
roots. Thus,
(4)
A state containing such a root makes a contribution of 9j- = If to h. Uncharged free
fermionic contributions to h are quantized in units of andfB !. Thus, masslessness of
gauge bosons constrain K to be a solution of,
1 m n
1= K + 16 +"2; m, n E {O, ;Z+} , (5)
which limits K to values in the set {I, 2, 4, 8, 16}.

In combination the constraints (1) and (5) permit only levels 1, 2, and 4 for SO(10)
and E6 , and, in addition to these, also levels 8 and 16 for SU(5). One result is that
massless 126's can never appear in free fermionic SO(10) SUSY-GUTs; 16's must serve
in their stead.
SUSY-GUTs From Free Fermionic Models
In light-cone gauge, a free fermionic heterotic string model[5, 6] contains 64 real

worldsheet fermions ljJm, where 1 ~ m ~ 20 for left-moving (LM) fermions and 21 ~
m ~ 64 for right-moving (RM). 'lj!1 and 'lj!2 are the LM worldsheet superpartners of the
two LM scalars embedding the transverse coordinates of four-dimensional spacetime;
the remaining 'lj!m are internal degrees of freedom.
The transformation property of a real fermion ljJm around one of the two non-
contractible loops of a torus is expressed by ljJm -+ - exp{ 7r i am}ljJm, and similarly for
the other loop if am is replaced by (3m- The am and (3m are the mth components of
64-dimensional boundary vectors (BVs) a and jj, respectively, and have values in the
range (-1,1].
If ljJm cannot be paired with another real fermion or if it is combined with another
to form a Majorana fermion (one LM and one RM fermion), its phases are periodic or
antiperiodic, i.e. am, (3m = 0 or 1. If a real LM (RM) 'lj!m is paired with another real
LM (RM) 'lj!n to form a Weyl fermion 'lj!m,n == 'lj!n + i'lj!m, the phases may be complex
(i.e. the BV components am,n == am = an and (3m,n == (3m = (3n may be rational).
A specific model is defined by (1) a set of BVs {a}, describing various combinations
of fermion transformations around the two non-contractible loops on the worldsheet
torus, and (2) a set of coefficients, {C(,~)}, weighing the contributions, Z(~), to the
35
partition function, Zferm, from the fermions described by each BV pair (ii, iJ).
Zferm = a~} C Gn Z (g) . (6)

f3E{f3}
The weights C(ff) can be either complex or real (1) phases when either ii or iJ
have rational, non-integer components, but only real phases when ii and iJ are both
integer vectors.
Modular invariance requires that {ii} and {,S'} be identical sets and that if two
vectors, iii and iij , are in {ii} then so too is their sum, iii + ii j . Thus, {ii} and {iJ}
can be defined by choice of some D'-dimensional set of basis vectors {"Vd,
D' D'
ii =L aNi (mod 2) , iJ = L biVi (mod 2). (7)
i=l i=l
Modular invariance also dictates the allowed form of the phase weights:
(8)
where 85 (8;3) is the spacetime component of ii (iJ),

while ki,j is rational and in the
range (-1,1]. There are three mutual constraints on Vi and ki,j:
ki,j + kj,i !y.. V

2 ' J
(mod 2),
Nk
J 1"J o (mod 2), (9)
ki,i + ki,o -8t +!4 V V
t t (mod 2).
N j is the smallest positive integer such that N j Vj = 0 (mod 2).

A complex Weyl fermion lj;n,m in a sector ii carries a U(l) charge Q",(lj;n,m) pro-
portional to O<m,n:
(10)
N is the fermion number operator and has eigenvalues 0, l. Each ii yields a set
of states that are excitations of the vacuum by various modes of the real {lj;m'} or
complex {lj;m,n}. These states, therefore, carry differing charge vectors Q",. Together,
the charges of all states in all sectors form a lattice upon which the roots and weights
of an algebra can be embedded. The BV s j3 contribute a set of GSO operators that
project out certain states in each sector: for a state in a sector ii = ai Vi to survive,
its charge vector Q", must separately satisfy the relation,
(11 )
for each basis vector V j .

Consider now level-K SO(10) (henceforth denoted SO(lO)K) models. As the prior
section showed, the only allowed levels are 1, 2, and 4 corresponding to Q;oot = 1,
1/2, and 1/4, respectively. Level-2 and level-4 algebras require charge lattices of di-
mension greater than the rank of SO(10), i.e. more than five associated U(l) charges
are required for each embedding. For example, the minimal level-2 embedding re-
quires a six-dimensional charge lattice, with charge vectors for the five SO(10) simple
roots given by (0,0,0,1,0,0), G,-~,-~,-~,O,O), (0,0,1,0,0,0), (O,~,-~,O,-~,~),
36
and (O,!, -!, O,!, -!). The extra degree of freedom on the lattice corresponds to an
additional U(l) algebra.
Although the total central charge for SO(lOhxU(l) is 10, the charge lattice for
SO(10h xU(l) only yields a central charge of 6 (since each complex fermion contributes
1 to the central charge). Additional central charge must come from unpairable real
fermions (URFs), i.e. real fermions that cannot form Weyl or Majorana fermions.[5]
URFs assume the role of increasing the central charge without increasing the number
of local U (1) charges. It is in this manner that free fermions can match the effect of
increasing the level of a KM algebra.
Lewellen demonstrated that the smallest possible URF set is formed from 16 real
fermions (containing a central charge of 8). This set can contribute half of its cen-
tral charge to realize the required SO(lOhxU(l) central charge of 10. (Existence of
a remaining URF central charge of 4 = 8 - 4 denotes the presence of a discrete sym-
metry among the URFs.) Lewellen's SO(10h embedding presents an example of how
free fermionic representations of higher level KM algebras involve both charged and
uncharged sectors. All of the "infancy stage" attempts at three generation SO(10h
models[9, 10] involved both the minimal (six-dimensional) SO(10hxU(1) charge em-
bedding and the minimal c = 8 URF set. One direction of my current reRearch is to
proceed beyond these minimal embeddings and comprehensively classify and investi-
gate the further possible SO(10h charged and uncharged embeddings. Each physically
inequivalent choice of charged and uncharged embeddings should define a new class of
SO(10h models. When my investigation of S0(10h models is complete, I will proceed
on with similar treatment of SO(10k One important requirement for all such models
is that they have N = 1 spacetime SUSY, which is the topic of the next section.
Classes of N=l Spacetime SUSY Models
In D-dimensional heterotic free fermionic models, the 3(10 - D) real internal LM

fermions (henceforth denoted by Xl rather than 'ljJm) non-linearly realize a worldsheet
SUSY through a supercurrent of the form[19, 5]
TF = 'IjJ/l. 8X/l. + frJKX I X J XK . (12)
The iIJK are the structure constants of a semi-simple Lie algebra .c of dimension
3(IO-D). Four-dimensional models can involve anyone ofthe three I8-dimensional Lie
algebras: SU(2)6, SU(2)SU(4), and SU(3)SO(5). When TF is transported around
the non-contractible loops on the worldsheet, it must transform identically as 'IjJ/l. does:
periodically for spacetime fermions and antiperiodically for spacetime bosons. This
requirement severely constrains the BVs in consistent models. Since frJKX I XJ XK must
transform as 'IjJ/l. does, each BV necessarily represents an automorphism (up to a minus
sign) of the chosen algebra.
The simplest such four-dimensional modular invariant model is non-supersymmet-
ric. Its single basis vector is the all-periodic Yo; therefore it contains only the sectors
V 0 and 0 == V 0 +V 0 (the all antiperiodic sector). The graviton, dilaton, antisymmetric
tensor, and spin-l gauge particles all originate from the 0 sector. Each of the three
possible worldsheet SUSY choices for Lie algebra allows various possibilities for an ad-
ditional basis vector Si that both satisfies the automorphism constraint and contributes
massless gravitinos. Every {V, Si} set generates an N = 4 supergravity model. Addi-
tional basis vectors (with related GSO projections) must be added to reduce the number
of spacetime supersymmetries below four. Ref. [16] showed that neither SU(2)0SU( 4)
37
nor SU(3)0S0(5) algebras can be used to obtain N = 1 spacetime SUSY. This work
also presented two examples of different basis vector combinations (one being the N AHE
set of LMs)[17] that can yield N = 1 for SU(2)6, while it revealed one situation where
presence of a specific basis vector forbids N = 1.
I have finished the work initiated in [16]. That is, I have completely classified the
sets of LM BVs that can produce exactly N = 1 spacetime SUSY (and N = 4, 2, and
o spacetime SUSY solutions in the process). SU(2)6 is necessarily the supercurrent's
Lie algebra, which gives (12) the form of,
+ i L X3JX3J+1X3J+2.
6
TF = 'ljJlloXIl (13)
J=1
Each fermion triplet (X 3J , X3J+1, X3J+2) represents the three generators of the Jth
SU(2). N = 1 spacetime is only possible if the generators for each SU(2) are writ-
ten in the non-Cartan-Weyl basis of (J3 , J1> and J 2 ).
An automorphism of SU(2)6 is the product of inner automorphisms for the separate
SU(2) algebras and an outer automorphism of the whole SU(2)6 product algebra. [6,
16] The only inner automorphism for an individual SU(2) that could yield a massless
gravitino corresponds to one fermion in a triplet being periodic and the other two
being antiperiodic. An outer automorphism can be expressed as an element of the
permutation group P6 that mixes the SU(2) algebras.[16] The elements of P6 can be
resolved into factors of disjoint commuting cycles. These fit into eleven classes defined
by the different possible lengths, nk, of the cycles in the permutation such that L:k nk =
6. The set of these eleven classes (with a set of lengths written as nl . n2 ... ni) is
n E { 1 1 1 . 1 . 1 . 1, 2 1 . 1 . 1 1, 22 1 . 1, 222, (14)
3 1 1 . 1, 32 1, 33, 4 1 . 1, 42, 5 1, 6 }.
The first element in this set, 111111, is the h identity element, while 2 1111 is
the class with cyclic permutation between two SU(2) algebras (which two is indicated
by each class member's J subscripts). For example,
(15)
Similarly, an element of the 2 . 2 . 1 . 1 class permutes two separate pairs of algebras,

e.g.
21,2.23,4.1.1: (X 3, X\ X5 ) ++ (X 6 , X\ X8), (16)

(X 9 , XlO, Xl!) ++ (X 12 , X13, X I4 ). (17)
Of the eleven permutation classes, only those six involving an even number of
disjoint permutations correspond to BVs that can yield massless gravitino~.[16] The
other five would produce gravitino BVs that cannot satisfy all requirements of (9). The
six distinct gravitino BVs are listed in Table 3. (Note that, as with any BV, a tl n twisted
gravitino generator contains components of the form ~ where a and n are relative
primes in at least one component.) I have studied each gravitino generator and applied
all collsistent combinations of unique GSO projections to it.[ll] I have determined how
many of the initial N = 4 spacetime SUSYs survive various combinations of GSO
projections. My findings can be summarized as follows:
1. Only left-moving 7l 2 , 7l 4 , and 718 twists that correspond to automorphisms of

SU(2)6 are consistent with N = 1 in free fermionic models. All other LM tl n twists
obviate N = 1. Thus, neither gravitino generators 55 and 57 (both containing
38
Table 3. Distinct Free Fermionic Gravitino Boundary Vectors
BV Class Gravitino Boundary Vectors Allowed SUSY
111111 SI {1,1 (1; 0, 0)6} 4, 2, 1, 0

2211 S3 {1,1 (0, 1; _~, ~)2 (1; 0, 0)2} 4, 2, 1, 0
3 .1.1.1 S5 {1,1 (t, 1; -~, 0, O,~)
A A A
(1; 0, 0)3} 4, 2, 0
33 S7 {1,1 (~, 1; -~, 0, 0, ~)2} 4, 2, 0
42 S9 {1,1 (0 i l' 3 i i 3) (0, 1; -~, ~n 4, 2, 1, 0
'2' '-4'-4'4'4
51 SlO = {1,1 e5' 5'3 l''-5'4 -5'2 0, 0'5'"5
A A A A A
2 4)
A
(1; 0, On 4, 0
?L6 twists), nor SlO (containing ?LID twists) can produce N= 1 spacetime SUSY.
S5 and S7 only result in N = 4, 2, or 0, whereas SlO yields N = 4 or O.
2. N = 1 spacetime SUSY is possible for SI, S3, and S9. Six general categories of
GSO projection sets lead to N = 1 for SI, while three do for S3, and one does for
S9. The GSO projections in all these sets originate from LM BVs with ?L 2 , ?L 4 ,
and ?L s twists.
I have fully classified the ways by which the number of spacetime supersymmetries
in heterotic free fermionic strings may be reduced from N = 4 to the phenomeno-
logically preferred N = 1. This means that the set of LM BVs in any free fermionic
model with claimed N = 1 spacetime SUSY must be reproducible from one of the three
specific gravitino sectors in the set {SI, S3, S9}, combined with one of my LM BV sets
whose GSO projections reduce the initial N = 4 to N = 1. The only variations from my
BV s that true N = 1 models could have (besides trivial reordering of BV components)
are some component sign changes that I have shown do not lead to physically distinct
models.
Prior to my present SO(10h research, only the gravitino generator SI had been
used in N = 1 models. Reduction to N = 1 spacetime SUSY had always been ac-
complished through GSO projections from the NAHE set of LM BVs.[17] Thus, my
new N = 1 solutions should be especially useful for model building when the NAHE
set may be inconsistent with other properties specifically desired in a model. This,
indeed, appears to be the situation with regard to current searches for consistent three
generation SO( 10) level-2 models, at least when Lewellen's original minimal charged
and uncharged embeddings are chosen.
Concluding Comments
The result of the 1994 "infancy stage" of the search for string-derived three gen-
eration S0(10) SUSY-GUTs was essentially a no-go theorem for a particular choice
of charged and uncharged SO(10h embed dings that was combined with the standard
gravitino generator, SI. String GUTs has in the past year advanced to a more "ma-
ture" stage, with classification of non-minimal charged and uncharged embeddings now
underway. Further, complete classification of all directions to obtaining N = 1 space-
time SUSY has been completed. Relatedly, new classes of SO(10h models are now
under investigation. In parallel fashion, SO(10)4 models will also be examined. If three
generation free fermionic SO(10) SUSY-GUT models do exist, they will eventually be
found through the systematic search now in operation.
39
ACKNOWLEDGMENTS
G.C. wishes to thank the organizers of ORBIS SCIENTIAE 1996, in particular

Behram N. Kursunoglu, for producing such a stimulating and enjoyable conference.
References
1. P. Langacker, preprint UPR-0512-T (1992).

2. V. Kaplunovsky, Nucl. Phys. B307 (1988) 145;
V. Kaplunovsky and J. Louis, "On Gauge Couplings in String Theory," hep-th
/9502077; UTTG-24-94; LMU-TPW-94-24.
3. For a review of the feasibility of the various proposed solutions see K. Dienes,
"String Theory and the Path to Unification: A Review of Recent Developments"
IASSNS-HEP-95/97; hep-th/9602045.
4. D. Lewellen, Nucl. Phys. B337 (1990) 61.
5. H. Kawai, et. aI, Nucl. Phys. B288 (1987) 1; Nucl. Phys. B299 (1988) 431.
6. 1. Antoniadis, et. aI, Nucl. Phys. B289 (1987) 87; Nucl. Phys. B298 (1988) 586.
7. A. Font, et. aI, Nucl. Phys. B345 (1990) 389.
8. J. Ellis, et. aI, Phys. Lett. B245 (1990) 375.
9. G. Cleaver "Guts with Adjoint Higgs from Superstrings", in the proceedings of
PASCOS '94, May 1994, Syracuse, New York, p. 223; "SO(10) SUSY-GUTs Based
on Superstrings," in the Proceedings of DPF '94, August, 1994, Albuquerque,
New Mexico, p. 1442.
10. Chaudhuri, et. al" "Fermion Masses from Superstrings with Adjoint Scalars,"
Fermilab-PUB-94/137-T; "String Models for Locally Supersymmetric Grand Uni-
fication, " in the Proceedings of DPF '94, August, 1994, Albuquerque, New Mex-
ico, p. 1393.
11. G. Cleaver Nucl. Phys. B456 (1995) 219; What's New in Stringy SO(10) SUSY-
GUTs," in the Proceedings of Strings '95, March 1995, Los Angeles, California.
12. S. Chaudhuri, et. aI, Nucl. Phys. B456 (1995) 89; "Three Generations in the
Fermionic Construction," Fermilab- PUB-95 /349-T.
13. Aldazabal, et. ai, "Standard Grand Unification from Superstrings," in the Pro-
ceedings of SUSY '95, Paris, May 1995; Nucl. Phys. B452 (1995) 3; "Building
GUTs from Strings," FTUAM-95-27.
14. L. Hall and S. Raby, Phys. Rev. D51 (1995) 6524;

Anderson et. ai, Phys. Rev. D49 (1994) 3660.
15. V. Kac, Infinite Dimensional Lie Algebras, (Birkhauser, Boston, 1983);
V. Kac editor, Infinite Dimensional Lie Algebras and Groups, (World Scientific,
Singapore, 1989).
16. H. Dreiner, et. aI, Nucl. Phys. B320 (1989) 401.
17. A. Faraggi, et. aI, Nucl. Phys. B335 (1990) 347;
1. Antoniadis, et. al Phys. Lett. 194 (1987) 231.
40
18. P. Candelas, et. ai, Nucl. Phys. B258 (1985) 46.
19. I. Antoniadis, et. ai, Phys. Lett. B149 (1987) 231.
41
SEARCHING FOR DARK MATTER WITH THE FUTURE LHC
ACCELERATOR AT CERN USING THE CMS DETECTOR
Vasken Hagopian* and Howard Baer

Florida State University
Tallahassee, Florida, 32306
INTRODUCTION
The fraction of visible matter may be as small as about 1% of the total mass of
the universe 1 . The missing dark matter could be baryonic matter in the form of black
holes, Jupiter-like planets, white dwarf stars, etc. (collectively called MACHO's for
MAssive, Compact Halo Objects). This type of dark matter could account for much
of the discrepancy from galactic rotation curves, but would not be able to account for
the amount of dark matter needed on larger scales to explain galactic clustering or
the simplest inflationary cosmological models, which require the matter density of the
universe to be at the critical density. The missing dark matter could also be composed
of elementary particles. The latter come in two different forms: hot dark matter, such
as massive neutrinos which would be moving at speeds close to the speed of light,
and cold dark matter, such as axions or the lightest superpartner of supersymmetric
(SUSY) theories, which would typically be moving at non-relativistic velocities. In
R-parity conserving SUSY theories, the lightest SUSY particle (LSP) has to be stable.
A neutral LSP would interact weakly, like a neutrino, and would be very difficult to
observe directly. The new accelerator to be built at CERN, called the LHC, should
have sufficient energy to produce SUSY particles in abundance. Two detectors, CMS 2
and ATLAS3, should be able to detect the presence of SUSY particles, and will engage
in the search for supersymmetry. Thus, if dark matter is primarily made up of LSP's,
experiments at LHC should be able to discover dark matter indirectly, by finding an
excess of events with missing transverse energy, due to LSP's that escape the detector.
WHY SUSY IS NEEDED

The world of elementary particle physics is explained by the Standard Model (SM).
The SM predicts the existence of elementary particles and their interactions, but is not
a complete theory. Even though experimental results agree well with predictions of
the SM to fairly high accuracy, some fundamental properties cannot be explained. In
the SM, particles obtain mass via the Higgs mechanism. Unfortunately, in the SM the
Higgs mass is unstable, due to the so-called quadratic'divergence. In SUSY theories, the
extra symmetry results in the cancellation of quadratic divergences in the Higgs mass.
Furthermore, if one tries to embed the SM into a Grand Unified Theory (GUT), then
one finds the "gauge couplings" will have the appropriate unification only if the GUT is
supersymmetric. In this case, starting with appropriate GUT scale boundary conditions
for the gauge couplings, the resulting weak scale value of sin 2 Ow matches well with
experimental measurements. Finally, we remark that in the SM, electroweak symmetry
breaking is put in "by hand"; however, in supersymmetrized versions of the Standard
Model, electroweak symmetry breaking is a derived consequence of supersymmetry
breaking at or beyond the unification scale 4 5
43
IfSUSYparticles are created in the laboratory (or at the beginning ofthe universe),
they will necessarily be created in pairs in R-conserving theories. R-parity quantum
numbers of +1 are assigned to all SM particles, and -1 to all superpartners; thus, R-
parity is a multiplicatively conserved quantum number. Conservation of R-parity also
means that the lowest mass SUSY particle is stable. This will be a correct hypothesis as
long as baryon and lepton number conservation holds. So far the experimental results
on proton lifetime are consistent with baryon number conservation. In an accelerator,
pairs of SUSY particles will decay through a cascade which ultimately terminates with
the LSP, which will then escape the detector.
SUSY MODEL
The most conservative approach to SUSY model building is to take the highly
successful SM of particle physics and supersymmetrize it. This leads to the Minimal
Supersymmetric Standard Model (MSSM). In the MSSM, for each SM particle, such
as a quark or a lepton, there is a corresponding SUSY particle. Table 1 lists the
SUSY partners for each particle; the names of SUSY particles have the letter "s" in
front of them or are modified at the end by "ino." One important difference between
SUSY partners and the SM particles is that their spin differs by ~ unit of h. For
example, the gluino has spin ~ while the gluon has spin 1. Supersymmetry must
necessarily be a broken symmetry, meaning that SM and SUSY partners must have
different masses. The exact mechanism for SUSY breaking is unknown; in the MSSM,
one introduces "soft-supersymmetry breaking" terms by hand into the theory, which
should parameterize the effects of supersymmetry breaking in a more fundamental
theory. In the literature, one often sees the parameter set (mo, ml/2, A o, tan fJ and
sgn(p,.) Here, mo is the common mass of all spin-O particles at the GUT scale, ml/2
is the common mass of all spin ~ SUSY particles at the GUT scale, Ao is a common
GUT scale SUSY Lagrangian trilinear coupling, tanfJ is the ratio of Higgs field vacuum
expectation values, and p, is the superpotential Higgs mixing term (its magnitude is
specified by constraints from radiative electroweak symmetry breaking). When one
specifies the above parameter set, then one can derive (via 26 coupled renormalization
group equations) what the physical masses and couplings of the complete spectrum of
SUSY particles should be 5 .
Table 1. Minimal Supersymmetric Standard Model (MSSM)
Standard Model States SUSY Partners

quark squark(q)
lepton slepton(i)
neutrino sneutrino(ii)
gluon gluon(g)
charged higgs chargino
charged weak boson CW.)
light higgs
heavy higgs neutralino
pseudoscalar Higgs (i.)
neutral weak boson
photon
At an accelerator, the pair production of SUSY particles would be accompanied

by production of very many SM particles as well. Figure 1 shows diagrammatically
how a typical production event is modeled. The SUSY particles almost immediately
decay into hadron jets and eventually the LSP. Detectors have to be designed that can
observe the SM particles and have good energy resolution and hermeticity to be able
to detect the LSP's as missing transverse energy.
44
p
1) Hard scattering
2) Convolution with PDF's
3) Initial/ Final State showers
4) Hadronization
5) Beam remnants
Event generation in
LL - QeD
Figure 1: A typical production of SUSY particles.
THE LHC ACCELERATOR and CMS DETECTOR

The Large Hadron Collider (LHC) is a proton-proton collider with an energy of
7 Te V for each proton. The plan is to build this accelerator at CERN in the existing
27 km tunnel of the LEP accelerator in Geneva, Switzerland. The accelerator is
expected to be completed in the year 2004. The particle detectors are to be built at the
intersection regions, where the protons collide with an energy of 14 TeV. The planned
luminosity of the accelerator is = 1034 cm- 2 sec-I. For the computations reported
below, we have assumed an integrated luminosity of 10 fb- 1 .
Total Weight 14~OO t.

Overall diameter 14.60 m
Overall tength 21 .60 m
Magnetic field 4 Toslo
Figure 2: The CMS detector for the LHC at CERN.
The Compact Muon Solenoid detector (CMS) will be at one of the intersection
areas. The CMS detector shown is shown in figure 2 and the schematic in figure 3
45
shows the detector with an event. The basic components of the detector are as follows:
the intersection region is inside a silicon vertex detector. Outside the vertex detector
is the charged particle tracker. Both Vertex and Tracker are in a 4 tesla magnetic
field. Outside of the Tracker is the Calorimeter, which has both electromagnetic and
hadronic portions. Outside of the Calorimeter are the muon toroids. This detector has
very few dead regions and very good energy resolution. Since the detector is still being
designed, for our simulations, a simplified one was assumed. We simulate calorimetry
covering -5 < ", < 5 with cell size t:l.", X t:l. = 0.05 X 0.05. We take the hadronic energy
resolution to be 50%/VE EB 3% for 1",1 < 3, where EB denotes addition in quadrature,
and to be 100%/VE EB 7% for 3 < 1",1 < 5, to model the effective PT resolution of the
forward calorimeter including the effects of shower spreading. We take electromagnetic
resolution to be 10%/VE EB 1%. To detect SUSY particles, it was assumed that the
trigger will require 3 or more hadronic jets, with each jet having a transverse energy
above 100 GeV and a missing transverse energy of 100 GeV or more. In addition,
each SUSY candidate event will also have about 30 minimum bias events that form
the background during each beam bunch crossing; we neglect these in our simplified
approach.
--------------------------------------------------~----~ -
Figure 3: A schematic diagram of an event in the CMS detector.
SUSY SEARCHES
The program ISAJET version 7.166 was used to generate the events and apply
the resolution function and the trigger requirement. Even though the minimal SUSY
46
model was employed, there are still several variables that need to be specified. Figure
4 is a plot of the physical gluino and squark masses in the mo vs. ml/2 plane for a
typical combination of parameters. The excluded regions are both experimental from
lower energy data and theoretical. Changes in the parameters Ao, tan (3 and sgn(fL)
hardly changes the gluino and squark masses (but do change the excluded regions and
other super-particle masses). CMS can probe the region below the solid contour; the
maximal reach of LBC is attained by examining events with jets, missing ET and a
single isolated lepton 7 Figure 5 shows the corresponding reach plot, after mapping into
the more traditional gluino mass versus squark mass plane. The upper dashed curve is
the upper mass limit that CMS can probe.
400 ~~~~--____~____~~
Ii (1000)
q(500)
o
fia (GeV)
Figure 4: Universal scalar mass versus gaugino mass. The superimposed curves are the
gluino and squark masses at the weak scale. The solid line labeled "Reach," is the limit
of experimental sensitivity.
fig (GeV)
Figure 5: Gluino versus squark mass at the weak scale. The dashed curve is the upper
mass limit CMS can detect.
47
CONCLUSIONS
If SUSY is valid, then we should be able to observe squarks and/or gluinos up
to masses of about 1500-2000 GeV /c 2 , with 10 fb- 1 of data. The detector used in
the Monte Carlo computations is ideal, so in reality, the mass limits will be somewhat
different, but not by very much. If we observe SUSY particles, then this can explain
at least part of the dark matter that was produced in the first second of the Big Bang.
At this conference we also heard a talk by Dr. R. Arnowitt, where he gave convincing
arguments that SUSY particles cannot have masses above 700 Ge V / c 2 or so. Other
groups find similar values. The inescapable conclusion is that, if SUSY particles are
not found at the LHC, then weak scale, R-parity conserving supersymmetry is DEAD.
ACKNOWLEDGEMENTS
We would like to thank C-H. Chen, F. Paige and X. Tata for various discussions.
We also like to thank S. Blessing and S. Hagopian for helping with this paper. We
would especially like to acknowledge Dr. Chih-hao Chen, whose PhD. dissertation was
the basis of the SUSY plots. This research was supported in part by the US Department
of Energy.
REFERENCES
* Based on the presentation given by Vasken Hagopian at Orbis Scientiae 1996, in
Coral Gables, Florida, January 1996.
1. For reviews, see E. W. Kolb and M. S. Turner, The Early Universe, (Addison-
Wesley, Redwood City, 1989); G. Jungman, M. Kamionkowski and K. Griest,
SU-4240-605 (1995) (submitted to Physics Reports); see also J. Ellis, CERN-
TH. 7083/93 (1993).
2. CMS Collaboration, Technical proposal, CERN/LHCC 94-38 (1994).
3. ATLAS Collaboration, Technical proposal, CERN/LHCC 94-3 (1994).
4. For a review, see H. Baer et. al., to appear in Electroweak Symmetry Breaking
and New Physics at the Te V Scale, edited by T. Barklow, S. Dawson, H. Haber
and J. Seigrist (World Scientific) 1995.
5. For a review and further references, see M. Drees and S. Martin, to appear in
Electroweak Symmetry Breaking and New Physics at the Te V Scale, edited by
T. Barklow, S. Dawson, H . Haber and J. Seigrist, (World Scientific) 1995.
6. F. Paige and S. Protopopescu, in Supercollider Physics, p. 41, ed. D. Soper (World
Scientific, 1986); H. Baer, F. Paige, S. Protopopescu and X. Tata, in Proceed-
ings of the Workshop on Physics at Current Accelerators and Supercolliders,
ed. J. Hewett, A. White and D. Zeppenfeld (Argonne National Laboratory,
1993).
7. H. Baer, C-H. Chen, F. Paige and X. Tata, Phys. Rev. D52, 2746 (1995) and
FSU-HEP-951215 (1995) (Phys. Rev. D, in press).
48
A Scale Invariant Superstring Theory
With Dimensionless Coupling To
Supersymmetric Gauge Theories
M. Awada and F. Mansouri

Physics Department, University of Cincinnati, Cincinnati, OH 45221
Abstract
We show that there exist a unique dimensionless coupling between abelian su-
persyulllletric gauge theories and a superstring theory that respects all expected
symmetries. The coupling is expressed in terms of chiral currents and superfields
in superspace. The natural coupling to the superstring gives rise to a new ob-
servable that is "stringy" in nature and has no analogue in non-supersymmetric
gauge theories. We compute the expectation value of this "stringy" observable
and show that its regularization leads to the kinetic term of a new superstring
theory that is space-time scale invariant. We suggest a mechanism for breaking
dynamically the scale symmetry, which provides the string with tension and leads
to the Green-Schwarz theory.
1 Introduction
In the last two years, there have been significant developments in N = 1 and N =2 super-
symmetric gauge theories [lJ. In particular a mechanism for confinement was provided
by the condensation of monopoles. In finite supersymmetric gauge theories, such as
N=4 super Yang-Mills theory, and a class of N=2 super Yang-Mills theories coupled
to N =2 matter the corresponding beta functions vanish and the issues of confinement
and asymptotic freedom become unclear.
In ordinary gauge theories, the issues of confinement and asymptotic fn~edom can
be studied using a gauge invariant observable known as the Wilson loop [2,:3J. This ob-
servable is characterized by a dimensionless coupling constant and has found a variety
of applications in the study of gauge theories, ranging from phenomenology to topolog-
ical field theories [4J. One remarkable feature of this observable is that in lattice gauge
theories, strong coupling expansion shows that it produces the Area law suggesting
confinement. In a dynamical sense the area law corresponds to a string theory. It is
reasonable to expect that string theory will playa role in QeD confinement. Whetlwr
string theory could be equivalent to QCD at least in some limit, say large N, is still
debatable. This is because the behavior of the standard string models qualitatively
disagree with QCD at short distances. In particular, the exponential fall-off behavior
of the Nambu-Goto string scattering amplitudes contradict the power fall-off behavior
of the observed scattering amplitudes in deep inelastic scattering at very high ener-
49
gies. The absence of scale which is characteristic of power law behavior suggests that
a string theory consistent with QCD at short distances should have long range order
i.e no scale. Pursuing this end, Polyakov [5] considered modifying the Nambu action
by a renormalizable scale invariant curvature squared term (rigid strings). The theory
closely resembles the two dimensional sigma model where the unit normals correspond
to the sigma fields. In the large N approximation, this model does not undergo a phase
transition. Polyakov suggested adding a topological term to produce a phase transition
to a region of long range order. In another approach [6], the rigid string was instead
coupled to long range Kalb-Ramond field [7]. Since spin systems in two dimensions may
exhibit a phase transition with the inclusion of long range interactions, it is natural to
expect a similar behavior for rigid strings with long range Kalb-Ramond fields. Indeed
it was shown [6] that there is a phase transition to a region of long range order in the
leading and subleading order of large N approximation, where N is the dimension of
space-time.
Scaling is a vital property of any QCD string. Another way to look for such a scale
invariant string theory is to examine whether there exist a natural coupling to gauge
theories through some dimensionless coupling constant. It can be easily seen that there
is no consistent dimensionless coupling with at most two derivatives between strings
and non-supersymmetric gauge theories except for the parity violating expression
ie J
dO"F* . ( I)
In this expression, dO" is the surface measure, and F* is the dual of the gauge field
strength. However, there do exist higher derivative dimensionful couplings between
strings and gauge fields [8]. Therefore, it is of interest to explore whether supersym-
metric gauge theories can couple to a superstring theory with a dimensionless coupling
strength.
Recently [9] ,we showed that in abelian supersymmetric gauge theories,in addition
to an explicit chiral superfield expression for the supersymmetric Wilson loop in su-
perspace, there exist a unique dimensionless coupling to superstrings that respects all
the required symmetries. The coupling is expressed in terms of chiral currents and
superfields in superspace. The chiral current has support only on a two surface. If the
surface has a boundary, the natural superstring coupling gives rise to a new observable
that is "stringy" in nature and has no analogue in non-supersymmetric gauge theories.
The aim of this letter is to uncover the structure of the kinetic term of the super-
string theory which leads to the above dimensionless coupling. We accomplish this by
computing the expectation value of a "stringy" observable. Regularization and classical
renormalization lead to a spacetime scale invariant superstring theory characterized
with a dimensionless coupling constant. The lack of scale in this theory makes it fUll-
damentally different from the Green-Schawrz superstring or the heterotic string [10]. A
link between this theory and the known dimensionful theories could be established. We
suggest a mechanism for a dynamical breaking of the global space-time scale symmetry.
2 Coupling to Supersymmetric Gauge Theories

We begin with reviewing some of the results of [9]. For clarity of presentation, we
give the details for the supersymmetric Maxwell's theory. The generalization to the
supersymmetric non-abelian gauge theories will be presented in a forthcoming article.
Throughout most of this paper we will use the superspace t.wo component notation.
The coupling of superstrings to supersymmetric abelian gauge theories is expressed
50
by a new interaction term which, unlike the exponent of the supersymmetric Wilson
loop, is not of topological but of dynamical origin that has support only on the two
surface [9] :
(2)
where hab is the metric on the two surface, h is its determinant, and C.;'b(O are the
components of a manifestly supersymmetric invariant spinor tensor constructed from
the supersymmetric vielbeins given below. The quantities W,,(x(O, O(~)) are the su-
persymmetric invariant abelian chiral superfields representing the field strengths, and
h.c denotes hermitian conjugation. The interaction (2) is also invariant under tlH' local
scale transformation hab ~ A(Ohab of the world sheet metric.
The supersymmetric vielbeins are:
v~" = iJax""(O - ~(O"(OOaO"(O + O"(OOaO"(O)

v~ = OaO"(O (:3)
v~ = OaO"(O.
They are invariant under global space-time supersymmetry transformation rules defined
1 .
8x""(~) = '2(f"O"(O + f"O"(O)
80"(0 = f" (4)
80"(0 = f".
The requirement that, e.g, in four dimensions, the coordinates 0 satisfy the Majorana
condition demands that f be a Majorana. Thus we define:
(5)
The field contents of the supersymmetric Maxwell's theory are given by W,,(x,O)
[11] which satisfy the chirality conditions:
(6)
D"W" = D"W".
The W's are determined in terms of an unconstrained vector superfield V:
. .
W = -I [)2 D V . w = !:.. D2 D . V (7)
" 2 ", " 2 "
which are solutions of (6). The W',s are invariant under the gauge transformation
8V=i(A-A) (8)
where A (A) is a chiral (anti-chiral) parameter superfield. The component expansion

of V and W" in the Wess-Zumino gauge are respectively,
V = (0,0,0,0, A"", 1/J", 1/J", D) (90. )
W" -.' II{J r

- 1/)" - U )e;{J -
.II D
We; + 2"U
1
u",,'f/
01."
112>l (
9b )
where If; is the superpartner of the gauge field A"", j,,{J = ~O(""A3) is the Maxwell's
51
field strength and D is an auxiliary field. An important property of the Wa (Wi is
that it is invariant under the chiral (anti-chiral) supersymmetry transformations of the
component fields:
[j,,, W" = [j,o, Wi> = 0 . (lOa)

[jA"" = i(f"1t'" + /'1//
[j1/)" = fi3 j"i3 + if"D (lOb)

1 ..
[j D = 2",,(f"l/'" - f"lV)
We know of no way to construct the interaction term (2) in the absence of super-
symmetry. Moreover, in contrast to the supersyuunetric Wilson loop on a two surface
[9], this interaction cannot be reduced to an expression on the line. It is supersymmet-
ric, gauge, and reparemetrization invariant and characterized by a new dimensionless
coupling constant K which is different from the gauge coupling e. It is also locally scale
invariant on the world sheet.
3 A New Superstring-like Observable

We can now define a new super-gauge invariant "stringy" observable
(II)
If we take the surface ~ to have a boundary, a closed loop C, then the correlation
function of \fI(~) might be useful for describing pair creation and annihilation of loops,
particularly in strongly coupled super QED.
The superstring-like observable can be totally expressed in terms of chiral currents
on the surface. Define
( 12a)
where [j6(Z - z(O) is the chiral delta function defined in the chiral representation to be
= [j4(X - x(~))(B - B(~))2, and
( 12b)
The action (2) then takes the following form [9]:
Sint = Jd 6 z(J"W" + h.c) . (1:3)
which is chiral, manifestly supersymmetric, and gauge invariant.

C::
In the absence of supersymmetry b = 0, and we loose the superstring interaction.
Therefore the superstring-super gauge interaction (2) is unique to supersymmetry.
An important question is whether such a superstring theory or a superstring-like
observable can exist in a D + I dimensional Minkowski world of the form MDXR. This
existence depends crucially on whether or not the closed surface ~ can be embedded
ill MD. For D = :3 the answer is certainly negative in general. However, for simply
connected :3-manifolds such as the three sphere, the embedding exists. Furthermore as
shown and discussed in [12], there is a whole class of non-simply connected manifolds
in which any loop C embedded in them can be thought of as the boundary of a closed
surface ~ in M:J
52
4 The Computation of < \l1( C) >
Since the theory is abelian one can easily show that [13]:
(14)
The average in (14) is taken with respect to the BoltzmanIl factor of the super-
Maxwell gauge action
SSuper Maxwell = Jcf3zW"'(z) W", (z) ( 15)
Using the expression of W" in (9b) OIle finds:
( 16)
inserting this into (14) and doing one of the z integrations we obtain
Clearly (16) is divergent and requires regularization. We will regularize the delta
function by replacing it with
(18)
as f - . O. This regularization method can be interpreted geometrically by the manifold

splitting regularization method where one displaces I; in the first measure infinitesi-
mally away that in the second measure along some unit normal ni'(O. Thus we define
I;, to have coordinates yi' = xi' + mi' where xi' is the coordinate on I;. Inserting (18)
into (17) and taking tllP limit f - . 0 wt> obtain the following action:
( HJ)
where Gab is the inverse of the induced metric Gab on the world sheet:
(20)
[{ab = [{" [{"gab + h.c . (21 )

[{" = habC~b(O' gab = V~Vb",
It is straight forward to show that
(22a)
where
[{a = [{"Va" . (22b)
Further manipulations makes tht> action takes the simple form
(2:3a)
where
(2:3b)
53
Here ( is the covariant antisymmetric tensor. To put our action in a more familiar
form, we restore the 4-component notation, with J1 = 0, ... D - 1 being the space-time
index, and a = 1, .. .4 is the spinor index. We have
(24a)
where
(24b)
el' = tab Oa Vb
The space-time vectors el' are supersymmetric and of dimension of [mass]. They are
responsible for giving a dimensionless coupling constant " in contrast to the Greeu-
Schwarz superstring action where the e'.s are absent. The supersymmetric action (24a)
is also invariant under the two dimensional general coordinate transformations, global
space-time scale transformations, and local world sheet scale transformations:
(25)
In fact the action (24a) can be further simplified. Define the set of supersymmetric,
locally scale invariant vectors:
ab
I' _ E !'l I'
(j - - - I UaVb (26a)
(G)2
where c is the numerical antisymmetric tensor that transforms as a density (cab =
REab ). The action (24a) becomes
(26b)
The above action is expressed in a second order formalism. The corresponding first
order action has the form
(27)
It is clear that the action (27) is invariant under space-time supersYllunetry, space-time
global scale symmetry, local world sheet scale symmetry, and local reparametrizatiolls.
We can write down another action with the invariance properties of (27). Define the
following the composite field:
(28)
Then we also have
81 = "I
2
47r IE
r
d2~vChhabGabq, . (29)
The full action of our scale invariant superstring theory is therefore:
8superstring = So + SI
The hab equation of motion gives the usual vanishing energy- momentum tensor:
-
'='ab -"21 hab hcd '='cd
- = (:30)
54
-:='ab = - 1 ( "oTaTb
2 + "12(,'ab'i'"")
47l"
_ 1 i'
Ta - 2V a O"w
Quantum loop calculations of the above superstring theory interacting with the
abelian supersymmetric gauge theories (2) will relate the above dimensionless coupling
constants to that of the gauge theory.
The action (29) is very suggestive, in the sense that the Green-Schawrz theory can
result from spontaneously breaking the global space-time scale invariance of by giving
some expectation value to the composite fermionic operator <1>:
,,2
4~ < >= {lstring tension
To summarize, we have obtained a general gauge and supersymmetric invariant ex-

pression for the interaction of the superstring with a supersymmetric abelian gaugp fipld.
When restricted to a surface with a boundary, this expression lead to a superstring-likp
observable which has no counter part in non-supersymmetric gauge theories. By com-
puting the expectation value of this "stringy" observable we discover the kinetic term of
such a superstring theory. This superstring theory is scale invariant and characterizpci
by dimensionless coupling constants. A mechanism in which the scaling symmetry is
spontaneously broken could lead to the Green-Schwarz theory.
This work was supported, in part, by the department of energy under the contract
number DOE-FG02-84ER4015:~.
References
1. [l] N. Seiberg and E. Witten, Nuc!. Phys. B426 19 (1994).
2. [2] K.G.Wilson, Phys.Rev. DlO (1974) 2455.
:3. [:3] A.M. Polyakov, Phys. Lett. B59 (1975) 82; F. Wegener, J.Math. Phys. 12
(1971) 2259
4. [4] E. Witten, Comma. Math. Phys. 117, :35:3 (1988).
5. [5] A. Polyakov, Nue!. Phys. B268 (1986) 406, also Gauge fields, and Strings,
Vol.:3, harwood academic publishers,
6. [6] M. Awada and D. Zoller, Phys.Lett Bn5 (1994) 115 ;ibid 119 M. Awada,
Phys. Lett. B:351 (1995) 46:~ ;ibid 468, M. Awada, D. Zoller, and .J.Clark, Phys.
Lett. B:352 (1995) 428
7. [7] M. Kalb, and P. Ramond Phys. Rev. D Vol.9 (1974) 22:37
8. [8] L.N.Chang, and F. Mansouri, proceeding of the John Hopkins workshop, e(l.
G.Domokos and S.Kovesi Domokos, John Hopkins Univ. (1974).
9. [9] M. Awada and F. Mansouri, UCTP-1O:3-1996-hep-th/9512098.
10. [10] M.Green, .J.Schwarz, and E.Witten Superstring Theory, Cambridge Univer-
sity press 1987.
11. [11] J. Wess and J. Bagger, Introduction to supersymmetry, Princeton University
Press 198:3.
55
12. [12] M.Awada, Comma. Math. Phys. 129;329 (1990).
13. [13] M. Awada and F. Mansouri,in preparation.
56
SUPERSTRING SOLITONS AND
CONFORMAL FIELD THEORY
L. Dolan
Department of Physics, University of North Carolina

Chapel Hill, North Carolina 27599-3255, USA
INTRODUCTION
Theories with conventional superstring spectra can be generalized by allowing

non-vanishing vacuum expectation values (vevs) for all the fields in the low energy
supergravity, "rather than for only the metric tensor. The vevs are given by classical
solutions of the supergravity field equations. These solutions can be elementary in
that they carry only Noether charge, or solitons which have topological charge. The
bosonic supergravity fields, apart from the metric, are described by d-form potentials,
which fall into two categories: they are either NS-NS or RR, the latter occurring
only in Type II superstrings. These two types are distinguished by the d-forms
representing massless states having their origin in the Neveu-Schwarz/Neveu-Schwarz
or the Ramond/Ramond sector of the string theory.
The description of string theory in terms of a two-dimensional conformally in-
variant sigma model (worldsheet) action involves the target space metric tensor (and
possibly the two-form potential and dilaton) which has the string, whose trajectory
is XI-'(u, r), as its source.
In general, for other d-form field potentials with non-zero vevs, the worldbrane
action will be a d-dimensional field theory which couples the d-form supergravity field
potential to a fundamental d-dimensional extended object (a "(d-1)-brane") whose
trajectory is given by XI-'(e a ) for 0 ::; a ::; (d - 1). The formulation of these d-
dimensional worldbrane actions is a goal of defining string theory in its most general
context.
Soliton solutions are non-perturbative, their charge and mass per Unit volume is
large when the original coupling in the theory is small. Nonetheless, such classical
solutions can be found, such as the BPS dyon and the bound state of two magnetic
monopoles and an electron in four-dimensional N = 4 supersymmetric Yang Mills field
theory. Similarly, soliton states exist in string theory. In particular, in Type II string
models, they lead to states which enhance the original spectrum. This is interesting
in that Type II was viewed as an economical model, but one which appeared to fall
57
just short of phenomenological viability. Now these theories are seen to have as many
states as the heterotic string, by the mere choice of an appropriate internal space
to compactify to four dimensions. Furthermore, the existence of classical solutions
allow the matching of spectra in pairs of string theories, related to each other by
exchanging the electric and magnetic properties, i.e. exchanging weak and strong
coupling. String coupling is related to e<tP>, the vev of the scalar dilaton field.
Access from weak to strong coupling is given by < -+ - < . This is an
example of duality symmetry, discrete symmetries which map betweeen strong and
weak coupling of field theory and also of string theory. In this way all string theories
form a web, and the unique structure of this web may have a unique ground state.
The more dimensions compactified the more duality symmetries become visible. The
largest groups appears to be related to affine Es and the hyperbolic Kac-Moody
algebra Eg.
Classical solutions of supergravity theories have made clear the role of duality
symmetry on the level of the effective low energy Lagrangians. When these solutions
are exact in the sense that that they hold for all values of the dimensionful string
parameter 0/, then they correspond to a conformal field theory. Presumably every
conformal field theory leading to a consistent string spectrum can be viewed as cor-
responding to some exact solution of the low energy supergravity field theory. In
this talk we present the connections among the above ideas to demonstrate how new
states occur in string theory. As in most non-linear problems, of course the way to
the ground state, is to pick the theory (i.e. the set of variables) in which most of the
interesting physics is linear, and thus more readily accessible.
SUPERGRAVITY, d-FORMS, AND STRING COUPLING CONSTANT
The Chapline-Manton action for ten-dimensional (10D) N = 1 supergravity cou-

pled to N = 1 Yang-Mills with the antisymmetric tensor field strength H,_wp modified
by Chern-Simons terms, may be derived from the three-point bosonic tree amplitudes
of the heterotic string. The bosonic terms in the effective action are
1= Jd10xe { __l_R -
2K2
~o
2"
Do"D - ~e-V2I<DH H"vp - ~e- ~I<DFa F,.va} .
2 ,.vp 4 ,.v
(1)
Here 0 ::; JL ::; 9. (1) is the two-form potential version of lOD supergravity with H,.vp
given in (3). The Fock space states in the string, whose low energy limit is (1), have
unit norm. R is the conventional curvature scalar
(2a)
R,.v = R~vp (2b)

R:v>.. = ov r :>.. - o>..r~v + r~>..r~v - r~vr~>.. (2c)
r~,. = ~gPV(o>..gv,. + o,.g>..v - ovg>..,.) (2d)

The field strength for the antisymmetric tensor field B,.v generalized to include the
gauge field Chern-Simons three form is
1 K
H,.vp = 3( o,.Bvp + ovBp,. + opB"v-"4 tr[A,.Fvp + AvFp,. + ApF,.v - 2glOApA,.Avl)
(3)
58
where A~ == A~Ta, etc., with trTaT b = 28 ab , the gauge field strength F;:v = 8~A~ -
8vA~ + 91OfabeAtA~, and 910 is the IOD gauge coupling constant.
In terms of dimensionless fields 9~v, ' == v'2/'i,D, and the length dimension [L]-l
fields A~ == 910 A~ and T3~v == J;.9~oB~v, the effective action (1) is
1=_1_ Jd 10 x C;;9
2/'i,2 V -y
{-R - ~82 '/"8~'//
~'I' 'I'
- ~~e-'1i
12 9to ~vp
1i~vp - ~e-H'P!:
29~0 ~v
:F~va}
(4)
where
1 3
ll~vp = 3"( 8~T3vp + 8vT3p~ + 8pT3~v-2..(ii tr[A~:Fvp + Av:Fp~ + Ap:F~v - 2ApA~Av])
6 2
= ..(ii91OH~vp, (5)
:F;v == 8~A~ - 8vA~ + fabeA~A~ = 91O F;v

In IOD, the dimensions of the fields and coupling constants are as follows:
GlO == 2neV6 rv [L]8, a' rv [Lj2, /'i, == v'87rGlO rv [L]4, 9~v rv [L]O,
mpLANCK
.;rgr rv [L]O,
R rv [L]-2, J d10x.;rgr~ rv [L]O, D rv [L]-4, ' rv [L]O, B~v rv [L]-4, T3~v rv [L]-1,
AI' rv [L]-4, AI' rv [L]-1, 910 rv [L]3, H~vp rv [L]-5, 1il'vp rv [L]-l, F~v rv [L]-5,
:F~v rv [L]-2.
It seems that (4) is parameterized by an arbitrary dimensionless coupling con-

4
stant A == ~. But in fact we can remove 910 from the above Lagrangian by writing
4
= ' + In ~, since (4) is invariant under a shift of the dilaton ' ---+ ' + a, for
constant a, if 910 transforms as 91Oe- ~. Thus the supergravity theory has no free
dimensionless parameter, see (6). The important point is that the expectation value
of the field is not determined in the classical Lagrangian (4). As a quantum theory
it has a one-parameter family of vacuum states[11. Knowing what value the vev of
takes on in the quantum theory is crucial, since the value of the physical gauge
coupling constant, which we will call the square root of the string coupling constant,
-.;g; == et<ol<l>lo> = 9~0/'i,-~ d<ol<l>'lo> depends on which of the classical vacua is
physically relevant.
For ~ e-' = e-, then (4) is
glO
1=_1_
2/'i,2
Jd lO x F9 {-R - ~8
2 I'
8~ - ~
12
e-1i /-,vp 1i~vp - ..jK,
2
e-Hp!:~v :F~va} .
(6)
In order to have the heterotic string IOD action depend on the parameters differently
from (6), we could have chosen a field redefinition iJ~v == 6~' 9~oB~v instead of T3~v ==
*9~oB/-,v in (1). Then instead of (4,5), we have
1= _1_ Jd10x cg {-R - ~8 '8~' - /'i,4 e-' H Hl'vp - ~e-t<l>' p!: :F~va}
2/'i,2 V -y 2 I' 129toa/2 I'Vp 29~o I'V
(7)
59
1
1= -2K2 J d 10 x ~g {
V -y
1 ,l,8!-',I,-
-R- -8 - - -e-'I'H
2!-''I' 'I' 12
1 ;.- H!-'vp_
!-,vp
- a' 1;'
-e-2'1'Fa
2 !-'v F!-'va .
}
(9)
In terms of sigma model variables G!-'v = et g!-'v, i.e. the string metric G!-'v which
appears in the sigma model action
the heterotic string effective action (6) in lOD scales with the dilaton:
1=_1_2K2 Jd10x VrGI lUI

e- 2t/> {-R + 48!-,'I' 'I' - ~1l
,l,8!-',I, 12 !-,vp ll!-'VP - ..jK, Fa F!-'va} ,
2!-'v
(10)
since in general for g!-'v(x) -+ e2A (X)g!-'v(x) == G!-'v(x), we have detg!-,v -+ e2nA detg!-,v
in n dimensions, and
ViYT R(g) = e-(n-2)AJjGf {R(G) + (n - 2)( -n + 1)G!-,v8!-,A8vA}

- 8!-,(g!-'v ViYT8v A) . (11)
Then for n = 10, A = t

we have e!detg!-'v = detG!-'v, so Ji9i
= e-H JjGf,
etc. In (10), we have written the effective action (6) as e- 2t/> times a function that is
invariant under -+ + a constant.
So the overall factor of e- 2t/> in (10) allows us to interpret the vev of et/> as the
loop counting parameter, i.e. the square of the string coupling constant. This is the
familiar statement that the string coupling constant is precisely the expectation value
of a field - the dilaton field:
gs == e<olt/>(x)lo> . (12)
Similarly, in terms of sigma model variables, (9) is
1= _1_ 2i> {-R+48 18!-'1_ ~1l ll!-'vp_ a'Fa F!-'va}.

2K2 Jd10x VrGle-
lUI !-''I' 'I' 12 !-,vp 2!-'v
(13)
In order to eliminate K altogether from (9), we can "redefine the zero" of , i.e. define
== cp - In c?r' so that (13) is
1=_1_
2a,4 Jd10x VrGIlUI
e- 2. In terms ofthe metric G, dilaton
f"V
, two-form B, and gauge field strength F, the effective action (10) is
where we have made use of the notation of differential forms.

Let B!-'1!-'2 ...!-'P be an antisymmetric tensor field with p indices defined on a man-
ifold M of dimension n, i.e. 1 ~ 1-1. 1-2, ,I-p ~ n. Mathematicians call this field
60
a p-form, i.e. as a differential form of degree p. The exterior derivative operator d
is defined such that acting on any differential form such as a p-form , it gives a
(p + I)-form d whose components are
For ego
F=dA
H=dB
A differential form is closed if d = O. A differential form is exact if = dO'..

d2 O'. = 0 always, i.e. every exact form is closed d2 = O.
All massless lOD supergravity theories, including the low energy limits of the
heterotic and Type II strings, contain the following terms:
1 =1-
2K2
J 10 ~ -R--(o)
d xv-g 1 {- -1e -</> 1l
2
2
12
2} . (17)
COMMON SYSTEM OF SUPERGRAVITY AND SIGMA MODEL
In order to calculate elementary solutions for classical fields, including singular

points, we can consider the common system[2]
(18)
of the supergravity target space field theory action, and the worldbrane action which
couples an arbitrary d-form potential field to a brane source. We follow the notation
ofref [3].
Let an arbitrary d-form potential field A/-LI/-L2"'/-Ld in D spacetime dimensions
(0 :S: /-L :S: D - 1) interact with gravity gJ1-V and the dilaton :
Here the rank (d+ 1) field strength Fd+l = dAd and a(d) is a constant, a(2) = 1. These
fields couple to an elementary d-dimensional extended object (a "(d-l)-brane") with a
trajectory through the D-dimensional target space given by X/-L(~i) for 0 :S i :S d -1,
with worldvolume metric 'Yij(~) and tension Td, via the worldbrane action:
Sd = Td Jdd~[ - ~H'YijOiX/-LOjXVg/-LVea(d)</>/d + (d; 2) H
X/-LIl X/-L2
1 to iIi2 ... idl
- d! ViI Vi2 . .. ()id X/-LdA /-LI/-L2 /-Ld ] . (20)
Starting from (6), we define B/-Lv = 3A/-Lv so that for F/-Lvp = 0/-LAvp + ovAp/-L + opA/-Lv'
equations (18), (19) are identical. The elementary solutions are solutions of the
classical equations of motion generated by fJS = O.
61
NS BRANES and RR BRANES
The field content of the lOD supergravity theories that are low energy limits
of the IIA, lIB, I SO(32), and heterotic Es x Es and SO(32) string models can be
described as follows: In these string theories, the Lorentz generators separate the
massless states into irreducible SO(8) representations. The NS sector contains the
vector representation 8v and the R sector either an 88 or 8e spinor. Thus the NS-NS
massless sector is 8v x8 v = 1+28+35 t and RR bosons are either 88 x8 8 = 1+28+35 8
or 8 8 x 8e = 8v + 56 v . The fields associated with these representations are d-forms
given by the O-form </> for 1, the I-form Ap. for 8v , the 2-form Bp.v for 28, the 3-form
Ap.vp for 56 t , and the anti-self-dual 4-form Gp.vpu for 35 8 , In addition the metric
tensor gp.v is the 35 t . The fermion fields occur in the NS-R sectors: 8v x 88 = 8e + 56 8
or 8v x 8e = 88 + 56 e, so the gravitino 'l/Jp. is either 56 8 or 56 e, and the spinor field X
is 88 or 8 e .
Type IIA theory is non-chiral and the R sectors on the left and right have different
spinor representations. The massless boson fields in the 2-form potential version of
the theory are gp.v, </>, Ap., Bp.v, Ap.vp. The fermions are 'l/J~ in 56 8 , 56 e, and X in 8 8 , 8e
for 1 :::; i :::; 2.
Type lIB is chiral, so the R sectors have the same spinor representation. The
bosons are gp.v, </>, a, B~v, B~v, Gp.vpu, where </> is the dilaton, and a is the axion, i.e.
the scalar from the RR sector.
The heterotic massless bosonic fields are as in (6): gp.v, </>, Bp.v, and A~. The
Type I SO(32) open and closed string model also contains gp.v, </>, Bp.v, and A~.
We see that Type II string models have bosons from both the NS-NS and RR
sectors. Should any of these d-form potential fields acquire a non-zero vev, they will
couple to a (d-l)-brane extended object, giving rise to ad-dimensional worldbrane
action. Then, in general, the corresponding target space (lOD) spectrum of this
"string" theory will have a modified low energy limit.
We can define the field strength (p+1)-form F associated with each p-form field
potential A as F = dA as in (16). We also introduce the dual field strength, via the
Hodge dual operation * defined with respect to a metric gp.v as
1
* F P.lP.D-(p+l) =
- (p + 1)! g P.lVl g P.D-p-lVD-p-l E ... F:
V1 VD
VD-p . VD'
where EV1 ... VD is the D-dimensional totally antisymmetric symbol with E1...D-l = l.
The dual operation on the field strength corresponds to electric/magnetic duality in
point field theory (d=l, D=4). For our example, the dual field strength corresponds
to a (D - p - 2)-form potential. With a non-vanishing vev, this couples to (D - p - 3)-
brane.
In Type II theory, the NS-NS (d - 1)-branes can couple to d-form potentials
where d = 2,6. Here we have included the dual value as well. For d = 1, the 2-form
potential couples to the onebrane extended object, known as the string. For d = 6,
a 6-form potential couples to the fivebrane, whose trajectory through lOD can be
described by six-dimensional fields XP.(~Q), 0:::; a :::; 5. The classical soliton solution
[3,4,5] called the fivebrane, which describes a 2-form vev, also describes (in its dual
formulation) the vev of this 6-form potential.
Since RR-branes couple to RR field potentials, the RR-branes carry charge of
the RR bosons[6,7]. They have been dubbed Dirichlet p-branes, i.e. D-branes, as
62
they can be shown to correspond to states in type I theory with Dirichlet boundary
conditions[7,8]. In IIA theory, the RR type (d-1)-branes couple to d-form potentials
for d = 1,3,5,7, where the dual pairs are (1,7) and (3,5). For the example d = 3, the
RR boson potential field Al'vp couples to the membrane (twobrane). II B theory has
d = 2,4,6, where d = 4 is a self-dual potential, and (2,6) is a dual pair. As far as the
field potentials are concerned, II B also allows d = 0,8 corresponding to the axion
and its dual. The value d = 9 appears to correspond to a massive lOD supergravity
with non-vanishing cosmological constant; and d = 10 can be interpreted in terms of
type I strings[7].
CONFORMAL FIELD THEORY
In order to bring together the analysis of exact classical solutions of supergravity

theories and states in string theory (i.e. states in one-to-one correspondence with
conformal fields), we discuss the following example of a heterotic string soliton and
its associated conformal field theory. We show how to maintain hermiticity conditions
for the conformal generators.
The solution, in dual coordinates, provides a non-zero vev for a 6-form potential,
which then couples to a NS-NS type fivebrane extended object discussed above. In
the original (non-dual) coordinates, the fivebrane is solitonic, in that it has magnetic
charge. The symmetric fivebrane solution[4,5] is a solution to the equations of motion
for the heterotic string low energy action given in (14). It is exact in the sense that
it is a solution to all orders in a', and the incorporation of the vevs generated by this
solution in a sigma model results in an exact algebraic conformal field theory. To
describe the solution we split xl' = {xP-, yrn} with 0 ::; p, ::; 5; 6 ::; m ::; 9. The metric
tensor field of this fivebrane in the wormhole throat limit is given by
(21)
This solution carries topological 'magnetic' charge proportional to f, and f = 1 is the

minimal solution. The spacetime part of the metric is flat and therefore corresponds
to a free conformal field theory. The transverse part leads to a conformal field theory
where the target space metric describes a wormhole geometry. This can be viewed
as a variation of a Wess-Zumino-Witten (WZW) conformal field theory comprised of
a supersymmetric WZW model with Virasoro central charge Cwzw = n3';2 + ~ and a
supersymmetric Feigen-Fuchs model with cff = ~ + n!2'
where n + 2 = f. The total
central charge of the conformal field theory corresponding to the transverse part of
the fivebrane metric is Ctot = Cwzw + cff = 6. A conformal field theory with N = 4
global superconformal symmetry has been constructed which describes this[9,5]. It is
given in terms of a level n SU(2) Kac-Moody algebra Ja(z), a U(l) current JO(z)
and four free fermion conformal fields 'lj;a(z). The Virasoro current is
which has C = 6 ~~!~l. In order to realize Ctot = 6, we can define a shifted generator
(23a)
63
with a corresponding shift for the superVirasoro generators. In component form, we
have
L nnew -_ L n - 2np
1 JnO+ (-p - 21P) JOn + un,02
A 1 (p
-2 - pp
-) . (23b)
The mixing of the new superVirasoro generators with the Kac-Moody currents is
unchanged save for the mixing with JO(z).
In physical contexts, one may wish to have hermitian generators L~ = L_ n .
This restriction is not necessary mathematically but it can be maintained by the
following constraints. Hermiticity requires that Rep = 0 and Imp = 2Imp. With the
hermiticity restriction, then (23) becomes
Lnew(z) = L(z) + ImpaJO(z) + !pJO(z) + ~2Ppt

z 2z
L~ew = Ln - in Imp J~ + Rep J~ + 8n,0~ppt
cnew = c + 12(Imp)2 . (24)
Thus if we choose Imp = V2~ and L given in (22), then from (24) L new has
cnew = 6.
CONCL USIONS
The role of classical solutions in demonstrating new states in string theory is

discussed. In some cases, string soliton states are massless, which leads to an en-
hancement of the gauge symmetry of the theory. NS-NS bosons and RR bosons
generate massless states in Type II theories[lOl and both can couple to d-dimensional
extended objects (d - l)-branes. This in turn generalizes our conventional concept
of string theory described by a two-dimensional conformally invariant sigma model
worldsheet action to a wider class of theories given in terms of higher-dimensional
worldbrane actions.
REFERENCES
1. M.B. Green, J.H. Schwarz and E. Witten, Superstring theory, Vol. lf32, Cam-
bridge University Press, 1988.
2. A. Dabholkar, G. Gibbons, J.A. Harvey and F. Ruiz Ruiz, Nucl. Phys. B340
33(1990).
3. M. J. Duff, R. Khuri, and J.X. Lu, String Solitons, hep-th/9412184.
4. A. Strominger, Nucl. Phys. B274 253 (1986).
5. C. Callan, J.A. Harvey and A. Strominger, Nucl. Phys. B359 611 (1991); B367
60 (1991) .
6. C. Hull and P. Townsend, Nucl. Phys. B438 109 (1995), hep-th/9410167; Nucl.
Phys. B451 525 (1995), hep-th/9505073.
7. J. Polchinski, Dirichlet-Branes and Ramond-Ramond Charges, hep-th/9510017.
8. P. Townsend, D-branes from M-branes, hep-th/9512062.
9. A. Sevrin, W. Troost, and A. Van Proyen, Phys. Lett. B208 (1988) 447.
10. L. Dolan and S. Horvath, Nucl. Phys. B448 (1995) 220, hep-th/9503210.
64
Comments on Symmetry Breaking Terms
in Quark Mass Matrices
K. Tanaka
Department of Physics, The Ohio State University
174 West 18th Avenue, Columbus, OR 43210, USA
ABSTRACT
It is discussed how the symmetry breaking terms in the fermion mass matrices are
related to the charges of a U (1) x horizonal symmetry beyond the standard model
(SM). Then, we discuss anomaly cancellation through Green Schwarz mechanism
and mention an approach to find many solutions to the anomaly constraints.
We start with a democratic mass matrices of fermions of the same charge q for
three families in order to study the mass hierarchy of quark mass matrices,
M = Mq ( 11 11 1)
1 q = 2/3, -1/3, 1, O.
3 1 1 1
We add a symmetry breaking parameter [1] f. q <1

1
M~M'U
3
1 1
1 ) ,
1 1 1 + f. q
and diagonalize and obtain the form
M,~ G0
0
Zq2 o) .
Zq3
The second and third families of a given charge are massive. To provide mass to the
first family we add the second symmetry breaking term Cq l
o) ,
Zq3
65
and diagonalize and obtain
o
o),
mq3
h
were mql -- ~
3 (_1
31'q + 13 V/I'q2+ 81 Cq2) , m q2 -- ~
3 (1
31'q + 13 V/I'q2+ 81 Cq2) , mq3 --
(3
~ + ~I'q), and C~l = M;c~. The mass matrices are characterized by an over-
all mass term and 2 symmetry breaking parameters I'q and c q .
In order to find a relation between different parametrizations of masses, we use
a symmetry approach proposed by Ftoggatt and Nielsen [2] in which a new Abelian
horizontal gauge symmetry U(l)x is introduced beyond the standard model (SM).
We assume a minimal supersymmetric standard model (MSSM) with a new
horizontal symmetry U(l)x with an Abelian charge X [3-6]. U(l)x can be expressed
in terms of a family dependent (FD) and family independent (FI) parts, U(l)x =
U(l)FD + U(l)FI. We denote the (lefthanded) SM matter chiral superfields as Q U
d L e and label the U(l)FD part in the form
and the U(l)FI part as X o = (ao, bo, Co, do, eo). The Xi charges are traceless.
First consider, for simplicity, the Yukawa coupling matrix Yu for the up-quarks
q = 2/3, and let the X charge of the Higgs doublet be Hu. Assume at the tree level
that the Yukawa interaction (renormalizable) includes only the third family,
where Yt is the Yukawa coupling, and the remaining elements of the Yukawa matrix
Yu is zero at the tree level, but there are extra charges Xij, for renormalizable in-
teractions. We choose only the U(l)FD part or Xij to be symmetric, i.e., Xij = Xji.
Conservation of charge yields
We can restore the X charge conservation by a non-renormalizable term of an

operator of higher dimension with no hypercharge
Qi u j Hu(8 / M)nij(u) + Qi dj Hd(8 / M)'Piij(d) + Liej Hd(8 / M)'Piij(f) + L i TJj Hu(8 / M)nij(V) ,
where 8 is a singlet field of X charge -1 and Xij - nij = O. Actually one introduces a
pair of (8, 8f ) of singlets of respective charges (-1) and (+ 1) which acquire (8) = (8 f )
along a D-fiat direction [4]. Then Yukawa matrix for q = 2/3, for example, is
expressible as
lal + a2 - 2a31
21a2 - a31
la2 - a31
66
To obtain an interesting texture put al = 6, a2 = -4, a3 = -2, then
If we put a texture zero at Yn then [3, 4]
We compared this matrix tij with the previous M2 matrix and related the X charges
with the symmetry breaking terms cq and tEq [7].
We consider the approximate mass ratios at the electroweak scale. We use the
mass values
q = 2/3 u = 0.0051 C = 1.35 t = 180 GeV/c2
q = -1/3 d = 0.0089 s = 0.175 b = 5.3 GeV/c 2
q =-1 e = 0.000511 /-l = 0.10566 T = 1.777 GeV/c 2
q=O Ve = 0.10 X 10- 4 VI' = 0.29 X 10- 2 Vr = 0.12 eV/c 2 [8]
Express the mass ratios rl = mdm2' r2 = m2/m3 of a given charge q in terms

of powers of >.2 or >.4 to find a value of >. in the range>. rv 0.18 - 0.28. We find
the mass ratios of q = 2/3,0 are 0(>.4) with>' = 0.28 and that of q = -1/3, -1 are
0(>.2) with>' = 0.18. The result is given in the Table. The subscripts on the mass
designate the family.
Table
rl = ml/m2 r2 = m2/m3 >. = r;/4 >. = r;/4

q = 2/3 0.0038 0.0075 0.24 0.29
q=O 0.003 0.02 0.23 0.38
q = -1/3 0.051 0.033 0.22 0.18
q =-1 0.005 0.059 0.07 0.24
The mixed anomalies of U(I)x that are cancelled through the Green Schwarz
mechanism [9] are broken not far below the string scale, so the U(I)x symmetry
required to obtain the fermion mass matrices is not expected to survive down to
the electroweak scale. We note that the rough mass ratios that are valid at the
string scale [3] approximately survive with q = 2/3,0 satisfying mdm2 rv mdm 3 rv
0(>.4) = 0.006 and with q = -1/3, -1 satisfying mdm2 rv m2/m3 rv 0(>.2) = 0.032.
The mixed anomaly coefficients [3,6] are
C1 = TR(TxT?), C2 = TR(Tx TJu(2))' C3 = TR(Tx TJU(3)) , CyXX = TR(T;Ty) ,

where
67
3(3ao + do) + Hu + Hd
3(2ao + bo + co)
6( a6 - 2b6 + c6 - d6 + e6) + 2H~ - 2HJ
There are in addition Cx = TR(Tl) and Cg = 3(6ao+3bo+3co+2do+eo)+c~, where
c~ includes contributions from particles that do not appear in MSSM. We cannot
calculate the anomalies Cx and C~ without knowing the extra singlet particles with
chiral X charge other than those in MSSM models. Note because of the assumed
traceless condition of the familiy dependent part of the X charges I:r=l ai = 0 etc.,
only the family independent X charges with subscript zero appear in the anomaly
coefficients. Is Ci = 0 possible? Consider C3 = 3(2ao + bo + co). We assume that
ao, bo and Co have the same sign, so C3 = 0 is not possible.
The anomaly cancellation by Green Schwarz mechanism requires c;j Cj = k;j ki
where the ki are the levels for the Kac - Moody algebra. The unification of gauge
coupling in MSSM requires at string scale [10]
gVg~ = gVg~ = 5/3 g~ki = g~k2 = g~k3

or
5. .
C1 .. C2 .. C3 _
-
. . _
k1 . k2 . k3 - gl
-2.
. g2
-2. -2 _
. g3 -"3. 1. 1
The combinations
C2
8
C1 + C2 - -C3 2(Hu + Hd) - 6(ao + Co - do - eo) = 0,
3
CyXX 6(a6 - 2b6 + c6 - d6 + e6) + 2(H~ - HJ) = 0
yield the following 3 relations where 2h = Hu + Hd, ~ = Hu - Hd,
eo = 2ao - bo,
2h = 3( -an + bo + Co - do),
~ = [(5ao + bo)(ao - bo) + (co + do)(co - do)l![ao - bo - (co - do)],
and CyXX can then be written as
CyXX = 6[(5ao + bo)(ao - bo) + c6 - d6 + 2h~/3] = o.
We divide the solutions into 3 cases (1) Co = do, (2) ao = bo and (3) ao # bo, Co #
do. The X charges are denoted lao, bo, Co, do, eo, Hu, Hd] = lao, bo, Co, do, 2ao - bo, 2h +
z,-z].
(1) Co = do
then 2h = 3( -an + bo) ~ = 5ao + bo
There is the constraint
~ = 2h + 2z = 3( -an + bo) + 2z = 5ao + bo

or
z = 4ao - boo
68
Solution is
lao, bo, y, y, 2ao - bo, 3( -ao + bo) + z, -z].
When ao = bo = x we obtain Ibanez-Ross solution [4].
(2) ao = boo
We have [x, x, Co, do, x, 3(co - do) + z, -z],
2h = 3(co - do),
t::. = -(co + do).
Constraint is
t::. = 3(co - do) + 2z = -(Co + do), or z = -2co + do or Co = (do - z)/2

The solution is
[x, x, (do - z)/2, do, x, -(3do + z)/2, -z].
When do = y, we have Jain Shrock solution [6].
(3) Start with the expression [x + v,"x + 2v, y + w, y, x, 3(v + w) + z, -z],

then 2h = 3(v + Co + do),
t::. = [(6x + 7v)v - (2y + 2)w]/(v + w) = 3(v + w) + 2z,
where
w = -(3v+y+z)/2,
v = (z - 3x)/2.
Solution can be written
[x + v, x + 2v, (y - 3v - z)/2, y, x, -(3v + 3y + z)/2, -zJ.
This is a new solution. Many other solutions are available.
The author wishes to thank B. Kursunoglu and A. Perlmutter for their hospital-
ity.
References
1. See, for example, M. Tanimoto, Phys. Rev. D41, 1586 (1990); H. Fritzsch,
preprint MPI-PHT/94-77; K. Tanaka, Proceeding of Eight Meeting of the
Division of Particles and Fields of the American Physical Society, Albuquerque,
New Mexico, 1994.
2. C. Froggatt and H. B. Nielsen, Nucl. Phys. B147, 277 (1979).
3. P. Binetruy and P. Ramond, Phys. Lett. B350, 49 (1995).
4. L. Ibabez and G. G. Ross, Phys. Lett. B332, 100 (1994).
5. M. Leurer, Y. Nir, and N. Seiberg, Nucl. Phys. B398, 319 (1993).
69
6. V. Jain and R. Shrock, Phys. Lett. B352, 83 (1995).
7. K. Tanaka, preprint reported at First International Conference on Frontiers of
Physics (1995).
8. C. E. Albright and S. Nandi, Phys. Rev., D53, 2699 (1996).
9. M. Green and J. Schwarz, Phys. Lett. B149 117 (1984).
10. L. E. Ibanez, Phys. Lett. B303, 55 (1993).
70
SECTION III - GRAVITATIONAL WAVES
LIGO: AN OVERVIEW
Barry C. Barish
California Institute of Technology

Pasadena, CA 91125
ABSTRACT
The Laser Interferometer Gravitational-Wave Observatory (LIGO) is being developed to detect

gravitational waves emitted from Astrophysical sources. The facility will consist of two widely
separated laboratories housing highly sensitive long baseline interferometers using suspended test
masses. The technique and status of the project are discussed.
INTRODUCTION
The concept of gravitational waves was tirst proposed by A. Einstein as a consequence of the
theory of general relativity. A gravitational wave is a propagating distortion of spacetime which
occurs when masses accelerate. This emission is analogous to the generation of electromagnetic
waves when electric charges accelerate.
Evidence that gravitational waves indeed exist in nature has been provided by the beautiful
experiment of Hulse and Taylor, who measured the orbital period of a binary neutron star system,
over a 15 year interval with great precision. They observed, with 1% accuracy, a decreased in the
- 8 hour period of the orbit by about 10 seconds. This measurement is completely consistent
with expectations from general relativity due to the emission of gravitational radiation.
The present situation experimentally is reminiscent of what occurred after the emission of
neutrinos were proposed as the explanation for observations of apparent missing energy (and
angular momentum) from some nuclear beta decay reactions. That lead to a concerted effort to
'directly observe' neutrinos through their interactions. This finally occurred 20 years later by
Reines and Cowan, and Reines was awarded the Nobel Prize this year for this discovery and
subsequent research.
Since the time of the tirst direct observations, a rich tield of neutrino physics has developed
lxlth to study the neutrino itself (this continues with the search for neutrino mass and oscillations),
and the use of the neutrino as a sensitive probe of fundamental particle physics (quark structure of
nucleon, neutral currents, etc.).
73
For gravitational waves, following indirect observation, sensitive new instruments are now
being developed to directly detect these waves. Two main approaches are being employed -
resonant bar detectors and suspended mass interferometers. In this talk, I concentrate of the
interferometer approach, and particularly those on the earth's surface (there are long range
proposals for such devices in space) and in particular I discuss the U.S. Project (LIGO). There is
also a large French-Italian project (VIRGO).
-1
0.5 0.6 0.7 0.8

T (sec)
Figure 1. Binary neutron star coalescence 'chirp' signal
THE LlGO CONCEPT
LIGO, the Laser Interferometer Gravitational-Wave Observatory(l), is a joint MIT/Caltech

project funded by the National Science Foundation. It will consist of two widely separated sites
(3()(X) km), each having two 4 km arms in an L-shape and under high vacuum. These vacuum
systems will house sensitive suspended mass interferometers which will be used for coincidence
detection of gravitational waves. The experimental goal is to measure changes in distance of as
little as 1O18 m in the separation of the masses in the arms over a frequency interval of 1O-1()(X)
Hz. The expected signal from a binary neutron star system, like the one observed by Hulse and
Taylor, is a so-called 'chirp' signal (Figure I) that increases frequency and amplitude as it crosses
our frequency bands during the last seconds of the inspiral.
The 'benchmark' design goal of LIGO is to have sufficient sensitivity capability to observe
such systems. The best estimates for rates, based on extrapolations of the statistics of neutrons
stars in our galaxy yield - 200 Mpc (650 million light years) as the distance LIGO must be
sensitive to. in order to observe three neutron inspirals/year. More optimistic estimates yield a
distance of 23 Mpc, and ultraconservative estimates yield 1000 Mpc. With this guidance and
experimental practicalities, the strategy for LIGO is to build an initial device that will approach
the interesting region, be straightforwardly improvable to the best guess estimate. Furthermore,
the overall LIGO facility design is such that future more sensitive interterometers can reach the
conservative bound without being limited by the facility (e.g. the vacuum, the seismic isolation,
etc.).
The LIGO concept is shown in Figure 2. A high power laser (initially lOW), which has been
highly stabilized, is used as the light source. The laser is a Nd:YAG type with wavelength A. =
74
1064 11m. The laser beam is injected into the two arms of the interferometer by splitting the
beam, and then the beam is servo-locked to the length of the interferometer arms. The detectors
are set on a dark fringe and using the high photostatistics, small changes in distance are recorded
by very finely 'splitting' the fringe.
The reflectivity's of the mirrors are selected and positions controlled to build up the beam in
the resonant cavities and to optimize sensitivity.
THE SENSITIVITY OF LIGO
A comparison of LIOO sensitivities with predictions of various astrophysical sources is given

in Figure 3. The shape of the sensitivity curve is determined by three basic noise sources; seismic
noise at low frequencies (10-50 Hz); thermal noise at intermediate frequencies (50-100 Hz); and
shot noise at the highest frequencies (100 - 1000 Hz). Both the initial LIOO and an advanced
detector (where all three noise floors have been significantly improved) are shown. In addition to
binary neutron star systems, possible signal levels from binary blackhole systems and from
supernovae collapse are shown.
LIGOSTATUS
LIOO is presently making the transition from a design effort to an actual construction project.
This is an exciting time, when after years of preparation, dirt is being moved, concrete poured,
and the enormous vacuum system construction gotten underway.
The two sites for LIOO are in Hanford, Washington and Livingston, Louisiana as shown in
Figure 4. Each site consists of a large corner station, which houses the lasers, the input-output
optics and vacuum equipment. This station 'feeds' the long 4 km beam pipes with beams that
. ~ End Mirrors
T~
Recycling
Mirror
To
.......,; T=3% Detector
From
Laser
.figure 2. The optical system for LIGO
75
frequency, Hz
Figure 3. LIGO sensitivity compared with expected gravitational wave signals.
Figure 4. LIGO sites.
ret1ect off test masses at the far end. The vacuum chambers are 4 ft in diameter allowing
installation of multiple interferometers in the same vacuum. In fact, the Hanford facility will
initially house two interferometers (2 km and 4 km) to help reduce noise through a 'local'
coincidence.
Both sites have been cleared and graded, and the concrete slab on which the beam tube will be
erected. and the beam tube enclosure (Figure 5) are now being constructed in Washmgton.
The beam tubes are made from thin stainless steel, especially selected and carefully cleaned to
reduce outgassing properties. They are rolled in a special spiral mill designed for this purpose and
welded. The tube has stiffening rings to maintain the stability and bellows every 130 ft to allow
for expansion and contraction.
76
Figure 5. Beam tube and enclosure for LIGO.
Internal to the beam tube. baffles are inserted to reduce scattered light off the walls getting
into the system. Vibrations of the walls cause modulation of this light which can create phase
noise background. To reduce this problem, baffles of special material ('black glass' fired on
stainless steel) is absorptive at laser wavelengths and is serrated at the edge to minimize ditfractive
effects.
The vacuum requirements are stringent in order to" insure that gas scattering will never a
factor to the noise floor, even in advanced detectors. The pressure required to meet this goal is
that we obtain <10- 9 torr for all residual species. That has been achieved in a 130 Ii prototype
beam pipe assembly in all essential ways to be identical to the actual LIGO beam tube. Now. we
just need the quality control to build 16 km of beam pipe in the field of the same quality in order
to yield the same performance.
The conventional facilities for LIGO are well underway and will be completed in 1999. The
initial interferometers are under design, with the various subsystems projected to have final
designs and initiate construction within the next two years. By the year 2000 we expect to be
commissioning the initial interferometers, testing them and soon after we will do an initial search
for gravitation waves. Our goal is to reach our initial design sensitivity by 2002 at which point we
hope to detect gravitational waves.
In any case, the detector sensitivity will be improved systematically by incremental

improvements in the following years, allowing detection if the rates are low. If gravitational
waves have already been observed, it will afford the opportunity to improve the rates to where
LIGO will be able to initiate a new field of research - gravitational wave astrophysics.
REFERENCES
I. Abramovici, A. et. aI., "LIGO, the Laser Interferometer Gravitational Wave Observatory",
Science 256.325-333 (1992).
77
Cosmology and LIGO
Lee Samuel Finn
Department of Physics and Astronomy,

Northwestern University
Evanston, Illinois 60208-3112
INTRODUCTION
Overview
The most promising anticipated source for the United States Laser Interferometer
Gravitational-wave Observatory (LIGO) [1], or its French/Italian counterpart VIRGO
[4], is the radiation emitted during the final moments of inspiral before the coalescence
of a neutron star - neutron star (ns-ns) binary system [24]. The instruments operating
in both the LIGO and VIRGO facilities will evolve over time, eventually becoming
sensitive to neutron star binary inspirals at distances approaching 2 Gpc [10].
Binary inspiral observations in the LIGO or VIRGO detectors will be characterized
by their signal strength and "chirp mass" (a combination of the binary's component
masses and cosmological redshift). The distribution of observed inspirals with signal
strength and chirp mass depends on cosmological parameters that describe our uni-
verse (Hubble constant, deceleration parameter, density parameter), the distribution
of neutron star masses in binary systems, the overall density of coalescing binaries, and
the properties of the detector. In this paper I summarize recent investigations of the
binary inspiral event distribution (with signal strength and chirp mass) that the LIGO
and VIRGO detectors can expect to observe for different cosmological models.
In addition to their value as quantitative expectations of what LIGO and VIRGO
can expect to observe, these a priori distributions will playa pivotal role both in the
construction of templates for detecting binary inspirals and in the interpretation of the
observations. The distributions reviewed here, evaluated for our preconceived notion of
the binary inspiral rate and cosmological parameters, are the prior probabilities required
to form the likelihood function from the observed detector output [10]. Additionally,
the observed inspirals will be a sample drawn from a particular cosmological model
characterized by Ho, qo, no, neutron star mass distribution, and evolution characteristic
of our own universe. By comparing the observed distributions to the ones described
here we can measure those properties of our own universe.
These cosmological tests are analogous to the number-count tests of classical cos-
mology, which, in their simplest form, involve observing the distribution of a source
79
population as a function of apparent luminosity or redshift. The first suggestion that
binary inspiral number counts be used to measure interesting cosmological parameters
was made by Finn and Chernoff [10, 5] (although they did not use the language usually
associated with this technique of classical astronomy). Using Monte Carlo simulations
they demonstrated that the distribution of inspiral events with signal strength and
chirp mass could be used to measure the Hubble constant. In this work I summarize
a more general, explicit, and complete exposition of the properties of binary inspiral
observation catalogs [8].
The cosmological implications of gravitational wave observations of binary inspiral
were first recognized by Schutz [22]. He pointed out that each in spiraling binary is
a standard candle in the sense that, if observed in three independent interferometers,
its luminosity distance can be determined from the observed detector response. If an
observed inspiral is associated with one of the several galaxy clusters that reside in
its positional error box (whose determination also requires three interferometers), and
if the red shifts of those clusters are determined optically, then observation of several
inspiraling binaries would lead to a statistical determination of the Hubble constant that
is independent of the cosmic distance ladder and the uncertainties that lurk therein.
Markovic [16] proposed a variation on the general theme introduced by Schutz:
he observed that known neutron star masses were all close to 1.4 M0 and that, in
any event, there is a maximum neutron star mass. The observed chirp mass is a
function of the mass of the binary's two components and its redshift. Assuming that
the mass distribution in neutron star binaries does not evolve significantly over the
range of binary inspiral observations, examination of the chirp mass distribution in
binary systems at fixed luminosity distance would reveal the corresponding redshift.
Thus, gravitational radiation observations alone might suffice to determine the Hubble
constant.
Unfortunately, detailed calculations show that, even for the most advanced 1IGO
and VIRGO detectors that have been discussed, the fractional uncertainty in the
measured luminosity distance will be of order unity for events seen more frequently
than thrice per year (i.e., for events at distances greater than approximately 100 Mpc)
[13, 6], and the angular position error boxes for these events are likewise large (on order
10 deg 2 ). Consequently, cosmological tests that rely on accurate and precise measure-
ments of the distance and position of inspiraling binaries using LIGO and VIRGO are
not promising.
In contrast, the cosmological tests discussed here and in [5] require only gravitational
wave observation in a single interferometer. Furthermore, advanced LIGO detectors can
expect to observe approximately 50 ns-ns binary inspiral events per year, from distances
up to 2 Gpc, whose signal strength can be measured to better than 10% and whose
chirp mass can be measured to better than 0.1% [10, 6]. The rate, depth, accuracy
and precision of these single interferometer observations suggest that cosmological tests
based on the distribution of observed events with signal strength and chirp mass have
great promise.
COALESCENCE RATE DENSITY
Introduction
In order to describe the binaries included in a signal-to-noise limited catalog we must
first characterize the coalescing binary distribution. Assume that coalescing binaries
are distributed homogeneously and isotropically with the cosmological fluid and define
the binary coalescence local specific rate density N by
80
(1)
where dV is a co-moving cosmological fluid volume element, dt is a proper time interval

measured in the fluid rest frame, Mo is the system's intrinsic chirp mass,
Mo == p,3/5 M 2 / 5 (2)
and p, and M are the binary's reduced and total mass.
Gravitational-wave detectors like LIGO or VIRGO measure
M == Mo(l+z), (3)
where z is the system's redshift with respect to the detector. To distinguish between
the observed quantity M, which involves the system's redshift, and Mo, which depends
only on the binary's intrinsic properties, refer to the former as the observed chirp mass,
or simply the chirp mass, and the latter as the intrinsic chirp mass.
In terms of N the total co-moving rate density on the surface of homogeneity at
redshift z is thus
n(z) = J N(Mo, z) dMo (4)
If we assume that, over the depth observable by either LIGO or VIRGO, evolution
in N is negligible 1 then 71, = no is a constant independent of z. The best cur-
rent estimate of the ns-ns binary coalescence rate density no at the current epoch
is 1.1 x 1O- 8 h Mpc- 3 ycl, where h is the Hubble constant measured in units of
100 Km S-1 Mpc- 1 [18, 19]. This estimates relies on the 3 observed binary pulsar sys-
tems that will coalesce in less than a Hubble time (PSRs 1913+16, 1534+12, and
2127+11C [23, 26, 20]). Since the number of observed systems is small the actual rate
is quite uncertain: Phinney [19] has estimated that, while unlikely, a rate two orders of
magnitude higher or lower could be reconciled with current observations. Black hole -
black hole (bh-bh) and bh-ns binaries are believed to form at rates comparable to ns-ns
binaries; however, the masses of these systems and the fraction that merge in less than
a Hubble time are entirely uncertain [18, 19].
Both theoretical and observational evidence suggest that the neutron star mass
distribution is narrow [27, 7, 2]. The limited observations of the neutron star mass
distribution in binary pulsar systems provide independent 95% confidence intervals for
ml and mu [7]:
1.01 < mdM0 < 1.34

(5)
1.43 < m,.jM 0 < 1.64.
The most likely values of ml and mu are
1.29M0 ,
1.45M0 (6)
In this brief review I ass~me that Mo is a constant .Alto for all binary systems (for 1.4 M0
neutron star binaries, Mo ~ 1.2 M0)' In this limit the homogeneous and isotropic local
specific rate density can be written
1 As shown in section , advanced LIGO and VIRGO detectors will observe ns-ns binaries from
redshifts not expected to exceed z ~ 0.5 and with the preponderance of events ariSing from z ~ 0.1;
consequently, neglect of evolution is not an unreasonable approximation.
81
(7)
THE SIGNAL-TO-NOISE RATIO
Binary inspiral signal-to-noise ratio

The amplitude signal-to-noise ratio p corresponding to compact binary inspiral is [10]
( _M
5/6
_ _~ __ ) 1/2
P - 8e dL 1.2 M0 ((fmax), (8)
where
(9)
(10)
(11)
fmax is the system's orbital angular frequency at coalescence, dL is its luminosity dis-
tance (a function of cosmological redshift), and Sn (f) is the detector noise power spec-
tral density. In equation 8 e, M and dL depend on the particular binary system under
consideration and TO is a characteristic distance that depends only on the detector's
noise power spectral density Sn(f). The dimensionless function (, which also depends
only on the detector's noise spectrum, increases monotonically from 0 to 1. Its argu-
ment fmax is the redshifted instantaneous orbital frequency when the inspiral terminates
either because the compact components have coalesced or because the orbital evolution
is no longer adiabatic and coalescence is imminent.
The characteristic distance TO and characteristic function (describe different aspects
of a gravitational-wave detectors sensitivity to binary inspiral gravitational radiation.
For a fixed binary, the larger TO, the greater p; for fixed p, the larger TO, the farther the
detector can "see." Decreasing the noise power in any band increases TO, but (owing
to the factor f- 7/ 3 in the integrand of the expression for X7/3) improvements at low
frequency are more effective than those high frequency.
The function ( reflects the overlap of the signal power with the detector bandwidth.
The orbit of an inspiraling compact binary evolves adiabatically owing to gravitational
radiation emission until an innermost circular orbit (ICO) is reached at an instantaneous
orbital frequency fIco. At the lCO the orbit evolves on a dynamic timescale and the
components coalesce quickly. Thus, the adiabatic inspiral waveform ends when the
orbital frequency reaches fIco. 2 The quadrupole radiation frequency observed at the
detector when the orbital frequency is fIco is
21, - 2fICO (12)

max- l+z
Since the binary orbit evolves adiabatically from low frequencies to fICO, the ob-
served quadrupole radiation spectrum has significant power only up to frequency 2fmax.
If 2fmax is very much greater than the frequency where the detector noise is minimized,
than the signal power bandwidth overlaps completely with the detector bandwidth and
82
(Umax) ~ 1. On the other hand, if 2fmax is much less than the frequency where the de-
tector noise is minimized, than the overlap of the signal power and detector bandwidths
is negligible, (Umax) ~ 0 and p ~ o.
The orbital frequency at the transition of the binary orbit from adiabatic inspiral
to plunge and coalescence has been studied using high-order post-Newtonian methods
[14]. For symmetric binaries (i. e., those with equal-mass components) the instantaneous
red shifted instantaneous orbital frequency of the ICO is given by
frco
fmax = (13)
1 + z'
710. Hz (2.8M(o))
(14)
l+z M '
99. Hz (20. M(o)) (15)
l+z M '
where M is the binary's total mass. More generally, for binaries with compact compo-
nents frco depends inversely on M and directly on a function of the dimensionless ratio
IL/ M; Kidder [14, fig. 4] shows M frco for binary systems of arbitrary mass asymmetry.
For a ns-ns binary the component's proper separation at the ICO is greater than a neu-
tron star diameter; so, coalescence occurs after the transition from inspiral to plunge
[14]. Tidal dissipation is important in determining the inspiral rate only in the last few
orbits before contact [15, 3]; consequently, equation 14 is applicable for ns-ns binaries
and should be applicable for black hole binaries as well.
Distribution of the orientation function 8

Interferometric gravitational-wave detectors are not isotropic antennae; neither are in-
spiraling binaries isotropic radiators. The signal-to-noise ratio of an inspiraling binary
in a single interferometric detector depends on the relative orientation of the source
and the detector to the line of sight connecting the two. The dependence of p on these
angles is confined to the angular orientation function 8, which is independent of all
other properties of the binary system. From observations of binary inspiral in a sin-
gle interferometer we can measure p and M but not 8. Even though 8 cannot be
measured, because it depends only on the angles that relate the detector and binary
to the line of sight we actually know quite a bit about it: in particular, averaged over
many binaries, we know the probability that 8 takes on any value. This probability
distribution is found numerically in [10]; here I note only that 0 ~ 8 ~ 4 and that to
an excellent approximation
Pe(8) = { 508 (4 - 8)3/ 256 if 0 < 8 < 4 (16)

otherwise.
Note that, since p is directly proportional to 8, we immediately have the probability
that a source with intrinsic chirp mass Mo at redshift z (and luminosity distance dL )
appears to us with signal-to-noise ratio p:
(17)
2The radiation waveform from the final plunge and the early stages of coalescence is not yet known,
though attempts to determine it numerically are underway [9]. If, after coalescence, a black hole forms,
then the final radiation reflects the black hole's quasi-normal modes, which are damped rapidly. The
inspiral waveform, and with it our ability to model the detector response, ends when the orbital
frequency reaches fIco.
83
where
P dL [ 1.2 M]
0
5/6
(18)
x == 8" ro(1/2 Mo(1 + z)
ro and ( for the LIGO and VIRGO detectors
LIGO will consist of three interferometers: one in Livingston, Louisiana and two in
Hanford, Washington. The Louisiana interferometer and one of the Washington inter-
ferometers will have 4 Km arm-lengths; the second Washington interferometer will have
a 2 Km arm-length and share the same vacuum system as the 4 Km interferometer.
For the proposed LIGO interferometers as described in [25J and modeled in [10], ro
ranges from 13 Mpc (for "initial" interferometers) to 237 Mpc (for "advanced" inter-
ferometers).3 As the LIGO detectors develop, incremental improvements will increase
roo
The orientation of the Washington and Louisiana interferometers were chosen to
be as close to parallel as possible; consequently, a simple approximation treats the
LIGO detector (all three interferometers operating in "triple-coincidence") as a single
interferometer of arm length
(19)
For this "super-interferometer" ro ranges from 19 Mpc to 355 Mpc.

Here and below, reference to the LIGO detector refers to the three interferometers
operating as a single detector, while reference to a LIGO interferometer indicates one
of the 4 Km interferometers operating in isolation.
As the LIGO detector evolves, ro will increase from approximately 20 Mpc toward
350 Mpc. VIRGO will consist of a single 3 Km interferometer. The target noise curve
for the initial operation of the VIRGO interferometer is described in [4J; for this inter-
ferometer ro is 13 Mpc. Like the LIGO interferometers, incremental improvements in
VIRGO will increase ro and it is reasonable to assume that these improvements could
raise ro for VIRGO to approximately 200 Mpc.
For neutron star binaries fmax ~ 710Hz (eq. 14) and ((fmax) ~ 1 in any of the
proposed LIGO and VIRGO interferometers.
DISCUSSION
Cosmology Tests
Consider a catalog of binary inspiral events with signal-to-noise ratio P greater than a
threshold Po. Suppose that for each event in the catalog p and M are known. How do
we make use of the catalog to test cosmological models against cataloged observations?
The simplest test involves the distribution of events with p. As the catalog limit
Po decreases, sources at increasingly larger distance become members. The number of
added sources depends on the increase of the spatial volume, the density of sources at
the corresponding red shift , and (since these sources are events that occur at a given
rate) the cosmological redshift. Thus, adopting a cosmological model C (e.g., Hubble
3The noise model used in [25] and [10] assumed a thermal noise spectrum corresponding to vis-
cous damping in the pendulum suspensions and the internal modes of the test masses. It is now
recognized that these modes axe structure damped [21, 11, 12] with a correspondingly different noise
spectrum. Preliminary estimates indicate that this improved noise model reduces ro for initial LIGO
interferometers, but leaves it unchanged for advanced ones.
84
constant Ho, deceleration parameter qo and cosmological constant A) implies an ex-
pected distribution P(plpo, C, V) for the catalog events taken with detector V. Denote
the cataloged binary inspiral signal-to-noise ratio observations by {pip> PO}. Suppose
that, before studying these observations, we have reason or are otherwise prejudiced to
believe the probability that C is the correct cosmological model is P(C). Using Bayes'
law of conditional probabilities, the a posteriori probability that we assign to model C
after considering the observations is
(20)
where Pi is the signal-to-noise ratio of the ith catalog observation and Vi represents the
detector configuration (ro, () when the observation was made4 . This test is exactly
analogous to number-flux cosmological tests using distant galaxies.
More subtle tests involve other distributions of cataloged events. For example, each
source in the catalog is characterized by its observed chirp mass M. The observed
chirp mass depends on both the intrinsic chirp mass Mo and the redshift z (see eq. 3);
consequently, the distribution of events in the catalog with M depends on the cosmo-
logical model. Denote the cataloged chirp mass observations by {Mlp > PO}. Adopting
a cosmological model C implies an expected distribution P(Mlpo, C, V). As before, if
we initially favor model C with probability P(C), then after studying the observations
the probability we ascribe to model C is
(21)
where Mi represents the ith cataloged chirp mass observation.

Of the four parameters that describe, at quadrupole order, binary inspiral observed
in a single interferometer, only p and M convey astrophysically interesting informa-
tion. The distributions P(plpo, C, V) and P(Mlpo, C, V) are each integrals over the
distribution P(p, Mlpo, C, V) that completely characterizes the catalog:
P(pIPo,C, V) = ! P(p,Mlpo,C, V)dM (22)
P(MIPo,C, V) = 1 00
PO
P(p,Mlpo,C, V)dp. (23)
These integrals are summaries of the catalog contents: as such, they are less informative
than P(p, Mlpo, c, V). The most sensitive test that we can make using the catalog
involves n.ot a summary, but the full information available: given the observations
{p, M Ip > po}, the probability that cosmological model C describes our universe is
(24)
Cosmological tests based on the summary distributions are still useful to make.
Summary distribution often depend only weakly or not at all on some of the param-
eters of the model C; in that case, the effective dimensionality of C is reduced in the
summary test and we may be able to distinguish more closely among cosmological mod-
els described by the remaining parameters than with a test using the full distribution.
This is particularly true when the number of observations in the catalog is small.
4Changes in the detector noise spectrum change the relation between p and the cosmological model.
This is not a problem in the.interpretation of a catalog as long as the detector properties are properly
associated with each cataloged observation.
85
In the remainder of this section I review some results for the expected distribution of
sources in a signal-to-noise limited binary inspiral observation catalog. For derivations
and a more detailed discussion see [8].
Sample depth
The signal-to-noise ratio of an inspiraling neutron star binary system with intrinsic
chirp mass Mo is
(25)
Since e is between 0 and 4, even an optimally oriented binary system has p less than
Po when z is greater than zo, where Zo satisfies
4_ POdL(zo) (1. 2M0)5/6 (26)

- 8rO(1/2(Jmax)5/6 Mo
Evaluation of Zo requires ((Jmax), which depends on the details of the detector's noise
power spectral density (through () as well as the binary's component masses and red-
shift (through fmax). For advanced interferometers ( ~ 1 as long as fmax ~ 70 Hz. Since
fmax ~ 710 Hz [2.8 M0/(1 + zo)M] for symmetric binary inspiral (see eq. 14) we can ap-
proximate ( ~ 1 as long as 1 + Zo ;S 10 (2.8 M0/M). For small Zo we can approximate
dL ~ z/ Ho; then
Zo (27)
(28)
which is much less than 10. Thus, the approximation ( ~ 1 is a good one for binary
systems with solar mass components, but not for binaries whose components are on
order 5M 0 .
For these approximately solar-mass binary systems Zo satisfies
(29)
where x is a root of
O -- X 12 + 8 qoaox11 + 16qoaox
2 2 10 - 2qox6 - 8 ao ( 2qo - 1) x 5 + 2qo - 1. (30)
and
8Horo A) 5/6
( Mo
Po 1.2M 0
(31)
(32)
The appropriate root of this equation can be found as a power series for small ao:
86
Zo = 4ao + 3"8 (3qo + 2) a o2 + 8 (6qo - 1) a o3 + 81
160 (54qo
2 + 9qo - 7 ) a 4
o
5600 ( 2 447 323) 5 (6)

+ 243 27qo - 2"5 qo + 100 ao+0 ao (33)
This truncated expansion is accurate to better than 0.2% for ao < 0.12 and 0::; qo ::; 1;
for ao < 0.1 and 0 ;S qo ;S 3/4 the accuracy is better than 0.01%.
Figure 1 shows Zo as a function of h for neutron star binaries with Mo = 1.19 M0
(corresponding to 1.37 M0 neutron stars), Po = 8, TO = 355, and three different values
of qo corresponding to an open (qo = 1/4, dotted curve), fiat (qo = 1/2, solid curve) and
closed (qo = 3/4, dashed curve) cosmological model. The redshift of the most distant
such source is less than 1/2 for h < 0.8 and does not exceed 7/10 for h < 1. The
sensitivity of Zo to qo is modest but significant: inspection of equation 33 shows that
Zo ~ 4ao [1 + ao (2qo - 1)] +0 [a~, (2qo - 1)2] , (34)
where ao ~ 0.12h for advanced LIGO interferometers. For open spatial geometries
(2qo - 1 <
0) the most distant sources are at smaller redshifts than in closed spatial
geometries (where 2qo - 1 > 0).
/
/
/
0.6 /
/
/ ....
/
/
0.5 /
/ .. /
c
/
N / ./
/
0.4 / .'
I I ......... ....
I .. /
/ .. /
0.3 ,0~...... /
,0 .. /
~.1/
~./
0.4 0.6 0.8

h
Figure 1: The distance to the farthest inspiraling binary system with signal-to-noise
ratio P greater than a threshold Po depends on the detector noise spectrum, the binary
system component masses, and the cosmological model. Shown here is the redshift to
the farthest ns-ns binary system observable with p ~ 8 in an advanced LIGO detector as
a function of the Hubble parameter h (the Hubble constant in units of 100 Km/s/Mpc).
The three curves represent matter dominated Friedmann-Robertson-Walker cosmologi-
cal models with different qo: a closed model with qo = 3/4 (dashed curve), a fiat model
(qo = 1/2, solid curve), and an open model with qo = 1/4 (dotted curve).
87
In luminosity distance, the sample depth is
4ao [ 40 40 2 1280 3
dL,o = Ho 1 + 3 ao +3 (2qo + 1) a o + 81 (9qo - 2) a o
+ 2~03 (2052q5 - I80qo - 143) a~ +0 (a g)] . (35)
Figure 2 shows dL,o for the same cases (h, qo) as figure 1 shows zoo Advanced LIGO
interferometers may observe neutron star binaries with p greater than 8 at luminosity
distances of order 2 Gpc.
Coalescence rate above threshold

The distribution of catalog events with p is given by
d2N/dtdp
P(plpo, C, 1)) = { dN/dt p> Po, (36)
o P < Po,
where
1 00
dp-
dtdp
~N
(37)
J
PO
d3N
(38)
dM odtdpdMo'
dtdpdMo
J dz dr 41ra~r2 no
dz VI - kr2I + z
6 (Mo - Alto)
x Pp(plz, M o, C, 1)), (39)
I+z
2000
()
0.
::E
.,j
1800
0.4 0.6 0.8

h
Figure 2: The same as figure 1, except shown here is the luminosity distance instead
of the redshift.
88
r measures co-moving distance on the cosmological spacelike hypersurfaces of homo-
geneity, ao is the cosmological scale-factor at the present epoch, k is +1, -1, or 0 for
closed, open, or flat spatial geometries, and Pp is given in equation 17.
Since d3 N/dtdpdM o depends on P only through Pp (see eq. 17), the integral over
p in equation 37 can be evaluated explicitly:
1 00
Po
dpPp(pIMo, z,C, D) 1 00
d8Pe(8) (40)
= Ce(x), (41)
where
5/6
= Po dL ( 1.2~0 )
(42)
X
8 ro (1 + z) 5/6'
Mo
The probability density Pp is a conditional one: it depends on M o, z, and the cosmo-
logical model. In contrast, the distribution Pe of 8 is universal: it is independent of
all the specific properties of the binary or the detector. Ce(x) is universal in this same
way and can be evaluated with Pe.
With our approximate expressions for Pe and zo(Po) we can evaluate the integrals
for d2 N / dt dp and dN/ dt as power series in small ao:
dN
dt
(43)
(44)
where ao is given by equation 31 and
(45)
In finding P(pIPo,Mo,C, D) we also found dN/dt, the anticipated rate of binary

in spirals with p > Po as a function of the cosmology, detector, and the expected intrinsic
chirp mass. Figure 3 shows dN/ dt as a function of h in an Einstein-deSitter cosmological
model (qo = 1/2) for a reasonable signal-to-noise ratio limit in an advanced LIGO
detector (Po = 8 and ro = 355 ~pc), typical neutron star masses (Mo = 1.19 ~0)'
and two different coalescence rate densities. The solid curve shows dN/dt for the "best
guess" coalescence rate density, no = 1.1 x 1O- 7 h ~pC-3 S-l, which is proportional
to h; the dashed curve shows dN/ dt for a constant coalescence rate density no =
8 x 10- 8 ~ pc -1 yr- 1 . The dashed curve shows how dN/ dt scales for constant no in
advanced interferometers with interesting h. Present estimates of no suggest that, if
89
0.4 0.6 0.8
h
Figure 3: The rate of ns-ns binary inspiral observations with signal-to-noise ratio greater
than 8 in an advanced LIGO detector is largely insensitive to the neutron star mass
range or the deceleration parameter in matter dominated Friedmann-Robertson-Walker
cosmological models. The solid curve shows the expected rate in an Einstein-deSitter
model as a function of the Hubble parameter h assuming the co-moving ns-ns binary
coalescence rate density at the current epoch is 1.1h Mpc 3 yr- 1 (solid curve); the
dashed curve shows the same assuming the rate density is 8 x 10- 8 Mpc- 3 ye 1 , which
is independent of h.
h = 0.75, advanced LIGO detectors can expect to observe approximately one neutron
star binary inspiral event per week.
The distribution P(plpo, C, 1)) is also sensitive to h. Let Co represent a flat and
static cosmological model. Then
P(plp C 1)) = { 3p~/ p4 p > Po (46)

0, 0, 0 p < Po
The first order correction (in 1/ p) owing to the expansion of the universe depends only
on the rate of expansion (Ho); corrections owing to the curvature of space (k "# 0) enter
only at second order in 1/ p. Since, for ns-ns binaries and reasonable limiting signal-to-
noise 00 2 is less than 1% even for advanced interferometer designs, cosmological tests
that focus only on the observed distribution of p will be insensitive to qo even though
the redshift to the most distant sources in the catalog is large. This weak dependence
on qo is characteristic of number-flux cosmological tests (see, e.g., [17, pg. 798]). On the
other hand, for advanced detectors and interesting cosmological models, the distribution
P(plpo, C, 1)) is sensitive to h at the 10% level, which makes possible a measurement of
h from the observations {pip> Po} alone.
Chirp mass spectrum P(MIPo,I)

In a homogeneous and isotropic cosmology the rate at which binary inspiral signals
corresponding to chirp mass M and p greater than Po are observed is
90
d2 N
dtdM
l dz dr
o dz
47Ta~r2
VI -
no 8
kr 2 1 + z
(~ -
1+ z
Mo) Ce [po dL
8 ro(1/2(Jrnax)
(1.2 M0)
M
5/6]. (47)
The catalog's a priori chirp mass distribution is thus
~N/dtdM
P(Mlpo, C, V) = dN/dt (48)
where
~N no dr (Z) 47Ta~r2(Z) C [po dL(Z) (1. 2M 0)5/6] (49)
dtdM M dz Jl - kr2(Z) e 8 (1 + Z)5/6ro Mo '
Z ~ -1. (50)
Mo
Figure 4 shows the distribution P(Mlpo,C, V) for catalogs with p greater than 8
compiled by an advanced LIGO detector (ro = 355 Mpc) in several matter dominated
FRW cosmological models. The intrinsic chirp mass of all systems is assumed to be
1.19 M0 . Six different models are shown, exploring two different h (0.5 and 0.8) and
three different qo (1/4, 1/2 and 3/4). The closely spaced curves with co-located extrema
are of the same h and differ only in qo. Note the strong dependence of P(Mlpo,C, V)
{)
/r
Ii \~
./! \
6 qo=3/4
5 qo=1/2
qo=1/4
I
o
1.2 1.4 1.6 1.8
Figure 4: A binary system's observed chirp mass M depends on its redshift; conse-
quently, a ns-ns binary inspiral sample will show a range of chirp masses corresponding
to the range of system redshifts. Shown here is the expected distribution of M for
binary systems consisting of two 1.37 M0 neutron stars with p > 8 in advanced LIGO
detectors for open (qo = 1/4), flat (qo = 1/2) and closed (qo = 3/4) matter-dominated
Friedmann-Robertson-Walker cosmological models with h = 0.5 and h = 0.8. In all
cases, as qo increases the tail of the chirp mass spectrum is extended.
91
on h and the weaker, but still significant, dependence on qo: the dotted and solid
curves correspond to flat cosmological models (qo = 1/2), the long-dash and dot-Iong-
dash curves correspond to open cosmological models (qo = 1/4), and the short-dash
and dot-short-dash curves correspond to the closed models (qo = 3/4). In general the
smaller qo, the more compressed the spectrum and the smaller the tail at large M.
Since the abscissa M is related to redshift according to M = (1 + z)M o figure 4
also shows the redshift of the preponderance of catalog events. For h = 0.8 most events
are at a redshift of 9%, while for h = 0.5 most events are at a redshift of 14%.
CONCLUSIONS
Observations of binary inspiral in a single interferometric gravitational wave detector

can be cataloged according to signal strength (as measured by signal-to-noise ratio p)
and chirp mass M. The distribution of events in a catalog composed of observations
with p greater than a threshold Po depends sensitively on the Hubble expansion and
less sensitively on the deceleration parameter and cosmological constant, as well as the
distribution of component masses in binary systems (neutron star binary observations
evolution is not expected to be important). In this paper I review some recent work [8]
on the anticipated distribution of binary inspiral events that will be observed by the
LIGO and VIRGO gravitational-wave antennae.
The distributions described here have two immediate, practical uses in gravitational-
wave data analysis: first, when evaluated for the cosmological parameters reflecting our
current best understanding of the universe, they are the prior probabilities which, to-
gether with the matched-filtered detector output, form the likelihood function and
determine the posterior probability that an in spiral has been detected; second, when
compared with the observed distribution in p and M of many separate binary inspiral
observations, they are used to infer new and more informed estimates for the cosmo-
logical parameters that describe the universe.
For the purpose of analyzing their sensitivity to binary inspiral, gravitational wave
detectors can be described by a characteristic distance TO and a bandwidth overlap
function (. The characteristic distance ro depends only on properties of the detector
and gives an overall sense of the depth to which the detector can "see" solar mass
binary systems whose radiation traverses the full detector bandwidth. The character-
istic function ( describes the overlap of the radiation from a particular binary system
with the detector bandwidth. Advanced LIGO interferometers are expected to be most
sensitive to binary inspiral radiation in the bandwidth 30-200 Hz: over 90% of the
signal-to-noise ratio is contributed by signal power in this narrow band.
For advanced LIGO detectors, the most distant neutron star binary inspiral events
with signal-to-noise ratio greater than 8 will arise from distances not exceeding ap-
proximately 2 Gpc, corresponding to a redshift of 0.48 (0.26) for h = 0.8 (0.5). The
depth is only weakly dependent on the range of neutron star masses or the deceleration
parameter. As the binary system mass increases so does the distance it can be seen, up
to a limit: in a matter dominated Einstein-deSitter cosmological model with h = 0.8
(0.5) the limit is approximately z = 2.7 (1.7) for binaries consisting of approximately
10 Mev black holes.
ACKNOWLEDGMENTS
It is a pleasure to thank the organizers of the Coral Gables conference for their invitation
to discuss this work at Orb is Scientiae. I am also grateful for the support of both the
Alfred P. Sloan Foundation and the National Science Foundation (PHY 9308728).
92
References
[IJ Alex Abramovici, William E. Althouse, Ronald W. P. Drever, Yekta Giirsel,

Seiji Kawamura, Fredrick J. Raab, David Shoemaker, Lisa Sievers, Robert E.
Spero, Kip S. Thorne, Rochus E. Vogt, Rainer Weiss, Stanley E. Whitcomb, and
Michael E. Zucker. LIGO: The laser interferometer gravitational-wave observatory.
Science, 256:325-333, April 1992.
[2] H. A. Bethe and G. E. Brown. Observational constraints on the maximum neutron

star mass. Astrophys. J. Lett., 445:LI29-LI32, 1 June 1995.
[3J Lars Bildsten and Curt Cutler. Tidal interactions of inspiraling compact binaries.
Astrophys. J., 400:175-180, November 1992.
[4J C. Bradaschia, R. del Fabbro, A. di Virgilio, A. Giazotto, H. Kautzky, V. Monte-

latici, D. Passuello, A. Brillet, O. Cregut, P. Hello, C. N. Man, P. T. Manh, A. Mar-
raud, D. Shoemaker, J. Y. Vinet, F. Barone, L. Di Fiore, 1. Milano, G. Russo,
J. M. Aguirregabiria, H. Bel, J. P. Duruisseau, G. Le Denmat, Ph. Tourrenc,
M. Capozzi, M. Longo, M. Lops, I. Pinto, G. Rotoli, T. Damour, S. Bonazzola,
J. A. Marck, Y. Gourghoulon, 1. E. Holloway, F. Fuligni, V. Iafolla, and G. Natale.
The VIRGO project - a wide band antenna for gravitational-wave detection. Nucl.
Instrum. Methods Phys. Research, A289:518-525, 1990.
[5J David F. Chernoff and Lee Samuel Finn. Gravitational radiation, inspiraling bi-
naries, and cosmology. Astrophys. J. Lett., 411:L5-L8, 1993.
[6J Curt Cutler and Eanna Flanagan. Gravitational waves from merging compact
binaries: How accurately can one extract the binary's parameters from the in spiral
waveform. Phys. Rev. D, 49(6):2658-2697, 1994.
[7J Lee Samuel Finn. Observational constraints on the neutron star mass distribution.
Phys. Rev. Lett., 73(14):1878-1881,3 October 1994.
[8] Lee Samuel Finn. Binary inspiral, gravitational radiation, and cosmology. Phys.
Rev. D, 53(6):2878-2894, 15 March 1996.
[9J Lee Samuel Finn. A numerical approach to binary black hole coalescence. gr-
qc/9603004; to appear in the proceedings of GRI4, March 1996.
[10J Lee Samuel Finn and David F. Chernoff. Observing binary inspiral in gravitational
radiation: One interferometer. Phys. Rev. D, 47(6):2198-2219, 1993.
[I1J A. Gillespie and F. Raab. Suspension losses in the pendula of laser interferometer
gravitational-wave detectors. Phys. Lett. A, 190:213-220, 25 July 1994.
[12] A. Gillespie and F. Raab. Thermally excited vibrations of the mirrors of laser
interferometer gravitational-wave detectors. LIGO pre print 94-3, November 1994.
Submitted to Physical Review D.
[13] P. Jaranowski and A. Krolak. Detectability of the gravitational-wave signal from a

close neutron-star binary with mass-transfer. Astrophys. J., 394(2):586-591, 1992.
[14] Lawrence E. Kidder, Clifford M. Will, and Alan G. Wiseman. Coalescing binary
systems of compact objects to (post)5/2-Newtonian order. III. The transition from
in spiral to plunge. Phys. Rev. D, 47:3281-3291, 1993.
93
[15] Christopher S. Kochanek. Coalescing binary neutron stars. Astrophys. J., 398:234,
1992.
[16] Dragoljub Markovic. Possibility of determining cosmological parameters from mea-

surements of gravitational waves emitted by coalescing, compact binaries. Phys.
Rev. D, 48(10}:4738-4756, 15 November 1993.
[17] C. W. Misner, K. S. Thorne, and J. A. Wheeler. Gravitation. Freeman, San

Francisco, 1973.
[18] Ramesh Narayan, Tsvi Piran, and A. Shemi. Neutron-star and black hole binaries
in the galaxy. Astrophys. J. Lett., 379:LI7-L20, 1991.
[19] E. Sterl Phinney. The rate of neutron star binary mergers in the universe: Minimal
predictions for gravity wave detectors. Astrophys. J., 380(1}:LI7-L21, 1991.
[20] T. A. Prince, S. B. Anderson, S. R. Kulkarni, and A. Wolszczan. Timing observa-

tions of the 8 hour binary pulsar 2127+11C in the globular cluster M15. Astrophys.
J. Lett., 374:L41-L44, June 1991.
[21] Peter R. Saulson. Thermal noise in mechanical experiments. Phys. Rev. D,

42(8}:2437-2445, 1990.
[22] B. F. Schutz. Determining the Hubble constant from gravitational-wave observa-
tions. Nature (London), 323:310-311, 1986.
[23] J. H. Taylor and J. M. Weisberg. Further experimental tests ofrelativistic gravity

using the binary pulsar PSR 1913+16. Astrophys. J., 345(1}:434-450, October
1989.
[24] Kip S. Thorne. Gravitational radiation. In Stephen Hawking and Werner Israel,
editors, 300 Years of Gravitation, pages 330-458. Cambridge University Press,
Cambridge, 1987.
[25] Rochus E. Vogt, Ronald W. P. Drever, Frederick J. Raab, Kip S. Thorne, and
Rainer Weiss. A Laser Interferometer Gravitational~wave Observatory (LIGO).
Proposal to the National Science Foundation, California Institute of Technology,
1989.
[26] A. Wolszczan. A nearby 37.9-ms radio pulsar in a relativistic binary system. Nature
(London), 350:688-690, 1991.
[27] S. E. Woosley and T. A. Weaver. Theory of neutron star formation. In David Pines,
R. Tamagaki, and S. Tsuruta, editors, The Structure and Evolution of Neutron
Stars, pages 235-249. Addison-Wesley, Redwood City, California, 1992.
94
Interferometry for Gravity Wave Detection
Peter Fritschel, MIT, LIGO Project
Nature of gravity waves: salient features for detectors

Much of the physics of general relativity (GR) is contained in the metric, gllv. The inherent
non-linearity of GR in general makes the computation of the metric very difficult if not impos-
sible, but here on earth the gravitational fields are always weak enough so that they can be
described by the so-called linearized theory. In this weak field limit, the metric can be approx-
imated as g~v ~ ll~v + h~v .Ih~vl 1 . The weak-field Einstein equations in vacuum (for a suitable
choice of gauge) then take the form of a wave equation: 02h~v = o. For a plane wave traveling
in the z direction, the field can have components in the x and y directions; i.e., the field is
transverse, with two independent polarizations.
The effect of the gravity wave on free masses can be understood by examining how the proper
distance, ds 2 = g~v dx ~dx v, between two masses, one located at the origin and the other at x=xo'
y=z=O, is affected: ~s = l"lgxxlll2dxz [1 + h12]xo. The gravity wave produces a relative 'dis-
placement' of two particle~, in the sense of the proper distance between them, that is propor-
tional to the wave amplitude and the initial separation. The effect is that of a tidal force, and it
is usual to regard h as a strain - a gravity wave of field amplitude h produces a strain of magni-
tude h.
Estimates of the field strength at the earth can be made using the linearized theory. Waves are
produced by oscillating multipole moments of the mass distribution of a system; the lowest
order than can emit energy is quadrupole radiation, and thus is usually the dominant form. The
wave field is proportional to the second time derivative of the source quadrupole moment, Q,
and decays as (lIr), r being the distance to the source. Dimensional analysis give the field
strength to be h - (Glc 4 )Qlr. Since the quadrupole moment is of order Q _MR2 (M is the mass
and R the size of the source), Q is roughly that fraction of the kinetic energy associated with
the quadrupolar, or non-spherical, motion of the source, Ekin ns:
h -G- -ns
E ldn (ns )
E ldn
- 1 0 _ 20 _ __ (10""MPC)
__
c4 r Msolar r
The maximum strain amplitude on earth thus ranges from 10- 17 for sources in our galaxy, to
10- 20 for sources in the Virgo cluster, to 10-23 for sources at cosmological distances.
How interferometric detectors work

An idealized approach to detecting gravity waves would be to allow free test bodies to float in
inertial space-time, and to continuously monitor the time of flight of light beams traveling
between them to sense changes in their proper separations. A Michelson interferometer is a
near ideal tool for monitoring the proper separations. Consider a Michelson interferometer
with a 50/50 beamsplitter at the origin which directs the input light (coming from a laser)
along the x and y axes, and two arm mirrors, one at x=l, y=O, the other at x=O, y=l. For one
95
phase of a gravity wave traveling in the z direction, the change in the x arm length is
A(, = h1l2, as shown above; the change in the y arm length is Al,. = - hll2 , consistent with the
quadrupolar nature of the wave. There is thus a differential phase shift between the two arms -
of magnitude Aelld = 41thll'A. - which precisely what a Michelson interferometer measures. The
free fall condition is clearly not practical for an earth-based detector, and in practice the 'test'
masses are hung from pendula so that they are free to move at frequencies above their pendu-
lum resonance. The real practical difficulties in implementing such a detector arise from the
extremely small length change that must be measured; even with the several kilometer arm
lengths of the large-scale projects, a wave amplitude of h - 10-21 produces a displacement of
only Ax - 10-18 m, or 10- 12 of the laser wavelength.
The arm length change, or signal, increases linearly with the interferometer length up to a
point: if the light travel (storage) time in an arm is longer than half the gravity wave period,
then the optical phase shift built up during one half-cycle of the wave will be removed during
the next half-cycle. For gw frequencies detectable on earth (j> 10 Hz), this corresponds to a
desired length of _106 m. A straight path of such length is clearly not practical, and instead
the light path is folded to achieve the same effective path, or storage time, in much smaller
physical lengths. The light thus samples the test mass displacement more than once, and the
length-to-phase conversion (dIP1dl) is increased compared to the simple Michelson interferom-
eter described above. A conceptually simple technique for doing this is to use optical delay
lines in the arms of the Michelson; the length-to-phase conversion is increased by the number
of bounces the beam makes on one of the delay line mirrors. In practice there would be some
tens of bounces on each mirror, and the large mirror surfaces this requires is a major disadvan-
tage of this technique. An alternative technique - and the technique to be used in both LIGO
and VIRGO - uses a resonant Fabry-Perot cavity as the storage element in each arm. The
length-to-phase conversion for this scheme is increased over the simple Michelson by a factor
of 2Flrc, where F is the finesse of the Fabry-Perot cavity. Typically a finesse of -100 is used,
corresponding to an optical storage time of'ts -1 msec, which gives optimum sensitivity at
-100 Hz.
The length-to-phase conversion factors in the preceding paragraph are correct only for gw
periods much longer than the optical storage time of the interferometer. For shorter gw peri-
ods the variation in optical phase shift over the cycle of the wave must be taken into account.
This is typically done by computing the frequency response of a particular optical configura-
tion to gravity waves of frequency f g . This can be done by considering that the gw-induced
modulation of the optical phase produces sidebands on the laser light at frequencies
Vo - fg and Vo + f g , where Vo is the laser frequency. The response of the optical system to light
at all three frequencies is then calculated to determine the frequency response to gravity
waves. Performing such a procedure for a Michelson interferometer with resonant Fabry-
Perot cavities in the arms gives an expression for the optical phase shift per wave amplitude;
in the limit where the interferometer length is small compared to the gravitational wavelength,
it has a simple form:
Belli- 4vots
i
h - JI + ( 41t f g t /
For high frequency waves, 41tfg ts I , the response is independent of the storage time, and 'sat-
urates' at a level IBeII/hl = 2vo/f g
The above discussion has assumed that the gravity wave is incident normally to the plane of
the interferometer, and that it is polarized such that the components of h are parallel to the
arms. The angular response of the interferometer is in fact relatively isotropic; for un-polar-
ized waves, the response is zero only for waves incident in the plane of the detector, at 45 0 to
the arm axes (for arms oriented 90 0 to each other).
96
Limitations to sensitivity: noise sources
The sensitivity of an interferometer depends not only on the conversion of gravity wave strain
to optical phase shift, but also on the ability to measure the optical phase. There are several
useful ways to classify the noise sources which limit the measurement. There is a natural divi-
sion into what are called sensing and displacement noise sources. Sensing noise is the noise
that would still be present if the test masses were fixed in inertial space and at absolute zero
temperature. At best this would be limited by the quantum statistics of the laser light (Poisson
statistics, not considering the possible use of squeezed light); in general sensing noise is of
optical origin. Displacement noise arises from actual forces acting on the test masses.
Another classification can be made into fundamental versus technical noise sources. Funda-
mental noise sources arise from unavoidable physical processes and present fundamental lim-
its to detection; an example is the quantum noise of the light. Specific fundamental noise
sources are further discussed below. Technical noise sources are due to imperfections in the
measurement technique, and in principle can be engineered away; an example is laser fre-
quency noise, which is a problem if (because) the interferometer arms are not perfectly
matched - it can be made insignificant by frequency stabilization and better matching of the
arms.
Noise sources can be further classified according to their statistical characteristics. Stationary
noise is ubiquitous, and is most conveniently described by power spectra. Quasi-stationary
noise changes on time scales of seconds, such as seismic noise. Periodic 'noise' may be arise
from the continual excitation of mechanical resonances in the system. Impulsive noise events
on time scales of milliseconds or shorter are particularly problematic to searches for impulsive
gravity wave events, and must be removed through correlation of the outputs of two or more
'independent' detectors.
Fundamental limits: Quantum noise. The equivalent phase noise due to the Poisson noise, or
shot noise, of the light depends on the specific technique used to detect the signal - generally
some sort of radio frequency phase modulation-demodulation technique is used for signal
detection - but the limiting sensitivity due to shot noise can be obtained by considering the
direct detection of the light at the mid-point of a fringe. The signal sensitivity - the change in
light power for a given phase shift - increases linearly with the light power in the interferome-
ter. The noise due to shot noise increases as the square root of the power: oP = J2hvP. The sig-
nal-to-noise ratio thus increases as JP. The equivalent strain sensitivity due to shot noise also
improves as JP:
where Pin is the input optical power to the interferometer. The frequency response factor,
IO$/hl, means that the sensitivity gets worse with frequency above the interferometer 'pole fre-
quency.'
The complementary aspect of the quantum noise is radiation pressure. Vacuum fluctuations
entering the output port of the interferometer produce quantum-limited intensity fluctuations
that are anti-correlated in the two arms. The resulting photon pressure fluctuations thus pro-
duce differential arm length fluctuations, which scale with frequency as l/j 2 and with the
square root of the optical power. These competing scalings of the shot noise ( - JP -1) and the
photon pressure noise ( - JP ) imply an optimum optical power. The strain sensitivity at the
optimum power is known as the 'quantum limit.' The differing frequency dependencies of the
two effects mean that the optimum power depends on the optimization frequency; for a simple
Michelson interferometer the optimum power is Pop, = TtcAmf 2 , where m is the mirror mass.
For m = 1Okg and A = 1 Ilm, Popt = 1 MW ! Practical powers for initial interferometers are
much lower (- 10 kW), and the quantum limit is not a concern initially.
97
In fact obtaining sufficient optical power (- 100 - 1000 W are needed) directly from a laser is
not practical, and a technique called power recycling is used to increase the effective power
and thus improve the shot noise limited sensitivity. Power recycling is made possible when the
interferometer is operated so that the Michelson output port is nominally at a minimum of the
fringe (at the 'dark fringe'). When the optical losses are low, then nearly all the input power is
reflected back towards the laser. From the viewpoint of the laser, the interferometer looks like
a retro-reflecting mirror, and so it is possible to form a resonant cavity out of the interferome-
ter and an additional mirror placed between the laser and the beamsplitter. With a particular
choice of the reflectivity of this additional mirror, it is possible to 'impedance match' the laser
light to the interferometer, so that no light is reflected; in this case the power buildup in the
interferometer is maximized. The possible power buildup, or recycling gain, is the inverse of
the fractional optical loss in the interferometer. For the large-scale interferometers, a recycling
gain factor of 30-50 is expected, thereby improving the shot-noise limited sensitivity by the
square root of this factor.
Fundamental limits: thermal noise. The fluctuation-dissipation theorem states that any dissi-
pative system will experience thermally driven fluctuations in its modes. Such thermal noise
effects arise significantly in two places in an interferometer: through dissipation in the flexure
of the suspension systems used to make the mirrors 'free' above the pendulum resonance; and
through dissipation of the internal modes of the mirrors. Nearly all of each mode's kTs worth
of thermal energy is concentrated at the resonant peak, but a small fraction is distributed in
frequency above and below the resonance. In the gw frequency band, the suspension thermal
noise is due to above-resonance excitation of the fundamental mode, and the mirror internal
thermal noise comes from below-resonance motion. The quantitative behavior depends on the
material properties, but in general the thermal noise contribution to the mirror displacement
varies as I\x - Jr /(mQ) , where T is the temperature, m the mirror mass and Q the quality factor
of the oscillator. A major goal of the mechanical design of the mirrors and suspensions is the
achievement of low mechanical loss (high quality factors).
-21
10
-23
10
RESIDUAL GASS, 1 o~ TORR H 2
-~L-~~~~--~~~~~~~~~~~~~
10 1 10 100 1000 10000
Frequency (Hz)
Figure 1. INITIAL INTERFEROMETER SENSITIVITY
98
Fundamental limits: gravity gradients. Time varying mass distributions in the vicinity of the
test masses produce fluctuating gravity gradient forces, which in tum produce uncorrelated
test mass motions above a few Hertz. Surface seismic compressional waves and atmospheric
pressure changes are uncontrollable sources; the resulting unshieldable forces set a lower limit
of about 10 Hz on the gravity wave band of an earth-based detector.
The preceding two fundamental limits are due to forces on the test masses, and thus produce
less strain noise linearly as the arm length increases. It is primarily the need to reduce the
effect of thermal noise that leads to the requirement of kilometer-size baselines in the LIGO
and VIRGO detectors.
To conclude, Figure 1 shows the spectrum of the projected limiting sensitivity of the initial
LIGO interferometer. The limits due to individual noise sources are shown in terms of their
equivalent strain noise; because of the random nature of the noise, they are most naturally dis-
played in terms of amplitude spectral density of strain: (5lI1)/(JHz). In addition to the funda-
mental noise sources mentioned above, the effects of seismic noise and residual gas pressure
are also shown. The latter effect is the fluctuation in the optical phase produced by a pressure
fluctuation in the beam path, due to the forward scattering off the gas molecules.
99
SECTION IV - NEUTRINO MASSES
LSND NEUTRINO OSCILLATION RESULTS
W. C. Louis, representing the LSND Collaboration!
Los Alamos National Laboratory

Physics Division
Los Alamos, NM 87545, U.S.A.
INTRODUCTION
In the past several years, a number of experiments have searched for neutrino
oscillations, where a neutrino of one type (say vll ) spontaneously transforms into a
neutrino of another type (say ve). For this phenomenon to occur, neutrinos must
be massive and the apparent conservation law of lepton families must be violated.
In 1995 the LSND experiment published data showing candidate events that are
consistent with vil ~ ve oscillations. 2 Additional data are reported here which
provide stronger evidence for neutrino oscillations. 3
THE LSND EXPERIMENT

The Liquid Scintillator Neutrino Detector (LSND) experiment at LAMPF 3
is designed to search with high sensitivity for vil ~ ve oscillations from /1-+ decay
at rest. LAMPF is a most intense source of low energy neutrinos due to its 1
mA proton intensity and 800 MeV energy. The neutrino source is well understood
because almost all neutrinos arise from 7[+ or /1-+ decay; 7[- and /1-- are readily
captured in the Fe of the shielding and Cu of the beam stop. The production of
kaons and heavier mesons is negligible at these energies. The ve rate is calculated
to be only 4 X 10- 4 relative to vil in the 36 < Ev < 52.8 MeV energy range, so that
the observation of a significant ve rate would be evidence for vil ~ ve oscillations.
The LSND detector consists of an approximately cylindrical tank 8.3 m long by
5.7 m in diameter. The center of the detector is 30 m from the neutrino source. On
the inside surface of the tank 1220 8-inch Hammamatsu phototubes provide 25%
photocathode coverage. A schematic of the LSND detector is shown in Fig. 1. The
tank is filled with 167 metric tons of liquid scintillator consisting of mineral oil and
0.031 gil of b-PBD. The low scintillator concentration allows the detection of both
Cerenkov light and scintillation light and yields a relatively long attenuation length
of more than 20 m for wavelengths greater than 400 nm. 4 A typical 45 MeV electron
created in the detector produces a total of rv 1500 photoelectrons, of which rv 280
photoelectrons are in the Cerenkov cone. The phototube time and pulse height
signals are used to reconstruct the track with an average r.m.s. position resolution
103
16 Jt

/~
C )<J
')t)
c
- (';'l>. ~
..)'
~
p~
i'J
tl
I';;'
<::! 8~
- "" <;;:u ~ ().,
-
ri
k:) ?:ilg
-
Access
I~
"
f.; . - fl - - f')
- ~
"" ~r
Door
, ')
Q
_, c ~ '" )
.) 'C
fl' 0 ,"
.-
~. - 0
1"-'- , > ) , ( ) ). )..,) ,' o ) )
l" :,) ~
,~
!~" , 0
~
uu .
~ 1\ P'l 2
~
," "l () '

~
~
~
t) ,~
c , V ~I.i
'-' ('
IX ['I(D-( ~ - ~
i1 _"~I [~ '~
9y~IX
e j
-J..J .
v
.
~ "fC:J . -JJ
'" ~
.. ~ Pb Shielding
Beam~
5.72 m ..,I a,75m . ~
Figure 1. A schematic of !he LS ND detector.
L 10 I
" " I .u
10
R
Figure 2. Measured R distribution for events wi!h the y correlated (solid) and uncorrelated (dashed)
wi!h !he primary event. The dotted curve is the Monte Carlo R distribution for events wi!h the y correlated.
') i I I II i II T ..""n,n", 'I---,-- r

""1"'- . ,
~ 10 3
"'"OJ
OJ
>(
OJ
E
.is"
"'/
10

I
--1--
!,! ,I
10 10
\
Figure 3. The R distribution, beam on minus beam off excess, for events that have energies in the
range 20 < Ee < 60 Me V. The solid curve is the best fit to the data, while the dashed curve is the compo
nent of the fit with an uncorrelated y.
104
TABLE 1. The 22 beam-on events with R > 30 and energy in the 36 < E. < 60 MeV range.
For each event is given the year recorded, energy, spatial position, and distance from the PMT
surfaces.
Event Year E(MeV) X(cm) Y(cm) Z(cm) D(cm)
1 1993 47.6 -66 -84 -77 115
2 1993 51.1 56 -96 53 103
3 1994 40.1 -36 196 -203 53
4 1994 44.2 69 -146 153 53
5 1994 39.4 -169 96 -347 39
6 1994 36.3 -156 -79 -207 84
7 1994 52.9 21 106 71 143
8 1994 37.0 31 156 -105 93
9 1994 42.4 -14 -121 -239 78
10 1994 37.7 -91 119 209 109
11 1994 54.3 -91 191 269 47
12 1994 55.8 71 -99 -259 100
13 1994 43.8 6 211 173 38
14 1995 49.2 -184 10 58 75
15 1995 56.5 128 -150 199 49
16 1995 37.4 45 -92 -239 1071
17 1995 45.1 -186 105 -126 45
18 1995 46.7 179 -93 -108 57
19 1995 40.2 -37 -71 160 128
20 1995 45.9 -161 87 -337 49
21 1995 36.3 46 150 107 100
22 1995 37.6 -73 107 -257 129
of rv 30 cm, an angular resolution of rv 12 degrees, and an energy resolution of rv 7%.

The Cerenkov cone for relativistic particles and the time distribution of the light,
which is broader for non-relativistic particles, give excellent particle identification.
Surrounding the detector is a veto shield 5 which tags cosmic ray muons going
through the detector.
DATA
The signature for a ve interaction in the detector is the reaction veP ---+ e+n
followed by np ---+ d'Y (2.2 MeV). A likelihood ratio, R, is employed to determine
105
TABLE II. A list of all backgrounds with the expected number of background events in the
36 < E. < 60 MeV energy range for R 2: 0 and R > 30. The neutrinos are from either 7r and J.L
decay at rest (DAR) or decay in flight (DIF). Also shown are the number of events expected for
100% v" -> v. transmutation.
Background Neutrino Source Events with R 2: 0 Events with R > 30
Beam Off 146.5 3.2 2.52 0.42
Beam-Related Neutrons < 0.7 < 0.1

4.8 1.0 1.10 0.22
2.7 1.3 0.62 0.31
7r -> ev and J.L -> evv DIF 0.1 0.1 0
Total with Neutrons 7.6 1.6 1.72 0.38
7r+ -> J.L+v" DIF 8.1 4.0 0.05 0.02
J.L+ -> e+v"v. DAR 20.1 4.0 0.12 0.02
J.L+ -> e+v"v. DAR 22.5 4.5 0.14 0.03
ve -> ve J.L+ -> e+v"v. DAR 12.0 1.2 0.07 0.01
ve -> ve 1.5 0.3 0.01 0.01
7r -> eVe DAR 3.6 0.7 0.02 0.01
0.2 0.1 0
7r -> ev and J.L -> evv DIF 0.6 0.1 0
Total without Neutrons 68.6 7.4 0.41 0.04
Grand Total 222.7 8.2 4.65 0.57
100% Transmutation 12500 1250 2875 345
whether a , is a 2.2 Me V photon correlated with a positron or is from an accidental

coincidence. R is the likelihood that the, is correlated, divided by the likelihood
that it is accidental. R depends on the number of hit phototubes for the" the re-
constructed distance between the positron and the " and the relative time between
the, and positron. Fig. 2 shows the expected R distribution for accidental pho-
tons and correlated photons. Fig. 3 shows the R distribution, beam on minus beam
off, for events with positrons in the 36 < E < 60 Me V energy range. The dashed
histogram is the result of the R fit for events without a recoil neutron, and the solid
histogram is the total fit, including events with a neutron. After subtracting the
neutrino background with a recoil neutron there is a total excess of 54.8:::~~:~ 8.2
events, which if due to neutrino oscillations corresponds to an oscillation probability
of (0.33:::~:~~ 0.05)%.
Fig. 4 shows the electron energy distribution, beam on minus beam off excess,
for events (a) without a , requirement and (b) events with an associated, with
106
0
20 30 40 50 60
~ 10
">
~"~ (b)
u
><
"e
'"
1l
60
positron energy (MeV)
Figure 4. The energy disttibution for events with (a) R ~ 0 and (b) R > 30. Shown in the figure are the
beam- excess data, estimated neuttino background (dashed), and expected disttibution for neuttino oscilla-
tions at large 8m 2 pulse estimated neuttino background (solid).
~ 6
!: 15
" 4
"u><
"e 2
"'"
.c
0
-200 0 200 -200 0 200

positron x-position (em) positron x-position (em)
C 6
"i; 15
4
"u
:l 10
e 2
"'"
.c
0
0
-200 0 200 -200 0
positron y-position (em) positron y-position (em)
~
"> 15
~"
~
>< 10
"e
'""
.c
0
0
-200 0 200
positron z-position (em) positron z-position (em)
Figure 5. The spatial disttibutions for beam-excess data events with 36 < E. <60 MeV. (a) - (c) are for R ~
oand (d) - (f) are for R > 30.
107
-,-:-_--~:~====~~~~-
L~~~d2-"
o~ 10 ,::_?--_-?--J
-I
10
-3
10 10 10 2 1
sin 29
Figure 6. Plot of the LSND &n 2 vs sin 2 29 favored regions. They correspond to 90% and % like-
lihood regions after the inclusion of the effects of systematic errors.
R > 30. For this latter requirement, the total 2.2 MeV, detection efficiency is
23% and the probability that an event has an accidental, in coincidence is 0.6%.
The dashed histogram shows the background from expected neutrino interactions.
There are 22 events beam on in the 36 < E < 60 Me V energy range and a total
estimated background (beam off plus neutrino-induced background) of 4.6 0.6
events. Table 1 lists the properties of these 22 events, while Table 2 shows the
background estimate for events in the 36 < Ee < 60 Me V energy range with R 2: 0
and R > 30. Fig. 5 shows the spatial distributions for the beam on-off excess events
with R ~ 0 and R > 30. The probability that this excess is due to a statistical
fluctuation is < 10- 7 . If the observed excess is due to neutrino oscillations, Fig. 6
shows the allowed region (90% and 99% likelihood regions) of sin 2 28 vs Llm 2 from
a maximum likelihood fit to the LIE distribution of the 22 beam on events. Some
of the allowed region is excluded by the ongoing KARMEN experiment at ISIS,6
the E776 experiment at BNL,7 and the Bugey reactor experiment. 8
CONCLUSION
In summary, the LSND experiment observes an excess of events with positrons

In the 36 < E < 60 Me V energy range that are correlated in time and space
with a low energy,. If the observed excess is interpreted as vI' ~ ve oscillations,
it corresponds to an oscillation probability of (0.33~~:g 0.05%) for the allowed
108
regions shown in Fig. 6. More data taking is planned for the experiment, and
the performance of the detector is under continuous study. Both of these efforts
are expected to improve the understanding of the phenomena described here. If
neutrino oscillations have in fact been observed, then the minimal standard model
would need to be modified and neutrinos would have mass sufficient to influence
cosmology and the evolution of the universe.
REFERENCES
IThe LSND Collaboration consists of the following people and institutions: K.
McIlhany,1. Stancu, W. Strossman, G. J. VanDalen (Univ. of California, Riverside);
W. Vernon (Univ. of California, San Diego and IIRPA); D. O. Caldwell, M. Gray,
S. Yellin (Univ. of California, Santa Barbara); D. Smith, J, Waltz (Embry-Riddle
Aeronautical Univ.); A. M. Eisner, Y-X. Wang (Univ. of California IIRPA); 1.
Cohen (Linfield College); R. L. Burman, J. B. Donahue, F. J. Federspiel, G. T.
Garvey, W. C. Louis, G. B. Mills, V. Sandberg, R. Tayloe, D. H. White (Los
Alamos National Laboratory); R. M. Gunasingha, R. Imlay, H. J. Kim, W. Metcalf
(Louisiana State Univ.): K. Johnston (Louisiana Tech Univ.); B. D. Dieterle, R. A.
Reeder (Univ. of New Mexico); A. Fazely (Southern Univ); C. Athnassopoulos, L.
B. Auerbach, R. Majkic, J. Margulies, D. Works, Y. Xiao (Temple Univ.).
2C. Athanassopoulos et. al. , Phys. Rev. Lett. 75, 2650 (1995).
3C. Athanassopoulos et. al. , submitted to Phys. Rev. C.
4R. A. Reeder et. al. , Nucl. Instrum. Methods A 334, 353 (1993).
5J. J. Napolitano et. al. , Nucl. Instrum. Methods A 274, 152 (1989).
6B. Bodmann et. al. , Phys. Lett. B 267, 321 (1991); B. Bodmann et. al. , Phys.
Lett. B 280, 198 (1992); B. Zeitnitz et. al. ,Prog. Part. Nucl. Phys., 32 351
(1994).
7L. Borodovsky et. al. , Phys. Rev. Lett. 68, 274 (1992).
BB. Achkar et. al. , Nucl. Phys. B434, 503 (1995).
109
THEORETICAL IDEAS ABOUT NEUTRINO MASS
Lincoln Wolfenstein
Carnegie Mellon University
Pittsburgh, PA 15213
USA
What insight does theory give us about neutrino masses? When parity was dis-
covered in 1957 it was immediately suggested that the neutrino was a massless Weyl
particle having only two components, the left-handed VL and the right-handed anti-
particle VR. With the advent of the standard model all fermions start out as Weyl
particles: these are the left-handed doublets
(1)
and the right-handed singlets
It is perfectly natural to also include right-handed neutrino singlets, but in the

usual form of the standard model they are arbitrarily excluded in order to make the
neutrinos massless.
Fermions obtain masses in the standard model via the Yukawa interaction (that
couples the left-handed doublets, the right-handed singlets, and the Higgs doublet)
after the Higgs obtains its vacuum expectation value. The left and right Weyl particles
merge to form a Dirac fermion. This is the ugliest part of the standard model since the
original interaction Lagrangian contains three arbitrary 3 x 3 matrices in flavor space,
one for ups, one for downs, and one for leptons. Indeed one of our hopes is that the
pattern of the neutrino masses may provide a new clue to the better understanding of
masses in general.
There exists a striking qualitative symmetry between the quarks and leptons in
(1). There are three generations of doublets and the charged leptons display a mass
111
hierarchy similar to that of the down quarks. The lepton doublets have the same weak
coupling to the Was do the quarks. Thus there exists the exciting possibility that this
is a broken symmetry which becomes exact at some high mass scale. Pati and Salam
called this symmetry the SU(4) of color. In the grand unified theory (GUT) SO(lO)
leptons and quarks are different components of a single representation (the 16). The
lepton-quark symmetry requires the existence of VR.
In SO(lO) it is natural that the Yukawa interactions connect the VR to VL with
the couplings of the same general magnitude as the up quarks. One might then expect
neutrinos to become Dirac particles with masses like the quarks. Gell-Mann, Ramond
and Slansky addressed this problem of SO(lO). Their solution was that when the
quark-lepton symmetry was broken, as part of the process of breaking SO (10) to the
standard SU(3) x SU(2) x U(I), the VR obtained large Majorana masses of order M,
the scale of this symmetry breaking. There resUlts a 6 x 6 mass matrix of the form
( 0 mD) (2)
mD mm
where mD is the usual Dirac mass matrix from the Yakawa interaction and mm is the
large Majorana mass matrix. In Mohapatra's talk the question is raised as to whether
the left-hand "zero" can be ignored. From Eq. (2) one obtains the see-saw formula
m2
m(v),...., -
M
where m is of the order of up-quark masses.
Quantitative predictions of SO(lO) depend not only on the magnitude of M but
also on the detailed structure of the matrices mD and mm. There are countless detailed
schemes in the literature giving quite different results for neutrino masses and mixings.
There are some general qualitative results that hold in most schemes but not all:
1. Neutrino masses follow the same hierarchy as other fermions
2. Mixings are neither very large or extremely small with the mixing of the first and
third generation the smallest.
Within SO(lO) there are two possible approximate magnitudes for M, each of
which leads to a scenario for solving the solar neutrino problem. In the simplest non-
SUSY SO(10) the breaking takes place in two stages with the second stage leading to
the breaking of quark-lepton symmetry with M of order 10 12 Gev. In this scenario
Ve - vlJ. MSW oscillations can explain the solar neutrinos with m 2 (vlJ.) about 10- 5 ev 2
In this case m (v.,.) could be of order of ev and thus be relevant for dark matter provided
vlJ. - v.,. mixing is small enough to have evaded detection. Alternatively m (v.,.) could
be close to 10- 1 ev, in which case, if vlJ. - v.,. mixing were larger than the simplest
expectation, one could fit the atmospheric neutrino anomaly.
In the simplest SUSY SO(10) there is a single breaking scale M of order 10 16 Gev.
In this case ve - v.,. MSW oscillations could provide the standard solar neutrino solutions
with m 2 (v.,.) about 10- 5 ev 2 . However in this case there is an additional possibility that
Ve - vlJ.oscillations may also effect the solar neutrinos. If you look at the standard plots
of neutrino survival rates with MSW oscillations it appears that m (v.,.) around 10-2 ev
could explain the chlorine and Kamiokande results but not the gallium result, but that
112
some additional Ve - vI-' oscillations with vI-' around 10-4 ev could provide additional
suppression for low energy neutrinos.
Now I would like to examine a more extreme possibility if one forgets about the
standard solar model (SSM) and simply asks what can we deduce from the observed
neutrino fluxes alone. The main purpose of this exercise is to emphasize what can be
learned from forthcoming experiments. We assume that we do not understand the sun's
interior but that the sun's energy arises from the standard nuclear reactions and that
the luminosity is essentially constant so that the present luminosity measures the energy
output of the reactions. Then the solar neutrinos provide the information as to the roles
of the PPI, PPII, PPIII, and CNO cycles in producing this energy. Carrying out this
exercise recently, Bahcall, Fukugita, and Krastev suggested that it was possible that
nearly all the neutrinos observed originated from the CNO bicycle. This is obviously
an extreme deviation from the SSM since it requires the central temperature Tc some
30% greater than usual calculations.
I have found a somewhat less extreme possibility. Let the 8 B neutrino true flux
be 2 to 3 times the SSM values with 90 to 100% of these Ve converted to V r In this
case most of the neutrinos detected by the Kamiokande experiment are due to Vr - e
scattering and the Ve flux at earth is less than 20% of the SSM value. (Note that
the cross-section for Vr - e (or vI-' - e) scattering is 6 to 7 times smaller than Ve - e
scattering. Thus the observation that the Kamiokande detected rate is one-half the
SSM prediction could be explained by a solar Ve flux 3 times the SSM value with
essentially all Ve converted to v r .) The Davis chlorine detector is only sensitive to V e , of
course, so that with these assumptions most of the neutrinos detected by Davis must
be from 7 Be. The flux of 7 Be Ve arriving at earth is then 1 to 2 times the SSM value. In
this case at least one-half and perhaps nearly all the neutrinos observed in the gallium
experiment are due to 7 Be. Since we are explaining the solar neutrino luminosity by
the P P cycles one must assume that most of the pp neutrinos have been converted to
vI-' in order to explain the observed gallium detection rate.
This represents an extreme version of the second scenario mentioned above. The
Vr mass near 10- 2 ev leads to ve - vr MSW oscillations that convert most of the
8 B neutrinos but not the 7 Be neutrinos. The Ve - vI-' MSW oscillation with the vI-'
mass near 10-4 ev converts most of the pp neutrinos but a smaller fraction of the 7 Be
neutrinos. Thus in contrast to the standard MSW scenario the main survivors are the
7 Be neutrinos.
If this possibility were true the SNO experiment would observe a rate for charged
current reactions a factor of five or more below the SSM prediction but a neutral current
rate a factor of 2 or more greater. This serves to emphasize the importance of the SNO
results. Similarly with this possibility the Borexino experiment would measure a rate
greater than the SSM. Of course it should be emphasized that there is no reason to
expect these deviations from the SSM.
In conclusion I believe there is good reason to believe that neutrinos may be
massive and that many different experiments are needed to probe the possibilities.
This work was supported in part by the U.S. Department of Energy Contract No.
DE-FG02-91ER40682.
113
A Bayesian Analysis of Solar Neutrino Data
Harrison B. Prosper
Introduction
One of the greatest scientific achievements of this century was the experimental con-
firmation that the sun is a copious emitter of neutrinos and the consequent confirmation
of Sir Arthur Eddington's hypothesis! that the sun shines because of the energy released
through nuclear reactions that take place in its core. This hypothesis was developed in
detail by Hans Bethe and others and their work forms the basis of mathematical mod-
els of the sun, called standard solar models2 , that have been developed into accurate
descriptions of the sun's observed characteristics.
These models are so accurate in fact that discrepancies between their predictions
and the experimental observations are taken seriously. One discrepancy has plagued
the field for over twenty-five years: the observed flux of solar neutrinos is lower than
that predicted by standard solar models.
For a long time the Chlorine solar neutrino experiment at the Homestake 3 mine
was the only source of the discrepancy. With only one experiment challenging theory
it was easy to brush aside the problem. Today, however, the problem is more acute:
the Gallium experiments, SAGE4 and GALLEX 5 together with the water experiment
Kamiokande6 not only confirm the discrepancy but also provide crucial information
about its nature. The problem is of such interest that it has inspired the development
of a new generation of experiments-SNO(1996), SuperKamiokande(1996), BOREX-
INO(1997), ICARUS(199g) and HELLAZ(2000)-whose collective aim is to resolve the
issue once and for all. But what precisely is the problem?
The Solar Neutrino Problem
The Homestake experiment detects neutrinos through the reaction
(1)
The Argon atoms are chemically extracted and counted. The SAGE and GALLEX
experiments rely on the reaction
(2)
115
while the Kamiokande experiment uses neutrino electron scattering:
(3)
The results of the radiochemical experiments, Homestake, SAGE and GALLEX, are
usually quoted in solar neutrino units (SNU): one SNU is 10-36 neutrino reactions per
atom per second. The neutrino flux for the water experiment is usually given as a
fraction of that predicted by a standard solar model, for example that of Bahcall and
Pinsonneault .
In the table below I give the predictions of two standard solar models, those of
Bahcall and Pinsonneault (BP) and those of Turck-Chieze and Lopes (TCL), together
with the experimentally measured rates. The results for SAGE and GALLEX have
been combined.
Experiment Measured Rates BP TCL
(CI) Homestake (SNU) 2.55 0.25 81 6.4 1.4
(Ga) SAGE and GALLEX (SNU) 74 9.5 132 7 123 7
(H 2 0) Kamiokande (BP) 0.51 0.07 1 0.14 0.77 0.19
The solar neutrino problem is the discrepancy between the measured and predicted
rates in all experiments.
But are the rates really discrepant? And if so by how much? And what information
can we extract about the individual neutrino fluxes? These questions have been investi-
gated intensively by many authors 7 all of whom agree about the physics of the problem
and come to the same broad conclusions. However, as noted by Hata and Langacker 7
there are differences, most of which are traceable to differences in the definitions of the
X2 functions used by different authors and their differing definitions of confident levels
and limits. Unfortunately, the definitions tend to be ad hoc and so if we wish to make
detailed comparisons of results we are, to some degree, forced to compare "apples with
oranges". The problem though is deeper than this: we are using a method of analysis,
the X2 fit, that was not designed to answer the kinds of questions that we wish to ask.
In the rest of the paper I illustrate an alternative approach that is better able to answer
a greater variety of questions, simpler: Bayesian inference.
Information Mining using Bayesian Inference
Given a model that connects neutrino fluxes cp to measured rates x, what is the
simplest way to infer information about the individual neutrino fluxes? The answer is
to use Bayes' theorem:
P(cpi I) = L(xicp,I)P(cpiI) (4)
x, i<pL(xicp,I)P(cpiI)'
where L(xicp,I) is the likelihood assigned to the measured rates x, P(cpiI) == f(cp)dcp
is the prior probability assigned to hypotheses about the fluxes cp and P( cpix, I) is the
posterior probability for cp given that we have measured x. The I reminds us that
these probability assignments depend on all kinds of other information not explicitly
stated. The key concept here, that is absent in the X2 approach, is the notion of the
"probability of an hypothesis". This notion allows us to answer any sufficiently clearly
stated question about the fluxes.
Let us first consider the likelihood function L(xicp,I). The model I shall use is
x = R cp, (5)
116
6.2 1.2 0.0 0.6)
R = ( 13.8 35.2 70.8 12.2 . (6)
1.0 0.0 0.0 0.0
The fluxes are given in units of the predicted fluxes of the 1992 Bahcall-Pinsonneault
standard solar modeF. The elements of the matrix R have been culled from Bahcall's
book 1 . They show how the different neutrino fluxes B, Be, pp and o, from the 8 B,
7 Be, pp and other nuclear reactions, respectively, contribute to the measured rates
XCI, XGa and XH2 0. We note that the water experiment measures exclusively the flux of
neutrinos from the 8 B reactions, the Chlorine experiment samples the neutrinos from
the 8 Band 7 Be reactions and the Gallium experiments sample all neutrinos, including
those from the proton-proton (pp) chain, which account for the bulk of the neutrinos
emitted by the sun. In the absence of information to the contrary I'll assume the mea-
sured rates to be uncorrelated so that the error ma.trix u associated with the rates
xT = (2.55,74,0.51) is diagonal, with diag-u = (0.25,9.5,0.072). The best represen-
tation of these data-in the sense that their probability assignment has the highest
entropy-is the Gaussian likelihood function
(7)
where the vector z = u-1.(x-R.). We should be clear what this does not mean: it does
not mean that we have assumed the experimental errors to be Gaussian distributed,
although this may be the case. What it means simply is this: given the reported
data, and no further information, a Gaussian likelihood is the best description of this
knowledge.
What about the prior probability P( 11)? Well, one thing we know for sure is that
> 0; but not much else! We shall assume that our lack of knowledge about the fluxes
can be represented by taking P( 11) ex constant. Other choices are possible, and the
result will depend upon which choice is made. However, provided that the data are
sufficiently precise, which they now are, any reasonable prior probability will do and
will give, within the uncertainties, the same inferences about the fluxes.
Applying Bayes' theorem, Eq.(4), we obtain
(8)
From the posterior probability we can derive many results. For example, by integrating
over the fluxes B, Be and o
(9)
we derive the posterior probability P( pplx, 1) for the pp flux alone; and likewise for B
and Be' The posterior probabilities for these fluxes are shown in Fig. 1. The results
indicate a slight suppression of the pp neutrino flux, whereas the 8 B neutrino flux is
down by about a factor of 2 and that of 7 Be even more. These conclusions, of course,
are not new; they have been reached by many authors. The point I wish to make is
the directness and simplicity with which conclusions can be drawn within the Bayesian
approach. Another noteworthy, but unrelated, point is that nowhere have I imposed
the solar luminosity constraint, as is often done in other analyses. By not imposing the
constraint I allow for the possibility that the constraint is violated.
117
3 . 6
-e-=
0.
0.
-e-
"t:I 2.5
-. ~ 5
----
~~
2
----
~
4
~
-~..
-e-= 3
-0.
-e-o. 1.5
'-' '-'
~ ~
"t:I 1 "t:I 2
0.5 1
00 1.5 0.5 1 1.5

<Ppp <P Be
-e-=
12
~ 10
----
~
~
-=
-e-
8
'-'
~
6
"t:I
4
2
J
00 0.5 1 1.5
<PH
Figure 1: Posterior probabilities for the pp, 7 Be and 8 B normalized neutrino fluxes.
We noted that the 7 Be neutrino flux appears to suffer a greater suppression than
that of 8 B. This is very difficult to explain by any plausible modification of solar
models. This opens the exciting possibility that the resolution of the solar neutrino
problem may require new physics. But how significant is this differential suppression?
Or, to ask a more precise question: What is the probability of the hypothesis that
Be < B? To answer this question we need the probability
(10)
from which we can calculate readily Prob( Be < B). The answer is about 86%. In the
standard x 2 approach questions of this nature cannot be answered so simply because
the notion "probability of an hypothesis" does not exist.
Another interesting question that we can answer is how discrepant are theory and
experiment? Let's define the discrepancies tl == th - e., between the theoretical
fluxes and those extracted from the measured rates. We can apply Bayes' theorem
again, but this time taking as our prior probability for the fluxes ePB and ePBe that
which has been learnt from the measured rates, namely: P( ePB, ePBelx, 1). Hence,
(11)
where y = (1.0,1.0) are the normalized theoretical 8 Band 7 Be neutrino flux values
and the likelihood function-which now pertains to the theoretical results only-is a
2-D Gaussian that takes account of the known correlation between these fluxes and
118
the accuracy with which they are predicted. Figure 2 shows a contour plot of the
"discrepancy" distribution P( tl<PB, tl<PBe lx, y,1). From it a host of interesting questions
can be answered by performing appropriate integrations.
= 1.2
~
o.s
0.6
0.4
0.2
00 1 1.2
Ll<PRe
Figure 2: Contour plot of the discrepancy distribution. Notice that the contours are
centered far from the origin which indicates a significant discrepancy between theory
and experiment.
Conclusions
The solar neutrino problem appears to be real. However, quantitative statements

about the nature of the problem become entangled in details of the statistical analysis,
that vary somewhat amongst authors. I believe the principal problem is that we are
using a tool, namely a X2 fit, that is just too crude for the task at hand. A better tool
to use is Bayesian inference. It has the virtue of being both simple and powerful. It
can cope with all kinds of complications, for example constraints such as "the fluxes
are non-negative" or "the data contain systematic uncertainties", that would leave the
standard approach flailing hopelessly. Of course, the Bayesian approach has its own
complication: an appropriate prior probability must be chosen, which choice is not at
all obvious when we know almost nothing about a parameter. But, for the moment, I
see nothing better and as our analyses grow ever more complicated, I think, the utility
of the Bayesian approach will become ever more apparent.
119
References
1. An excellent account of the subject can be found in, J.N. Bahcall, Neutrino Astro-
physics, (Cambridge University Press, Cambridge 1989).
2. J.N. Bahcall and M.H. Pinsonneault, Rev. Mod. Phys. 64, 885 (1992). Rev. Mod.
Phys., 67, 781 (1995).
3. B. T. Cleveland et al., Nucl. Phys. B (Proc. Suppl.) 38,47 (1995).
4. P. Anselmann et al., Phys. Lett. B327, 377 (1995).
5. P. Nico et al., in Proceedings of the XXVII International Conference on High Energy
Physics, Glasgow, 1994, ed. by P.J. Bussey and I.H. Knowles (lOP, Bristol), p.
965.
6. Y. Suzuki et al., Nucl. Phys. B (Proc. Suppl.) 38,54 (1995).
7. See, for example, N. Hata and P. Langacker, Phys. Rev. D50, 632 (1994) and S.
Parke, Phys. Rev. Lett. 74,839 (1995) and references therein.
120
ULTRAHIGH-ENERGY NEUTRINO INTERACTIONS
AND NEUTRINO TELESCOPE EVENT RATES *
Raj Gandhi,l Chris Quigg,2 M. H. Reno,3 and Ina Sarcevic4
1Mehta Research Institute,

10, Kasturba Gandhi Marg, Allahabad 211002, India
2Fermi National Accelerator Laboratory,
Batavia, IL 60510 USA
3Department of Physics and Astronomy,
University of Iowa, Iowa City, IA 52242 USA
4Department of Physics,
University of Arizona, Tucson, AZ 85721 USA
ABSTRACT
We present results for neutrino-nucleon cross sections for energies up to 10 2l eV, of

relevance to the detection of ultrahigh energy galactic and extragalactic neutrinos.
At the highest energies, our results are about 2.4 times larger than previous estimates.
Using these new cross sections, we predict neutrino telescope event rates for the upward
moving muons initiated by the neutrino interactions in the Earth and for the contained-
vertex events in the PeV range due to neutrino-electron interactions. We show that
future neutrino detectors, such as AMANDA, BAIKAL, DUM AND and NESTOR have
a very good chance of detecting neutrinos which originate in the Active Galactic Nuclei.
INTRODUCTION
The Active Galactic Nuclei (AGN), with typical luminosities in the range 1042 to
48
10 erg/s, are believed to be the most powerful individual sources of radiation in the
Universe. These extragalactic point sources are also considered as prodigious particle
accelerators presumably powered by the gravitational energy of matter spiraling in to a
supermassive black hole, though the mechanism responsible for the conversion of grav-
itational energy to luminous energy is not presently understood. Recent detection of
energetic photons (E'"( '" 100 MeV) from about 40 AGNs by the EGRET collaboration l
and of TeV photons from Mkn 421, Mkn 501 2 and most recently from lES2344+514
Talk presented by I. Sarcevic at International Conference on Neutrino Mass, Dark Matter and
Gravitational Waves, Miami Beach, Florida, January 25 - 28, 1996.
121
by the Whipple collaboration3 have created new excitement in the field of high-energy
gamma-ray physics. If the observed photons are decay products of 7rS produced in
hadronic interactions in the disk surrounding the AGN, then AGNs are also powerful
sources of ultrahigh-energy (UHE) neutrinos. 4 Unlike photons, which are absorbed by
a few hundred gm/cm 2 of material, TeV neutrinos have interaction lengths on the order
of 250 kt/cm 2 and thus can provide a direct window to the most energetic processes in
the universe.
The advantage of the long interaction length translates to a challenge in the detec-
tion of neutrinos. Interaction rates increase with energy, but the fluxes of UHE neutri-
nos are steeply falling functions of neutrino energy. Cerenkov detection of muons from
interactions of muon neutrinos in the rock or ice surrounding the detector is feasible. 5
More difficult is the detection of charged-current interactions of electron neutrinos.
Large-area air shower arrays or large volume underground detectors may be adequate
for the detection of electron neutrinos, especially near the W-boson resonance in Vee
collisions. Theoretical calculations of the neutrino-nucleon and neutrino-electron cross
sections are instrumental in evaluating event rates for neutrino telescopes.
Here we present results 6 for charged current and neutral current cross sections
for energies up to 10 21 eV obtained using new parton distributions measured in ep
collisions at HERA. 7 We also discuss how detection of UHE neutrinos depends on
these cross sections and on the neutrino fluxes from UHE neutrino sources. Event rates
for muon neutrino conversions to muons are compared with earlier results based on
older parton distribution functions. 8 We also present results for contained events with
higher threshold energies.
SOURCES OF UHE NEUTRINOS

A variety of sources may contribute to the neutrino flux at the surface of the
Earth. Three types of sources are discussed here: atmospheric neutrinos from cosmic-
ray interactions in the atmosphere, neutrinos from active galactic nuclei, and cosmic
neutrinos from extragalactic cosmic ray interactions with the microwave background
radiation. Model predictions for neutrino fluxes from these three types of sources are
shown in Figure 1. Atmospheric neutrinos 9 (ATM), while interesting in their own right,
mask extraterrestrial sources for E" < 1 TeV. Consequently, we restrict our discussion
to neutrino energies above 1 TeV.
The TeV photons observed by Whipple collaboration 2 may be byproducts of
hadronic cascades initiated by the protons generated within the AGN accretion disk of
gas, or in the jets, which interact with matter or radiation in the AGN disk, to produce
pions whose decay products include both photons and neutrinos. The structure of the
corresponding hadronic cascade is:
pp --+7r+X
PI --+7r+X
np --+Jr+X
7r 0 --+ 'Y + 'Y
Jr --+ vI' + I-"
I-" --+ VI' + Ve + e
If charged and neutral pions are produced in equal proportions and photons originate
in hadronic cascades, simple counting leads to equal fluxes of photons and VI' + Vw The
flux of Ve + ve equals half of the flux of vI' + Vw The observed photon energy spectrum
122
10-7~~~=r~mrn=~~~~~~~~~~
...,......,
I 10-8
...I
"..
rt.I 10- 9
N
rt.I 10- 10
I
E! 10- 11
'--'
()
10- 12
10- 13
10- 14
10- 15
10- 16
10- 17
10- 18
;>
r.:l 10- 19
10-20~~LL~~~~ww~uw~ww~~~~~~
1000 10 4 105 10 6 107
Ell [GeV]
Figure 1: Muon neutrino plus antineutrino fluxes at the Earth's surface: angle-averaged
flux from atmospheric neutrinos (ATM), diffuse flux from active galactic nuclei (AGN-NMB,
AGN-SP and AGN-SS) and cosmic neutrinos (CR-2 and CR-4). The Frejus upper limit 14 on
a neutrino flux in excess of atmospheric neutrino flux is indicated at 2.6 TeV. The dotted line
indicates the vertical flux of atmospheric Jl + fl from Ref. 15.
is a power-law with 2
dN-y "" E- 2
dE-y -y
for 100 MeV:::; E-y :::; 2 Te V, and the same for neutrinos. We have chosen three represen-
tative fluxes of neutrinos from AGN, each corresponding to diffuse flux integrated over
all AGNs. These fluxes are shown in Figure 1. The Nellen, Mannheim and Biermann
flux 10 (AGN-NMB), which comes from assuming that pp collisions are the dominant
neutrino source, is parameterized by:
with the Ve + ve spectrum assumed to be 1/2 of vI" + VI" The neutrino luminosity of
a source is normalized to the observed diffuse x-rays and ,-rays. The NMB param-
eterization is valid for E" :::; 4 X 10 4 GeV. In our calculations described in the next
section, we have used this parameterization up to E" = 108 GeV. A somewhat differ-
ent assumption of the luminosity is used by Szabo and Protheroel l (AGN-SP) in their
extended model of neutrino sources, yielding a higher normalization of dN/ dE" at 1
TeV. Above Ell > 106 GeV, the AGN-SP follows a steeper power law,
which accounts for the lack of protons at even higher energies required to produce
neutrinos. The Stecker and Salamon flux 12 (AGN-SS) contains contributions from
both pp and PI interactions in the accretion disk and has a nearly constant value of
dN/dE" up to E" "" 105 GeV.
123
Two models of neutrino fluxes from cosmic ray interactions with the microwave
background 13 are labeled CR-2 and CR-4 in Figure 1. The fluxes depend on the red shift
of the cosmic ray sources. Maximum redshifts contributing are Zmax = 2 and Zmax = 4,
respectively.
The electron neutrino plus antineutrino fluxes, to a good approximation, are equal
to half of the fluxes shown in Figure 1.
UHE MUON NEUTRINOS

The primary means of detection of muon neutrinos and antineutrinos is by charged-
current conversion into muons and antimuons. The long range of the muon means that
the effective volume of an underground detector can be significantly larger than the
instrumented volume. For example, a 10 TeV muon produced by a charged-current
interaction in rock will propagate several kilometers in water-equivalent distance units
before its energy is degraded to 1 TeV.
Backgrounds to AGN sources of vJ1, + vJ1, include atmospheric neutrinos and at-
mospheric muons. Muons produced by cosmic ray interactions in the atmosphere mask
astrophysical signals unless detectors are very deep underground, muon energy thresh-
olds are set very high, or one observes upward-going muons. We evaluate here event
rates for upward-going muons produced in the rock surrounding the detector, for muon
energy thresholds above 1 TeVand 10 TeV.
The neutrino-nucleon cross section comes into the calculation of the event rate
in two ways. The probability of conversion VJ1, --+ 11 is proportional to the vN charged
current cross section. In addition, the neutrino flux is attenuated by passage through
the Earth. In the next section we describe our calculation of the neutrino-isoscalar
nucleon (vN) cross section. The vN charged-current reaction is the dominant source of
neutrino interaction except in a very narrow energy window at the W-boson resonance.
SMALL-x PARTON DISTRIBUTION FUNCTIONS AND a(vN)

The inclusive cross section for VJ1, +N --+ 11- +X is given by
(1)
where x = Q2 12M v, y = v lEv, with _Q2 being the momentum transfer between the
neutrino and muon, and v the lepton energy loss in the lab frame, v = Ev - Ew M is
the mass of the nucleon and Mw is the mass of the W-boson, while the Fermi constant
is GF = 1.16 X 10- 5 GeV-2. Taking the target as isoscalar nucleons, in terms of the
parton distribution functions for the proton,
2)
q (x, Q =
UV +
2
dv Us + ds
+ --2- + Ss + bs (2)
_
q(x,Q
2) Us +
ds
=-2-+cs+ts (3)
where we have written explicitly valence (v) and sea (s) distributions.
The general form of the cross section shows that at low energies, where the four-
Fermi approximation is valid, a rv E. At higher energies, the W-boson propagator
plays an important role. The value of (Q2) saturates at rv Mar, and x rv Mar 1(2M EvY)
decreases. For neutrino energies above 10 5 GeV, the small-x (x ::; 3 x 10- 2) behavior
of the parton distribution functions becomes important for the evaluation of the cross
section.
124
1000
.... : .... Q2-M
C\2
......... 100 - W2
.........
........
N
CY 10
><
tij
+
........ 1 CTEQ-DIS
N
CY EHLQ-DLA
>< 0.1 CTEQ-DLA
15
'--' MRS D-,G,A'
><
0.01
Figure 2: Comparison of the light-quark sea at Q2 = Ma, for various parton distribution
functions. Of the MRS distributions, D_ (A') is the most (least) singular.
10- 30
10- 31 CTEQ-DIS
EHLQ-DLA
10- 32 CTEQ-DLA
MRS D-,G,A'
~ 10- 33
C\l
8 ()
10- 34
Z;:,. 10- 35
'-"
u
u
b 10- 36
10- 37
10-38Lu~~~~wL~~~~~~W=~~~~~~~
10 100100010 4 10 5 10 6 10 7 lOB 10 9 101010111012
Ell [GeV]
Figure 3: The charged-current cross section for the CTEQ-DIS, CTEQ-DLA, EHLQ-DLA,
MRS A', MRS G and MRS D_ parton distribution functions. The data point, an average of
ZEUS and HI, is from Ref. 17.
125
Neutrino charged-current interactions have been measured directly in laboratory
experiments for neutrino energies up to Ev = 300 GeV.16 Charged-current ep scattering
at HERA, equivalent to Ev = 47.4 TeV, can be translated to a value of a(vN)P Recent
ZEUS and HI measurements at HERA7 of F;P at small-x (10- 4 :S x :S 10- 2) and for
a large range of Q2, 4 GeV2 :s Q2 :s 1600 GeV2 have provided valuable information
about parton densities at small-x and low-Q2. To evaluate the neutrino-nucleon cross
section at ultrahigh energies, extrapolations beyond the measured regime in x and Q2
are required.
There are two main theoretical approaches in the evolution in Q2 of parton densi-
ties: Gribov-Lipatov-Altarelli-Parisi 18 (GLAP) evolution and Balitskii-Fadin-Kuraev-
Lipatov 19 (BFKL) evolution. In the GLAP approach, parton distribution functions are
extracted at modest values of Q2 and evolved to higher scales. The BFKL approach
involves a leading D:.ln(l/x) resummation of soft gluon emissions, which generates a
singular behavior in x at an initial scale Qo,
xqs(x, Q~) rv x-O. 5 (4)

for small x, which persists at higher values of Q. In our extrapolation of the parton
distribution functions outside the measured region, we use GLAP evolution with input
at Qo = 1.6 GeV,
(5)
The value of ). is determined by fits to deep-inelastic scattering and hadron-hadron data
by the MRS20 and CTEQ21 Collaborations. The MRS set A' has). = 0.17, the MRS
set G has). = 0.07 while the MRS set D_ has). = 0.5. All of the MRS distribution
function are fitted using the MS factorization scheme. The CTEQ-DIS, using the deep-
inelastic scattering factorization scheme, has ). = 0.33. These distribution functions
are extrapolated using the power law fit to the distribution functions at x = 10- 5 and
Q = Mw. We have also extrapolated the leading order CTEQ distributions using
the double-log approximation. 22 For reference, the Eichten et al. 23 parton distribution
functions, extrapolated using the double-log approximation, are also shown. The spread
in values for the parton distribution functions is an indication of the uncertainty in
evaluating the v N cross section.
For each of these sets of distribution functions, we have evaluated the neutrino-
nucleon cross section. Figure 3 illustrates the range of predictions as a function of
neutrino energy. Also shown is the average of HI and ZEUS effective neutrino nucleon
cross sections,17 There is excellent agreement among the predictions of the MRS D_,
G, and A' distributions and the CTEQ3 distributions up to Ev :::::J 107 GeV. Above that
energy, our DLA modification of the CTEQ3 distributions gives a lower cross section
than the full CTEQ3 distributions (CTEQ-DIS), as expected from its less singular
behavior as x - t O. At the highest energy displayed, the most singular (MRS D_)
distribution predicts a significantly higher cross section than the others. Above about
106 GeV, the EHLQ-DLA distributions yield noticeably smaller cross sections than
the modern distributions. Plots similar to Figure 3 for antineutrino-nucleon charged
current interactions, as well as neutral current interactions, can be found in Ref. 6.
For charged current and neutral current interactions, for 1015 eV :S Ev :S 10 21 eV, the
cross sections follow a simple power law, for example
E )0.402
acc(vN) = 2.69 x 1O-36 cm 2 ( 1 G:V
126
Table 1: Number of upward JL +P, per year per steradian for A = 0.1 km 2 and E::nn = 1
TeV.
Fluxes EHLQ-DLA CTEQ-DIS

AGN-SS12 82 92
AGN-NMBlO 100 111
AGN-SPll 2660 2960
ATM9 126 141
Table 2: As in Table 1, but for E::nn = 10 TeV.
Fluxes EHLQ-DLA CTEQ-DIS

AGN-SS12 46 51
AGN-NMBIO 31 34
AGN-SPll 760 843
ATM 9 3 3
NEUTRINO TELESCOPE EVENT RATE

In order to calculate the number of upward-moving muons that can be detected
with neutrino detectors such as AMANDA, BAIKAL, DUMAND II and NESTOR,5
we fold in the neutrino flux and its attenuation in the Earth with the probability that
a neutrino passing on a detector trajectory creates a muon in the rock that traverses
the detector.
The attenuation of neutrinos in the Earth is described by a shadow factor S(E,,),
equivalent to the effective solid angle for upward muons, normalized to 27r:
(6)
where NA = 6.022 X 10 23 mol- l = 6.022 x 10 23 cm- 3 (water equivalent) is Avogadro'S

number, and z(O) is the column depth of the earth, in water-equivalent units, which
depends on zenith angle. 25 The probability that the neutrino with energy E" converts
to a muon is proportional to the cross section and depends on the threshold energy for
the muon E::nn:
(7)
where the average muon range in rock is (R).26 A more detailed discussion appears in
Ref. 6.
The diffuse flux of AGN neutrinos, summed over all AGN sources, is isotropic, so
the event rate is
(8)
given a neutrino spectrum dN"/dE,, and detector area A. As the cross section increases,
Pp. increases, but the effective solid angle decreases.
Event rates for upward muons and antimuons for a detector with A = 0.1 km2
for E::nn = 1 TeV and E::un = 10 TeV are shown in Tables 1 and 2. The CTEQ-DIS
distribution functions are taken as representative of the modern parton distribution
127
Table 3: Downward resonance Vee - W- events per year per steradian for a detector
with effective volume Veff = 1 km3 together with the potential downward (upward)
background from VIJ and vlJ interactions above 3 PeV.
Mode AGN-SS12 AGN-SPll

W - vlJl' 6 3
W - hadrons 41 19
(vlJ,vlJ)N CC 33 (7) 19 (4)
(vlJ,vlJ)N NC 13 (3) 7 (1)
function sets, and compared with the EHLQ-DLA event rate predictions. The muon
range is that of Ref. 26.
The theoretical predictions for ultrahigh-energy neutrinos from AGNs yield event
rates comparable to, or in excess of, the background rate of atmospheric neutrinos for
E:;un = 1 TeV. The AGN-SP rate is large compared to the AGN-NMB rate because
additional mechanisms are included. Flux limits from the Frejus experiment are in-
consistent with the SP flux for 1 TeV< Ell < 10 TeV.14 The atmospheric neutrino
background is greatly reduced by requiring a 10 TeV muon threshold, though AGN
induced event rates are reduced as well. The flatter neutrino spectra have larger con-
tributions to the event rate for muon energies away from the threshold muon energy
than the steep atmospheric flux.
We have evaluated the event rates using the other parton distribution functions
shown in Figure 2. Event rate predictions are unchanged with the other modern parton
distributions because all these distributions are in agreement in the energy range Ell ,....,
1 - 100 TeV. However, our results for event rates are about 15% larger than for the
EHLQ structure functions. This is due to the fact that EHLQ parton distributions were
based on the CERN-Dortmund-Heidelberg-Saclay measurements of neutrino-nucleon
structure functions 16 which had low normalization of about 15%.
UHE ELECTRON NEUTRINOS
Finally we consider event rates from electron neutrino and antineutrino interac-
tions. For YeN (and YeN) interactions, the cross sections are identical to the muon
neutrino (antineutrino) nucleon cross sections. Because of the rapid energy loss or
annihilation of electrons and positrons, it is generally true that only contained-vertex
events can be observed. Since electron neutrino fluxes are small, an extremely large
effective volume is needed to get measurable event rates. There is one exceptional case:
resonant formation of W- in Vee interactions at Ell = 6.3 PeV. The resonant cross
section is larger than the vN cross section at any energy up to 1021 eV. In Fig. 4 we
present neutrino-electron cross sections.
We note that, at the resonance energy, upward-moving electron antineutrinos do
not survive passage through the Earth. However, the contained events have better
prospects for detection. The contained event rate for resonant W production is
10
Rate = -SVeffNA
JdEve (J'iiee(Eiie)S(Ev.)
. dN ve
dE- . (9)
1 II.
We show event rates for resonant W-boson production in Table 3. The background is
for events with Ell > 3 PeV.
From Table 3 we note that a 1 km3 detector with energy threshold in the PeV
range would be suitable for detecting resonant Vee - W events. However, the vlJN
128
10.31
10.33
10.35
10.37
10.39
106 108
Ev [GeV]
Figure 4: Cross sections for neutrino interactions on electron targets. At low energies, from
largest to smallest cross section, the processes are (i) Vee -+ hadrons, (ii) vp'e -+ {tVe, (iii)
Vee -+ Vee, (iv) Vee -+ vp.{t, (v) Vee -+ Vee, (vi) IJ"e -+ v"e, (vii) v"e --> v"e.
background may be difficult to overcome. By placing the detector a few km under-

ground, one can reduce atmospheric-muon background, which is 5 events per year per
steradian at the surface of the Earth for E" > 3 PeV.
SUMMARY
In summary, we find that detectors such as DUMAND II, AMANDA, BAIKAL
and NESTOR have a very good chance of being able to test different models for neutrino
production in the AGNs. 24 For E:;,in = 1 TeV, we find that the range of theoretical
fluxes leads to event rates of 900-29,600 upward-moving muons/yr/km 2 /sr originating
from the diffuse AGN neutrinos, with the atmospheric background of 1400 events/yr
Ikm 2 /sr. For E:;,in = 10 TeV, signal to background ratio becomes even better, with
signals being on the order of 500-8,400 events/yr/km 2/sr, a factor ",20-300 higher than
the background rate. For neutrino energies above 3 Pe V there is significant contribution
to the muon rate due to the ve interaction with electrons, due to the W-resonance
contribution. We find that acoustic detectors with 3 PeV threshold and with effective
volume of 0.2 km 3 , such as DUMAND, would detect 48 hadronic cascades per year
from W -4 hadrons, 7 events from W -4 {tV" and 36 events from V" and vp. interactions
with virtually no background from ATM neutrinos.
Acknowledgements
This work was supported in part by DOE grants DE-FG03-93ER40792, DE-FG02-

85ER40213, and NSF Grant PHY 95-07688. Fermilab is operated by Universities Re-
search Association, Inc., under contract DE-AC02-76CH03000 with the United States
Department of Energy.
REFERENCES
1. C. E. Fichtel, et al., Astrophys. J. Suppi. 94:551 (1994).
2. M. Punch, et ai. (Whipple Observatory Gamma Ray Collaboration) Nature (London)
160:477 (1992); A.D. Kerrick et ai., Astrophys. J. 438:L59 (1995); J. Quinn, et
al. (Whipple Observatory Gamma Ray Collaboration) fA U Circular 6169 (June 16,
1995).
129
3. T. C. Weekes, private communication.
4. See T. Gaisser, F. Halzen and T. Stanev, Physics Reports 258:173 (1995) for a review of
neutrino astronomy and sources of UHE neutrinos.
5. J. Babson et ai., (DUMAND Collaboration), Phys. Rev. D42:3613 (1990); Proceedings
of the NESTOR workshop at Pyios, Greece, ed. L. K. Resvanis (University of Athens,
1993); D. Lowder et. ai., Nature 353:331 (1991).
6. R. Gandhi, C. Quigg, M.H. Reno and I. Sarcevic, hep-ph/9512364, to appear in As-
troparticle Physics (1996).
7. ZEUS Collaboration, M. Derrick et ai., Phys. Lett. B316:412 (1993); HI Collaboration,
I. Abt et ai., Nucl. Phys. B407:515 (1993).
8. C. Quigg, M. H. Reno and T. Walker, Phys. Rev. Lett. 57:774 (1986); M. H. Reno and
C. Quigg, Phys. Rev. D37:657 (1988).
9. L. V. Volkova, Yad. Fiz. 31:1510 (1980) (Sov. J. Nucl. Phys. 31:784 (1980)).
10. L. Nellen, K. Mannheim and P. L. Biermann, Phys. Rev. D47:5270 (1993).
11. A. P. Szabo and R. J. Protheroe, Astropart. Phys. 2:375 (1994).
12. F. W. Stecker, C. Done, M. H. Salamon and P. Sommers, Phys. Rev. Lett. 66:2697
(1991); Errat.: Phys. Rev. Lett. 69:2738 (1992). Revised estimates of the neutrino
flux appear in F. W. Stecker and M. H. Salamon, astro-ph/9501064, submitted to
Space Sci. Rev.
13. S. Yoshida and M. Teshima, Prog. Theoret. Phys. (Kyoto) 89:833 (1993).
14. W. H. Rhode et ai. (Frejus Collaboration), Wuppertal preprint WUB-95-26, to appear
in Astropart. Phys.
15. P. Gondolo, G. Ingelman and M. Thunman, Uppsala preprint TSL/ISV-95-0120, hep-
ph/9505417.
16. S. R. Mishra, et ai. (CCFR Collaboration), Nevis Laboratory Report Nevis-1465 (1992),
and in Lepton-Hadron Scattering, Proceedings of the Nineteenth SLAC Summer In-
stitute on Particle Physics, edited by Jane Hawthorne, SLAC-REPORT-398 (1992),
p.407.
17. Rolf Beyer, 1995 Workshop on Weak Interactions and Neutrinos, Talloires, France.
18. V.N. Gribov and L.N. Lipatov, Sov. J. Nucl. Phys. 15:438 (1972); L.N. Lipatov, Sov.
J. Nucl. Phys. 20:181 (1974); Yu.L. Dokshitser, Sov. Phys. JETP 46:641 (1977);
G. Altarelli and G. Parisi, Nucl. Phys. B126:298 (1977).
19. E.A. Kuraev, L.N. Lipatov and V.S. Fadin, Sov. Phys. JETP 44:443 (1976); 45:199
(1977); Ya.Ya. Balitskii and L.N. Lipatov, Sov. J. Nucl. Phys. 28:822 (1978).
20. A.D. Martin, W.J. Stirling and R.G. Roberts, Phys. Rev. D47:867 (1993); Phys. Lett.
354B:155 (1995).
21. H. Lai et ai., Phys. Rev. D51:4763 (1995).
22. L. V. Gribov, E. M. Levin and M. G. Ryskin, Phys. Rep. 100:1 (1983). See also D. W.
McKay and J. P. Ralston, Phys. Lett. 167B:103 (1986).
23. E. Eichten, I. Hinchliffe, K. Lane and C. Quigg, Rev. Mod. Phys. 56:579 (1984);ibid.
58:1065 (1986).
24. See 7 of T. K. Gaisser, F. Halzen and T. Stanev, Ref. 4, for a review of several models
in addition to the models presented here.
25. A. Dziewonski, "Earth Structure, Global," in The Encyclopedia of Solid Earth Geo-
physics, ed. D. E. James (Van Nostrand Reinhold, New York, 1989), p. 331.
26. P. Lipari and T. Stanev, Phys. Rev. D44:3543 (1991).
130
SECTION V - DIRAC'S LEGACY: LIGHT-CONE
QUANTIZATION
DIRAC'S LEGACY: LIGHT-CONE QUANTIZATION
Stephen S. Pinsky
Columbus, OH 43210
ABSTRACT
In recent years light-cone quantization of quantum field theory has emerged as a

promising method for solving problems in the strong coupling regime. This approach
has a number of unique features that make it particularly appealing, most notably, the
ground state of the free theory is also a ground state of the full theory.
Introduction
One of the central problems in particle physics is to determine the structure of

hadrons such as the proton and neutron in terms of their fundamental quark and gluon
degrees of freedom. Over the past twenty years two fundamentally different pictures
of hadronic matter have developed. One, the constituent quark model (CQM), or the
quark parton model is closely related to experimental observation. The other, quantum
chromodynamics (QCD) is based on an elegant non-abelian quantum field theory. The
light-front formulation of QCD appears to be the only practical hope of reconciling
QCD with the CQM. This elegant approach to quantum field theory is a Hamiltonian
gauge fixed formulation that avoids many of the most difficult problems in the equal
time formulation of the theory. The idea of deriving a null-plane constituent model
from QCD actually dates from the early seventies, and there is a rich literature on the
subject [11, 9, 24]. The main thrust of this talk will be to discuss the complexities of
vacuum that are unique to the light-front formulation of field theories.
An intuitive approach for solving relativistic bound-state problems would be to
solve the gauge fixed, Hamiltonian eigenvalue problem. One imagines that there is an
expansion in multi-particle occupation number Fock states. It is clearly a formidable
task to calculate the structure of hadrons in terms of their fundamental degrees of
freedom in QCD. Even in the case of abelian quantum electrodynamics, very little is
known about the nature of the bound state solutions in the strong-coupling domain. A
calculation of bound state structure in QCD has to deal with many complicated aspects
133
of the theory simultaneously: confinement, vacuum structure, spontaneous breaking of
chiral symmetry (for massless quarks), while describing a relativistic many-body system
which apparently has unbounded particle number. The analytic problem of describing
QCD bound states is compounded not only by the physics of confinement, but also by
the fact that the wavefunction of a composite of relativistic constituents has to describe
systems of an arbitrary number of quanta with arbitrary momenta and helicities. The
conventional Fock state expansion based on equal-time quantization quickly becomes
intractable because of the complexity of the vacuum in a relativistic quantum field
theory. Furthermore, boosting such a wavefunction from the hadron's rest frame to a
moving frame is as complex a problem as solving the bound state problem itself.
Fortunately, "light-front" quantization, which can be formulated independent of
the Lorentz frame, offers an elegant avenue of escape. There are, in fact, many reasons
to quantize relativistic field theories at fixed light-front time. Dirac [12], in 1949, showed
that a maximum number of Poincare generators become independent of the dynamics
in the "front form" formulation, including certain Lorentz boosts. Unlike the equal-
time Hamiltonian formalism, quantization on a plane tangential to the light-front can
be formulated without reference to the choice of a specific Lorentz frame. The eigen
solutions of the light-front Hamiltonian have Lorentz scalars M2 as eigenvalues, and
describe bound states of arbitrary four-momentum and invariant mass M, allowing the
computation of scattering amplitudes and other dynamical quantities.
However, the most remarkable feature of this formalism is the apparent simplicity
of the light-front vacuum. In many theories the vacuum state of the free Hamiltonian
is an eigenstate of the total light-front Hamiltonian. This means that all constituents
in a physical eigenstate are directly related to that state, and not disconnected vacuum
fluctuations. The Fock expansion constructed on this vacuum state provides a complete
relativistic many-particle basis for diagonalizing the full theory .The natural gauge
for light-cone Har.liltonian theories is the light-cone gauge A+ = O. In this physical
gauge the gluons have only two physical transverse degrees of freedom, and thus it
is well matched to perturbative QCD calculations. The simplicity of the light-cone
Fock representation relative to that in equal-time quantization arises from the fact that
the physical vacuum state has a much simpler structure on the light cone. Indeed,
kinematical arguments suggest that the light-cone Fock vacuum is the physical vacuum
state.
The success of the CQM or the Feynman parton model is a powerful for a light-front
formulation of QCD. The ideas of the parton model seem more easily formulated in the
light-front picture of quantum field theory than in the equal-time formulation. This
is a highly desirable feature if one wishes to have a constituent picture of relativistic
bound states and describe, for example, a baryon as primarily a three-quark state plus
a few higher Fock states aLa Tamm and Dancoff.
Studies of model light-front field theories have shown that the zero modes can
in fact support certain kinds of vacuum structure. The long range phenomena of
spontaneous symmetry breaking [18, 5, 42, 22, 46] as well as the topological structure
[26, 43] can in fact be reproduced with a careful treatment of the zero mode(s) of the
fields in a quantum field theory defined in a finite spatial volume and quantized at equal
light-front time. These phenomena are realized in quite different ways. For example,
spontaneous breaking of Z2 symmetry in 4>1+1 occurs via a constrained zero mode of the
scalar field [5, ?, 46]. There the zero mode satisfies a nonlinear constraint equation that
relates it to the dynamical modes in the problem. At the critical coupling a bifurcation
of the solution occurs [19, 46, 5]. One must choose one solution to use in formulating the
theory. This choice is analogous to what in the conventional language one would call
the choice of vacuum state. These solutions lead to new operators in the Hamiltonian
134
which break the Z2 symmetry at and beyond the critical coupling. The various solutions
contain c-number pieces which produce the possible vacuum expectation values of </>.
The properties of the strong-coupling phase transition in this model are reproduced,
including its second-order nature and a reasonable value for the critical coupling[5, 42].
Apart from the question of whether or not VEVs arise, solving the constraint
equations really amounts to determining the Hamiltonian (and other Poincare genera-
tors). In general, P- becomes very complicated when the zero mode contributions are
included; this is in some sense the price one pays to achieve a formulation with a simple
vacuum. It may be possible to think of the discretization as a cutoff which removes
states with 0 < p+ < 7r / L, and the zero mode contributions to the Hamiltonian as ef-
fective interactions that restore the discarded physics. In the light-cone power counting
analysis of Wilson [50] it is clear that there will be a huge number of allowed operators.
Quite separately, a dynamical zero mode was shown in Ref. [26) to arise in pure
SU(2) Yang-Mills theory in 1+1 dimensions. A complete fixing of the gauge leaves
the theory with one degree of freedom, the zero mode of the vector potential A+. The
theory has a discrete spectrum of zero-P+ states corresponding to modes of the flux
loop around the finite space. Only one state has a zero eigenvalue of the energy P-, and
is the true ground state of the theory. The nonzero eigenvalues are proportional to the
length of the spatial box, consistent with the flux loop picture. This is a direct result of
the topology of the space. As the theory considered there was a purely topological field
theory the exact solution was identical to that in the conventional equal-time approach
on the analogous spatial topology [21, 29, 45]. In the present context, the difficulty
is that the zero mode in A+ is in fact gauge-invariant, so that the light-cone gauge
A+ = 0 cannot be reached. Thus one has a pair of interconnected problems: first, a
practical choice of gauge; and second, the presence of constrained zero modes of the
gauge field. In several recent papers [25, 26, 41] these problems were separated and
consistent gauge fixing conditions were introduced to allow isolation of the dynamical
and constrained fields.
The study of these low dimensional theories is part of a long-term program to
attack QCD3 +1 through the zero mode sectors starting with studies of lower dimensional
theories which are themselves zero mode sectors of higher dimensional theories [25,
26J. A complete gauge fixing has recently been given for discrete light-cone quantized
QED 3 +l which further supports this program [27] and one sees how zero modes naturally
arise and the special role that they play. In appears that the central problem in light-
front QCD will be to disentangle and solve the constraints for the dependent zero modes
interms of the independent fields in the context of a particular gauge fixing.
Constrained Zero Modes
As mentioned previously, the light-front vacuum state is simple; it contains no

particles in a massive theory. However, one commonly associates important long range
properties of a field theory with the vacuum: spontaneous symmetry breaking, the
Goldstone pion, and color confinement. How do these complicated phenomena manifest
themselves in light-front field theory?
If one cannot associate long range phenomena with the vacuum state itself, then
the only alternative is the zero momentum components or "zero modes" of the field
(long range ~ zero momentum). In some cases, the zero mode operator is not an
independent degree of freedom but obeys a constraint equation. Consequently, it is
a complicated operator-valued function of all the other modes of the field [35]. This
problem has recently been attacked from several directions. The question of whether
boundary conditions can be consistently defined in light-front quantization has been
135
discussed by McCartor and Robertson [38] and Lenz [30,31]. They have shown that for
massive theories the energy and momentum derived from light-front quantization are
conserved and are equivalent to the energy and momentum one would normally write
down in an equal-time theory. Heinzl and Werner et al. [19, 18] considered q} theory
in (1+1)-dimensions and solved the zero mode constraint equation by truncating the
equation to one particle and retaining all modes. They implicitly retain a two particle
contribution in order to obtain finite results. Other authors [17] find that, for theories
allowing spontaneous symmetry breaking, there is a degeneracy of light-front vacua and
the true vacuum state can differ from the perturbative vacuum through the addition
of zero mode quanta. In addition to these approaches there are many others [6].
The definitive analysis by Pinsky, van de Sande, Bender and Hiller [5, 48, 22] of the
zero mode constraint equation for (1+1)-dimensional 4 field theory with symmetric
boundary conditions shows how spontaneous symmetry breaking occurs within the
context of this model. This theory has a Z2 symmetry -+ - which is spontaneously
broken for some values of the mass and coupling. Their approach is to apply a Tamm-
Dancoff truncation to the Fock space. Thus operators are finite matrices and the
operator valued constraint equation for the zero mode can be solved numerically. The
truncation assumes that states with a large number of particles or large momentum do
not have an important contribution to the zero mode.
One finds the following general behavior: for small coupling (large g, where
9 ex l/coupling) the constraint equation has a single solution and the field has no
vacuum expectation value (VEV). As one increase the coupling (decrease g) to the
"critical coupling" gcriticai> two additional solutions which give the field a nonzero VEV
appear. These solutions differ only infinitesimally from the first solution near the crit-
ical coupling, indicating the presence of a second order phase transition. Above the
critical coupling (g < gcritical), there are three solutions: one with zero VEV, the "un-
broken phase," and two with nonzero VEV, the "broken phase" . The "critical curves"
shown in Figure 1, is a plot the VEV as a function of g.
Since the vacuum in this theory is trivial, all of the long range properties must occur
in the operator structure of the Hamiltonian. Above the critical coupling (g < gcritical)
quantum oscillations spontaneously break the Z2 symmetry of the theory. In a loose
analogy with a symmetric double well potential, one has two new Hamiltonians for the
broken phase, each producing states localized in one of the wells. The structure of
the two Hamiltonians is determined from the broken phase solutions of the zero mode
constraint equation. One finds that the two Hamiltonians have equivalent spectra.
In a discrete theory without zero modes it is well known that, if one increases the
coupling sufficiently, quantum correction will generate tachyons causing the theory to
break down near the critical coupling. Here the zero mode generates new interactions
that prevent tachyons from developing. In effect what happens is that, while quantum
corrections attempt to drive the mass negative, they also change the vacuum energy
through the zero mode and the diving mass eigenvalue can never catch the vacuum
eigenvalue. Thus, tachyons never appear in the spectra.
In the weak coupling limit (g large) the solution to the constraint equation can
be obtained in perturbation theory. This solution does not break the Z2 symmetry
and is believed to simply insert the missing zero momentum contributions into internal
propagators. This must happen if light-front perturbation theory is to agree with
equal-time perturbation theory.
Another way to investigate the zero mode is to study the spectrum of the field
operator . Here one finds a picture that agrees with the symmetric double well po-
tential analogy. In the broken phase, the field is localized in one of the minima of the
potential and there is tunneling to the other minimum.
136
1.5~----------~--------~--~--~~
-1.5L-~~--~~~~~~--~--~~~~
0.2 0.4 Q.6 0.8 1 1.2 1.4
CJ
Figure 1: /0 = V41r(OI4>IO) vs. 9 = 24rrJ.L2 j>' in the one mode case with N = 10.
Canonical Quantization
The details of the Dirac-Bergmann prescription and its application to the system
considered here are discussed elsewhere in the literature [35, 49]. For a classical field
the (4)4h+l Lagrange density is
J.L2 2 >. 4
C = 8+4>8_4> - -4> - 4,4> .
( )
1
2 .
One puts the system in a box of length d and impose periodic boundary conditions.
Then
4>(x) = ~~ qn(X+)eiktx - , (2)
where k;; = 2rrnjd and summations run over all integers unless otherwise noted.
The J dx- 4>(x)n minus the zero mode part is
1
En = ,
n.
L
il,i2I"o.,i n ,-fO
% qi2 ... qi" Dil +i2+ ...+i",0 (3)
and the canonical Hamiltonian is
p- _ J.L2q'5 2E >'q~ >.q'5E2 >.qOE3 >.E4

- 2 + J.L 2 + 4!d + 2!d + d + d . (4)
Following the Dirac-Bergmann prescription, one identify first-class constraints which

define the conjugate momenta
0= Pn - ik;;q_n , (5)
where
m,n =f. O. (6)
137
The secondary constraint is,
(7)
which determines the zero mode qQ. This result can also be obtained by integrating the
equations of motion.
To quantize the system one replaces the classical fields with the corresponding
field operators, and the Dirac bracket by i times a commutator. One must choose
a regularization and an operator-ordering prescription in order to make the system
well-defined.
One begin by defining creation and annihilation operators at
and ak,
(8)
which satisfy the usual commutation relations
[ak,atl = 0, [aLat] = 0, [ak' at] = Ok,l, k,l > O. (9)
Likewise, one defines the zero mode operator
qQ = {;f; aD (10)
In the quantum case, one normal orders En.

General arguments suggest that the Hamiltonian should be symmetric ordered [4].
However, it is not clear how one should treat the zero mode since it is not a dynamical
field. As an ansatz one treats aD as an ordinary field operator when symmetric ordering
the Hamiltonian. The tadpoles are removed from the symmetric ordered Hamiltonian
by normal ordering the terms having no zero mode factors and by subtracting,
(11)
In addition, one subtract a constant so that the VEV of H is zero. Note that this
renormalization prescription is equivalent to a conventional mass renormalization and
does not introduce any new operators into the Hamiltonian. The constraint equation
for the zero mode can be obtained by taking a derivative of P- with respect to aD. One
finds,
where 9 = 247r /L 2 / A. It is clear from the general structure of (12) that aD as a function
of the other modes is not necessarily odd under the transform ak -+ -ak, k # 0
associated with the Z2 symmetry of the system. Consequently, the zero mode can
induce Z2 symmetry breaking in the Hamiltonian.
In order to render the problem tractable, one can impose a Tamm-Dancoff trun-
cation on the Fock space. Define M to be the number of nonzero modes and N to
be the maximum number of allowed particles. Thus, each state in the truncated Fock
space can be represented by a vector of length S = (M + N)!/ (M!N!) and operators
can be represented by S x S matrices. One can define the usual Fock space basis,
138
Inl' n2,, nM). where nl + n2 + ... + nM :S N . In matrix form, ao is real and
symmetric. Moreover, it is block diagonal in states of equal P+ eigenvalue.
Perturbative Solution of the Constraints
In the limit of large g, one can solve the constraint equation perturbatively .
Then one substitutes the solution back into the Hamiltonian and calculates various
amplitudes to arbitrary order in 1/g using Hamiltonian perturbation theory.
It can be shown that the solutions of the constraint equation and the resulting
Hamiltonian are divergence free to all orders in perturbation theory for both the broken
and unbroken phases. The perturbative solution for the unbroken phase is
(13)
Substituting this into the Hamiltonian, one obtains a complicated but well defined
expression.
The finite volume box acts as an infra-red regulator and the only possible diver-
gences are ultra-violet. Using diagrammatic language, any loop of momentum k with
e internal lines has asymptotic form k- i . Only the case of tadpoles e = 1 is divergent.
If there are multiple loops, the effect is to put factors of In(k) in the numerator and
the divergence structure is unchanged. Looking at Eq. (13), the only possible tadpole
is from the contraction in the term
(14)
which is canceled by the L,3/k term. This happens to all orders in perturbation theory:
each tadpole has an associated term which cancels it. Likewise, in the Hamiltonian one
o
no zero mode
-0.5
. . Maeno's ordering
symmetric ordering
our ordering
large d limit
~
.....-i...,
;l
-1

0.
!il-l.S

-2
-2.5~__~~=;===;~==;=~~==~=J
o 2.5 5 7.5 10 12.5 15
d
Figure 2: Convergence to the large d limit of 1 --> 1 setting E = g/p and dropping any
constant terms.
139
has similar cancellation. As with the zero mode, such cancellations occur to all orders
in perturbation theory.
For the unbroken phase, the effect of the zero mode should vanish in the infinite
volume limit, giving a "measure zero" contribution to the continuum Hamiltonian.
However, for finite box volume the zero mode does contribute, compensating for the
fact that the longest wavelength mode has been removed from the system. Thus,
inclusion of the zero mode improves convergence to the infinite volume limit; it acts as
a form of infra-red renormalization. In addition, one can use the perturbative expansion
of the zero mode to study the operator ordering problem. One can directly compare
our operator ordering ansatz with a truly Weyl ordered Hamiltonian and with Maeno's
operator ordering ansatz [33J. As an example, let us examine 0(>.2) contributions to
the processes 1 -+ 1 which as shown in Figure 2. including the zero mode greatly
improves convergence to the large volume limit. The zero mode compensates, in an
optimal manner, for the fact that one has removed the longest wavelength mode from
the system.
Non-Perturbative Solution: One Mode, Many Particles
Consider the case of one mode M = 1 and many particles. In this case, the
zero-mode is diagonal and can be written as
N
ao = fo 10) (01 +L fk Ik) (kl . (15)
k=l
Note that ao in (15) is even under ak -+ -ak, k =f 0 and any non-zero solution breaks
the Z2 symmetry of the original Hamiltonian. The VEV is given by
1 1
(0110) = . ~(OlaoIO) = ~fo. (16)
V41f v41f
Substituting (15) into the constraint Eq.(12) and sandwiching the constraint equation
between Fock states, one get a recursion relation for {fn}:
0= gfn + fn 3 + (4n - l)fn + (n + 1) fn+l + nfn-l (17)
where n :S N, and one define fN+l to be unknown. Thus, {it, 12, , fN+d is uniquely
determined by a given choice of g and fo. In particular, if fo = 0 all the fk'S are zero
independent of g. This is the unbroken phase.
Consider the asymptotic behavior for large n. If fn 1 in this limit, then the fn 3
term will dominate and
(18)
thus,
lim fn'" (-ltexp(3n constant) (19)
n-->oo
One must reject this rapidly growing solution. Hence, one only seek solutions where fn
is small for large n. For large n, the terms linear in n dominate and Eq. (17) becomes
fn+l + 4fn + fn-l = 0 . (20)
There are two solutions to this equation:

(21)
140
One must reject the plus solution because it grows with n. This gives the condition
V3-3+g =K K=0,1,2 ... (22)

2V3 '
Concentrating on the K = 0 case, one finds a critical coupling
gcritical = 3- v'3 (23)

or
>'critical = 47r (3 + Va) P? ~ 60J.L2 (24)
In comparison, values of >'critical from 22f.L2 to 55f.L2 have been reported for equal-time
quantized calculations [10, 1, 14, 28]. The solution to the linearized equation is an
approximate solution to the full Eq. (17) for fa sufficiently small. Next, one needs to
determine solutions of the full nonlinear equation which converge for large n.
One can study the critical curves by looking for numerical solutions to Eq. (17).
The method used here is to find values of fa and 9 such that fN+l = O. Since one seeks
a solution where fn is decreasing with n, this is a good approximation. One finds that
for 9 > 3 - .J3 the only real solution is fn = 0 for all n. For 9 less than 3 - V3 there
are two additional solutions. Near the critical point Ifol is small and
(25)
The critical curves are shown in Figure 1. These solutions converge quite rapidly with
N. The critical curve for the broken phase is approximately parabolic in shape:
9 ~ 3- v'3 - 0.9177fg (26)
One can also study the eigenvalues of the Hamiltonian for the one mode case. The
Hamiltonian is diagonal for this Fock space truncation and,
(n IH I) f~ 2n + 1 2 n +1 2
n =
3 ()
2"n n - 1 + ng - 4 - -4-fn + -4-fn+1 + "4n fn-l
2
- C . (27)
The invariant mass eigenvalues are given by
p 2 ln) = 2P+ P-In) = n>.(nIHln) In) (28)

247r
In Figure 3 the dashed lines show the first few eigenvalues as a function of 9 without
the zero-mode. When one include the broken phase of the zero mode, the energy levels
shift as shown by the solid curves. For 9 < gcritical the energy levels increase above the
value they had without the zero mode. The higher levels change very because fn is
small for large n. In the more general case of many modes and many particles many of
the features that were seen in the one mode and one particle cases remain.
One can also investigate the shape of the critical curve near the critical coupling as
a function of the cutoff K. In scalar field theory, (OIIO) acts as the order parameter of
the theory. Near the critical coupling, one can fit the VEV to some power of g- gcritical;
this will give us the associated critical exponent fJ,
(OlaoIO) ex (gcritical - gt . (29)
They have calculated this as a function of cutoff and found a result consistent with
fJ = 1/2, independent of cutoff K. The theory (4)1+1 is in the same universality class
141
2
-- - ---- ---
....
--- ---
O~~;::::;:::;:::;::::;=;:::::;::;J
o 0.2 0.4 0.6 0.8 1 1.2
9
Figure 3: The lowest three energy eigenvalues for the one mode case as a function of
9 from the numerical solution of Eq. (27) with N = 10. The dashed lines are for the
unbroken phase fo = 0 and the solid lines are for the broken phase fo f O.
as the Ising model in 2 dimensions and the correct critical exponent for this universality
class is (J = 1/8. If one were to use the mean field approximation to calculate the critical
exponent, the result would be (J = 1/2. This is what was obtained in this calculation.
Usually, the presence of a mean field result indicates that one is not probing all length
scales properly. If one had a cutoff K large enough to include many length scales,
then the critical exponent should approach the correct value. However, one cannot
be certain that this is the correct explanation of our result since no evidence that (J
decreases with increase K is seen.
Spectrum of the Field Operator
How does the zero mode affect the field itself? Since is a Hermitian operator it
is an observable of the system and one can. measure for a given state la). i and IXi)
are the eigenvalue and eigenvector respectively of V4n:
(30)
The expectation value of V4n in the state la) is L:i i I(Xi la) 12.
In the limit of large N, the probability distribution becomes continuous. If one
ignores the zero mode, the probability of obtaining as the result of a measurement of
..;;r:;r for the vacuum state is
p () = J;7fr exp (- t:) d (31)
where r = L:~ll/k. The probability distribution comes from the ground state
wave function of the Harmonic oscillator where one identifies with the position op-
142
0.35
0.3
:>.
~ 0.25
rl
:iIt!:l 0.2
.g
~
0.15
0.. 0.1
0.05
0~--_~4~~~~~~L-~~~~--~4--~
Eigenvalue
Figure 4: Probability distribution of eigenvalues of ..j4; for the vacuum with M = 1,

N = 10, and no zero mode. Also shown is the infinite N limit from Eq. (31).
erator. This is just the Gaussian fluctuation of a free field. When N is finite, the
distribution becomes discrete as shown in Figure 4.
In general, there are N + 1 eigenvalues such that (Xi 10} -=f. 0, independent of M.
Thus if one wants to examine the spectrum of the field operator for the vacuum state,
it is better to choose Fock space truncations where N is large. With this in mind,
one examines the N = 50 and M = 1 case as a function of 9 in Figure 5. Note that
near the critical point, Figure 5a, the distribution is approximately equal to the free
field case shown in Figure 4. As one moves away from the critical point, Figures 5b-
d, the distribution becomes increasingly narrow with a peak located at the VEV of
what would be the minimum of the symmetric double well potential in the equal-time
paradigm. In addition, there is a small peak corresponding to minus the VEV. In the
language of the equal-time paradigm, there is tunneling between the two minima of
the potential. The spectrum of has been examined for other values of M and Nj the
results are consistent with the example discussed here.
Physical Picture and Classification of Zero Modes in Gauge Theories
When considering a gauge theory, there is a "zero mode" problem associated with
the choice of gauge in the compactified case. This subtlety, however, is not particular
to the light conej indeed, its occurrence is quite familiar in equal-time quantization on a
torus [34,39,29]. In the present context, the difficulty is that the zero mode in A+ is in
fact gauge-invariant, so that the light-cone gauge A+ = 0 cannot be reached. Thus one
has a pair of interconnected problems: first, a practical choice of gaugej and second, the
presence of constrained zero modes of the gauge field. In ref. [27] the generalize gauge
fixing in a discrete formalism is described by Kalloniatis and Robertson.
One defines, for a periodic quantity j, its longitudinal zero mode
(32)
and the corresponding normal mode part
(33)
The "global zero mode"-the mode independent of all the spatial coordinates is denoted
by (f):
143
0.3
0.2
0.25
:>. O.lS
-i-J
j 0.2
...;
rl rl
...;
rl rl
.0
{lO.lS
.2o 0.1 .0
o
H H
0.. 0..
0.1
o.os
o.os
o -7. S -S
.111
-2. S 0
II.
2.S S 7. S o -7. S -S
. 11111
-2. S 0
II
2.S S 7. S
Eigenvalue Eigenvalue
(a) (b)
0.4
0.4
g=-1.0 g=-2.0
0.3
0.3
:>.
-i-J
.r..;l
rl
.0
.2o 0.2
H
0..
0.1 0.1
o -7.5 -5
011111,,111
-2.5 0 2. S 5 7. S o -7.S -S
..111
-2.5
I II
0
.1 II.
2.5 5 7.5
Eigenvalue Eigenvalue
(c) (d)
Figure 5: Probability distribution of eigenvalues of ..j4ii for the vacuum with couplings
(a) g = 1, (b) g = 0, (c) g = -1, and (d) g = -2. M = 1, N = 50, and the positive
VEV solution to the constraint equation is used.
(34)
Finally, the quantity which will be of most interest to us is the "proper zero mode,"
defined by
fo == (f)o - (f) . (35)
By integrating over the appropriate direction(s) of space, one can project the
equations of motion onto the various sectors. The global zero mode sector requires
some special treatment, and will not be discussed here. Consider the proper zero mode
sector of the equations of motion
-alAo + = gJo+ (36)
(37)
144
(38)
One first observe that Eq. (36), the projection of Gauss' law, is a constraint which
determines the proper zero mode of A+ in terms of the current J+:
A o+ -_ -9 81
1 +
Jo . (39)
The equations 37 and 38 then determine the zero modes Ao - and AOi. Eq. (39) is
clearly incompatible with the strict light-cone gauge A+ = 0, which is most natural
in light-cone analyses of gauge theories. Here one encounter a common problem in
treating axial gauges on compact spaces. It is not possible to bring an arbitrary gauge
field configuration to one satisfying A+ = 0 via a gauge transformation, and the light-
cone gauge is incompatible with the chosen boundary conditions. The closest one can
come is to set the normal mode part of A+ to zero, which is equivalent to
(40)
This condition does not, however, completely fix the gauge-one is free to make ar-
bitrary x--independent gauge transformations without undoing Eq. (40). One may
therefore impose further conditions on Ai-' in the zero mode sector of the theory.
Acting on Eq. (38) with 8i . The transverse field AOi then drops out and one obtain
an expression for the time derivative of Ao +:
+ 1 .
8+Ao = 9 81 8Jo' . (41)
Inserting this back into Eq. (38) one then find, after some rearrangement,

(42)
Now the operator (8; - 8i 8j /8l) is nothing more than the projector of the two-
dimensional transverse part of the vector fields Ao i and Joi. No trace remains of the
longitudinal projection of the field (8i 8j /8l)Aoj in Eq. (42). This reflects precisely the
residual gauge freedom with respect to x- -independent transformations. To determine
the longitudinal part, an additional condition is required.
The general solution to Eq. (42) is
A oi -
- -9 8IJ,i
2 0 + 8i<P (+ )
X , Xl. , (43)
1.
where <P f!1ust be independent of x- but is otherwise arbitrary. Imposing a condition on,
say, 8;Ao' will uniquely determine <po In ref.[26], for example, the condition 8;Ao i = 0
was proposed as being particularly natural. This choice, taken with the other gauge
conditions has been called the "compactification gauge." In this case
1 .
<p = 9 (8l)28i JO' . (44)
Of course, other choices are also possible. For example, one can generalize Eq. (44)
to
(45)
145
with a a real parameter. Then the "generalized compactification gauge." condition
corresponding to this solution is
1 .
oiAo' = -g(1 - a) 010;10' . (46)
An arbitrary gauge field configuration BJ1. can be brought to one satisfying Eq. (46)
via the gauge function
(47)
This is somewhat unusual in that A(x ..tJ involves the sources as well as the initial
field configuration, but this is perfectly acceptable. More generally, cp can be any
(dimensionless) function of gauge invariants constructed from the fields in the theory,
including the currents J. For our purposes Eq. (46) suffices.
One now has relations defining the proper zero modes of Ai,
(48)
as well as Ao + Eq. (39). All that remains is to use the final constraint Eq. (37) to
determine Ao -. Using eqs. 41 and 46, one finds that Eq. (37) can be written as
olAo - = -gJo- - 2ag ;2

.L
0+0;10i . (49)
After using the equations of motion to express o+Joi in terms of the dynamical
fields at x+ = 0, this may be straightforwardly solved for Ao- by imerting the 01. In
what follows, however, one has no need of Ao -. It does not enter the Hamiltonian, for
example; as usual, it plays the role of a multiplier to Gauss' law Eq. (38), which one is
able to implement as an operator identity.
The extension of the present work to the case of QCD is complicated by the fact
that the constraint relations for the gluonic zero modes are nonlinear, as in the 4
theory. A perturbative solution of the constraints is of course still possible, but in this
case, since the effective coupling at the relevant (hadronic) scale is large, it is clearly
desirable to go beyond perturbation theory. In addition, because of the central role
played by gauge fixing in the present work, one may expect complications due to the
Gribov ambiguity[15], which prevents the selection of unique representatives on gauge
orbits in nonperturbative treatments of Yang-Mills theory. Preliminary step in this
direction on the pure glue theory in 2+ 1 dimensions is found in ref. [26]. There one
finds that some of the nonperturbative techniques used recently in 1+ 1 dimensions
[5, 42] can be applied.
Dynamical Zero Modes
Our concern in this section is with those zero modes that are true dynamical
independent fields. They can arise due to the boundary conditions in gauge theory
preventing one from fully implement the traditional light-cone gauge A+ = O. The
development of the understanding of this problem in DLCQ can be traced in Refs.
[36, 18,25,26]. It has its analogue in instant form approaches to gauge theory [34,21].
Consider the zero mode subsector of the pure glue theory in (1+1) dimension,
namely where only zero mode external sources excite only zero mode gluons. This is
146
not an approximation but rather a consistent solution, a sub-regime within the complete
theory. A similar framing of the problem lies behind the work of Luscher [32] and van
Baal [47] using the instant form Hamiltonian approach to pure glue gauge theory in
3+ 1 dimensions. The beauty of this reduction in the 1+ 1 dimensional theory is two-
fold. First, it yields a theory which is exactly soluble. This is useful given the dearth
of soluble models in field theory. Secondly, the zero mode theory represents a paring
down to the point where the front and instant forms are manifestly identical, which is
nice to know indeed.
Consider an SU(2) non-Abelian gauge theory in 1+1 dimensions with classical
sources coupled to the gluons. The Lagrangian density is
(50)
where Fp." = a"A" - a"Ap. - g[Ap., A,,]. With a finite interval in x- from -L to L, one
imposes periodic boundary conditions on all gauge potentials AI'-"
One cannot eliminate the zero mode of the gauge potential. The reason is evident:
it is invariant under periodic gauge transformations. But of course one can always
perform a rotation in color space. In line with other authors [3, 44, 13], one chooses
o
this so that At is the only non-zero element, since in our representation only (13 is
o
diagonal. In addition, one can impose the subsidiary gauge condition A3"= 0 The reason
is that there still remains freedom to perform gauge transformations that depend only
on light-cone time x+ and the color matrix (13.
The above procedure would appear to have enabled complete fixing of the gauge.
This is still not so. Gauge transformations
(51)
generate shifts, according to Eq.(47), in the zero mode component
(52)
All of these possibilities, labeled by the integer n, of course still satisfy o_A+ = 0, but as
one sees n = 0 should not really be included. One can verify that the transformations
V also preserve the subsidiary condition. One notes that the transformation is x--
dependent and Z2 periodic. It is thus a simple example of a Gribov copy [15] in 1+1
dimensions. Following the conventional procedure one demands
A+...J. mr
3 -r gL' n= 1 , 2 , .... (53)
This eliminates singular points at the Gribov 'horizons' which in turn correspond to a
vanishing Faddeev-Popov determinant [47].
For convenience we henceforth use the notation
o 0
o J+J+ o B
At= v, x+ = t, w2 = ---.......=.
g2
and J:;="2' (54)
The only conjugate momentum is

o 0
p == 113" = a-At = a-v. (55)
147
o
The Hamiltonian density T+- = {)- At ITa - leads to the Hamiltonian
(56)
Quantization is achieved by imposing a commutation relation at equal light-cone

time on the dynamical degree of freedom. Introducing the variable q = 2Lv, the
appropriate commutation relation is [q(x+), p(x+)J = i. The field theoretic problem
reduces to quantum mechanics of a single particle as in Manton's treatment of the
Schwinger model in Ref.[34J. One thus has to solve the Schr6dinger equation
~(_~ (2Lw)2 Bq).I. = .1. (57)

2 dq2 + q2 + 2L If-' If-',
with the eigenvalue = E / (2L) actually being an energy density.

All eigenstates 't/J have the quantum numbers of the naive vacuum adopted in
standard front form field theory: all of them are eigenstates of the light-cone momentum
operator P+ with zero eigenvalue. The true vacuum is now that state with lowest p-
o
eigenvalue. In order to get an exactly soluble system one eliminates the source 2B =J;;.
The boundary condition that is to be imposed comes from the treatment of the
Gribov problem. Since the wave function vanishes at q = 0 one must demand that
the wavefunctions vanish at the first Gribov horizon q = 2rr /g. The overall constant
R is then fixed by normalization. This leads to the energy density only assuming the
discrete values
2
c(v) = L(X(V)2
"m 8rr 2 m , m = 1, 2, ... , (58)
where X~) denotes the m-th zero of the v-th Bessel function J v . In general, these
zeroes can only be obtained numerically. Thus
(59)
is the complete solution. The true vacuum is the state of lowest energy namely with
m=l.
The exact solution is genuinely non-perturbative in character. It describes vacuum-
like states since for all of these states P+ = O. Consequently, they all have zero invariant
mass M2 = P+ P-. The states are labeled by the eigenvalues of the operator P-. The
linear dependence on L in the result for the discrete energy levels is also consistent with
what one would expect from a loop of color flux running around the cylinder.
In the source-free equal time case Hetrick [21J uses a wave function that is sym-
metric about q = O. For our problem this corresponds to
't/Jm(q) = N cos( ..j2fmq) . (60)
where N is fixed by normalization. At the boundary of the fundamental modular region

q = 2rr/g and 't/Jm = (-l)mN, thus ..j2fm2rr/g = mrr and
g2(m 2 - 1)
f = =--'--,-----'- (61)
8
Note that m = 1 is the lowest energy state and has as expected one node in the allowed
region 0 ::; 9 ::; 2rr / g. Hetrick [21J discusses the connection to the results of Rajeev [45J
but it amounts to a shift in f and a redefining of m --> m/2. It has been argued by van
Baal that the correct boundary condition at q = 0 is 't/J(O) = O. This would give a sine
148
which matches smoothly with the Bessel function solution. This calculation offers the
lesson that even in a front form approach, the vacuum might not be just the simple
Fock vacuum. Dynamical zero modes do imbue the vacuum with a rich structure.
Acknowledgments
It is a pleasure to thank the organizers of ORBIS SCIENTIAE 1996 for providing

such a stimulating and enjoyable atmosphere. This work was supported in part by a
grant from the the US Department of Energy.
*
References
[1] J. Abad, J. G. Esteve, and A. F. Pacheco, Phys.Rev. D 32, 2729 (1985).
[2] J.L. Anderson and P.G. Bergmann, Physical Review, 83, 1018 (1951).
[3] A. M. Annenkova, E. V. Prokhvatilov, and V. A.Franke,Zielona Gora Pedag. Univ.

preprint - WSP-IF 89-01 (1990).
[4] C. M. Bender, L.R Mead, and S. S. Pinsky, Phys.Rev.Lett. 56, 2445 (1986).
[5] C.M. Bender, S. Pinsky, B. Van de Sande, Phys. Rev. D48 (1993) 816.
[6] M. Burkardt, Phys. Rev. D 47, 4628 (1993).
[7] M. Burkardt, Phys. Rev. D49, 5446 (1994).
[8] Carlitz, R, D. Heckathorn, J. Kaur and W.-K. Tung, Phys. Rev. D 11, 1234
(1975).
[9] Carlitz, R, and W.-K. Tung, Phys. Rev.D13, 3446 (1976).
[10] S. J. Chang, Phys. Rev. D 13,2778 (1976).

[11] De Alwis, S.P., and J. Stern, 1974, Nucl. Phys. B 77, 509.
[12] P.A.M. Dirac, Rev. Mod. Phys. 21,392 (1949).
[13] V. A. Franke, Y. V. Novozhilov, and E.V.Prokhvatilov, Lett. Math. Phys. 5, 239;

437 (1981)
[14] M. Funke, V. Kaulfass, and H. Kummel, Phys. Rev. D 35, 621 (1987);
[15] V.N. Gribov, Nucl. Phys. B139 1 (1978).
[16] A. Harindranath, RJ. Perry and J. Shigemitsu, Bound state problem in light-front
Tamm-Dancoff: A numerical studyin 1+1 dimensions, Ohio State preprint
[17] A. Harindranath and J. P. Vary, Phys. Rev. D36,1141 (1987).
[18] T. Heinzl, S. Krusche and E. Werner, Phys. Lett. B256 (1991) 55.
[19] T. Heinzl, S. Krusche, S. Simburger, and E. Werner, Z.Phys. C 56, 415 (1992)
149
[20] Heinzl, S. Krusche, and E. Werner, Phys. Lett. B 275, 410 (1992);
[21] J.E. Hetrick, Nucl.Phys. B30 228 (1993) .
[22] J. Hiller, S.S. Pinsky and B. van de Sande, Phys.Rev. D51 726 (1995).
[23] K. Hornbostel, Phys. Rev. D 45, 3781 (1992).
[24] Ida, M., Progr. Theor. Phys. 54, 1199 (1975).
[25] A.C. Kalloniatis, and H.C. Pauli, Z. Phys. 63 161 (1993)
[26] A. C. Kalloniatis, H. C. Pauli, and S. S. Pinsky, Phys. Rev. D52 1176 (1995).
[27] A. Kalloniatis and D. Robertson Phys. Rev. D50 5262 (1994)
[28] H. Kroger, R. Girard, and C. Dufour, Phys. Rev. D 35, 3944 (1987).
[29] E. Langmann and G. W. Semenoff, Phys. Lett.B296 117 (1992).

[30] F. Lenz, in the proceedings of the NATO Advanced Summer Institute on Nonper-
turbative Quantum Field Theory, Cargese, France,Aug 8-18, 1989. Edited by D.
Vautherin, F. Lenz, and J.W.Negele. Plenum Press, N. Y. (1990).
[31] F. Lenz, M. Thies, S. Levit, and K. Yazaki, Ann. Phys.20B, 1 (1991).

[32] M. Luscher, Nucl.Phys. B219 233(1983).
[33] M. Maeno, Phys. Lett. B, 83 (1994).
[34] N.S. Manton, Ann.Phys.(N.Y.) 159220 (1985) .
[35] T. Maskawa, and K. Yamawaki, Prog. Theor. Phys. 56,270 (1976).
[36] McCartor, G., Z. Phys. C 41, 271 (1988).
[37] G. McCartor, Z. Phys. C52 ,611 (1991).

[38] C. McCartor and D.C. Robertson, Z. Phys. C53 (1992) 679.
[39] J.A. Minahan and A.P. Polychronakos, Phys.Lett. B326288 (1994).
[40] H.C. Pauli, Nucl. Phys. A560 (1993) 50l.
[41] S. S. Pinsky and A.Kalloniatis, Phys. Lett. B365 225 (1996).
[42] S. S. Pinsky, and B. van de Sande, Phys. Rev. D 49, 2001 (1994).
[43] S. Pinsky "Topology and Confinement in Light-Front QCD" The Proceedings
of the Fourth Internationa Workshop on Light-Front Quantization and Non-
Perturbative Dynamics. 14-25 August i994, World Scientific S. D. Clazek ed.
[44] E. V. Prokhvatilov and V. A. Franke, Sov. J.Nucl Phys. 49, 688 (1989)
[45] S.C. Rajeev, University of Rochester preprint UR-1283 (1992); S.G. Rajeev PLB
212 203 (1988)
[46] D. Robertson, Phys. Rev. D 47, 2549 (1993).
150
[47] P. van Baal, Nucl.Phys. B369 259 (1992).
[48] B. van de Sande and S.S. Pinsky, Phys.Rev.D49 2001 (1994)
[49] R. S. Wittman, in Nuclear and Particle Physics on the Light Cone, edited by M.
B. Johnson and L. S. Kisslinger (WorldScientific, Singapore, 1989).
[50] K G. Wilson, T. S. Walhout, A. Harindranath, W.-M.Zhang,R. J. Perry, St. D.
Glazek, Phys. Rev.D49, 6720 (1994).
151
LIGHT-CONE QUANTIZATION AND HADRON STRUCTURE *
Stanley J. Brodsky
. Stanford Linear Accelerator Center

Stanford University, P. O. Box 4349
Stanford, California 94309
INTRODUCTION
Quantum chromodynamics provides a fundamental description of hadronic and

nuclear structure and dynamics in terms of elementary quark and gluon degrees of
freedom. In practice, the direct application of QCD to reactions involving the structure
of hadrons is extremely complex because of the interplay of nonperturbative effects such
as color confinement and multi-quark coherence.
In this talk, I will discuss light-cone quantization and the light-cone Fock expan-
sion as a tractable and consistent representation of relativistic many-body systems and
bound states in quantum field theory. The Fock state representation in QCD includes
all quantum fluctuations of the hadron wavefunction, including far off-shell configura-
tions such as intrinsic strangeness and charm and, in the case of nuclei, hidden color.
The Fock state components of the hadron with small transverse size, which dominate
hard exclusive reactions, have small color dipole moments and thus diminished hadronic
interactions. Thus QCD predicts minimal absorptive corrections, i.e., color transparen-
cy for quasi-elastic exclusive reactions in nuclear targets at large momentum transfer.
In other applications, such as the calculation of the axial, magnetic, and quadrupole
moments of light nuclei, the QCD relativistic Fock state description provides new in-
sights which go well beyond the usual assumptions of traditional hadronic and nuclear
physics.
QCD ON THE LIGHT CONE
The bound state structure of hadrons plays a critical role in virtually every area of
particle physics phenomenology. For example, in the case of the nucleon form factors,
*Work supported by Department of Energy contract DE-AC03-76SF00515.
153
pion electroproduction ep -+ e7r+n, exclusive B decays, and open charm photoproduc-
tion ,p -+ DA e , the cross sections depend not only on the nature of the quark currents,
but also on the coupling of the quarks to the initial and final hadronic states. Exclusive
decay amplitudes such as B -+ f{*" processes which will be studied intensively at B
factories, depend not only on the underlying weak transitions between the quark flavors,
but also the wavefunctions which describe how the Band f{* mesons are assembled
in terms of their fundamental quark and gluon constituents. Unlike the leading twist
structure functions measured in deep inelastic scattering, such exclusive channels are
sensitive to the structure of the hadrons at the amplitude level and to the coherence
between the contributions of the various quark currents and multi-parton amplitudes.
The analytic problem of describing QCD bound states is compounded not only by
the physics of confinement, but also by the fact that the wavefunction of a composite
of relativistic constituents has to describe systems of an arbitrary number of quanta
with arbitrary momenta and helicities. The conventional Fock state expansion based on
equal-time quantization quickly becomes intractable because of the complexity of the
vacuum in a relativistic quantum field theory. Furthermore, boosting such a wavefunc-
tion from the hadron's rest frame to a moving frame is as complex a problem as solving
the bound state problem itself. The Bethe-Salpeter bound state formalism, although
manifestly covariant, requires an infinite number of irreducible kernels to compute the
matrix element of the electromagnetic current even in the limit where one constituent
is heavy.
Light-cone quantization (LCQ) is formally similar to equal-time quantization (ETQ)
apart from the choice of initial-value surface. In ETQ one chooses a surface of constant
time in some Lorentz frame on which to specify initial values for the fields. In quantum
field theory this corresponds to specifying commutation relations among the fields at
some fixed time. The equations of motion, or the Heisenberg equations in the quantum
theory, are then used to evolve this initial data in time, filling out the solution at all
spacetime points.
In LCQ one chooses instead a hyperplane tangent to the light cone-properly called
a null plane or light front-as the initial-value surface. To be specific, we introduce LC
coordinates
(1)
(and analogously for all other four-vectors). The selection of the 3 direction in this
definition is of course arbitrary. In terms of LC coordinates, a contraction of four-
vectors decomposes as
(2)
from which we see that the momentum "conjugate" to x+ is p-. Thus the operator P-
plays the role of the Hamiltonian in this scheme, generating evolution in x+ according
to an equation of the form (in the Heisenberg picture)
[<p, P-] = 2i :~ . (3)
As was first shown by Diracl , seven of the ten Poincare generators become kine-
matical on the LC , the maximum number possible. The most important point is that
these include Lorentz boosts. Thus in the LC representation boosting states is trivial-
the generators are diagonal in the Fock representation so that computing the necessary
exponential is simple. One result of this is that the LC theory can be formulated in a
154
manifestly frame-independent way, yielding wavefunctions that depend only on momen-
tum fractions and which are valid in any Lorentz frame. This advantage is somewhat
compensated for, however, in that certain rotations become nontrivial in LCQ. Thus
rotational invariance will not be manifest in this approach.
Another advantage of going to the LC is even more striking: the vacuum state
seems to be much simpler in the LC representation than in ETQ. Note that the longi-
tudinal momentump+ is conserved in interactions. For particles, however, this quantity
is strictly positive,
1
p+ = (p~ + p~ + m2) 2 + l > 0 . (4)
Thus the Fock vacuum is the only state in the theory with p+ = 0, and so it must
be an exact eigenstate of the full interacting Hamiltonian. Stated more dramatically,
the Fock vacuum in the LC representation is the physical vacuum state. To the ex-
tent that this is really true, it represents a tremendous simplification, as attempts to
compute the spectrum and wavefunctions of some physical state are not complicated
by the need to recreate a ground state in which processes occur at unrelated locations
and energy scales. Furthermore, it immediately gives a constituent picture; all the
quanta in a hadron's wavefunction are directly connected to that hadron. This allows
a precise definition of the partonic content of hadrons and makes interpretation of the
LC wavefunctions unambiguous. It also raises the question, however, of whether LC
field theory can be equivalent in all respects to field theories quantized at equal times,
where nonperturbative effects often lead to nontrivial vacuum structure. In QCD, for
example, there is an infinity of possible vacua labelled by a continuous parameter 0,
and chiral symmetry is spontaneously broken. The question is how it is possible to
identify and incorporate such phenomena into a formalism in which the vacuum state
is apparently simple.
The description of relativistic composite systems using light-cone quantization l
thus appears to be remarkably simple. The Heisenberg problem for QCD can be written
in the form
(5)
where H LC = p+ p- - Pi is the mass operator. The operator P- = po - p 3 is the gen-
erator of translations in the light-cone time x+ = x O + x3. The quantities p+ = po + p3
and Plo play the role of the conserved three-momentum. Each hadronic eigenstate IH)
of the QCD light-cone Hamiltonian can be expanded on the complete set of eigen-
states {In)} of the free Hamiltonian which have the same global quantum numbers:
IH) = L 1f'1~ (Xi, kloi' Ai)ln). In the case of the proton, the Fock expansion begins with
the color singlet state luud) of free quarks, and continues with luudg) and the other
quark and gluon states that span the degrees of freedom of the proton in QCD. The
Fock states {In)} are built on the free vacuum by applying the free light-cone creation
operators. The summation is over all momenta (Xi, kloi) and helicities Ai satisfying mo-
mentum conservation L:i Xi = 1 and L:i kloi = 0 and conservation of the projection J3
of angular momentum.
The wavefunction 1f'1~(Xi' kloi' Ai) describes the probability amplitude that a proton
of momentum p+ = po + p3 and transverse momentum Pl. consists of n quarks and
gluons with helicities Ai and physical momenta pi = XiP+ and Pl.i = XiPl. + kl. i .
The wavefunctions {1f'1~( Xi, kl.i' Ai)}, n = 3, ... thus describe the proton in an arbitrary
moving frame. The variables (Xi, kloi) are internal relative momentum coordinates.
The fractions Xi = pt IP+ = (p? + p7)/(P O + P 3), 0 < Xi < 1, are the boost-invariant
light-cone momentum fractions; Yi = log Xi is the difference between the rapidity of
the constituent i and the rapidity of the parent hadron. The appearance of relative
155
coordinates is connected to the simplicity of performing Lorentz boosts in the light-
cone framework. This is another major advantage of the light-cone representation.
The spectra of hadrons and nuclei as well as their scattering states can be identified
with the set of eigenvalues of the light-cone Hamiltonian Hw for QCD. Particle number
is generally not conserved in a relativistic quantum field theory, so that each eigenstate
is represented as a sum over Fock states of arbitrary particle number. Thus in QCD
each hadron is expanded as second-quantized sums over fluctuations of color-singlet
quark and gluon states of different momenta and number. The coefficients of these
fluctuations are the light-cone wavefunctions tPn(Xi, k1.i' Ai). The invariant mass M of
the partons in a given n-particle Fock state can be written in the elegant form
M2 = Ln k2 +m2
1.i (6)
i=1 Xi
The dominant configurations in the wavefunction are generally those with minimum
values of M2. Note that, except for the case where mi = 0 and k1.i = 0, the limit
Xi - t 0 is an ultraviolet limit, i.e., it corresponds to particles moving with infinite
momentum in the negative z direction: kt - t -k? - t -00. The light-cone wavefunctions
encode the properties of the hadronic wavefunctions in terms of their quark and gluon
degrees of freedom, and thus all hadronic properties can be derived from them. The
natural gauge for light-cone Hamiltonian theories is the light-cone gauge A+ = O. In
this physical gauge the gluons have only two physical transverse degrees of freedom,
and thus it is well matched to perturbative QCD calculations.
Since QCD is a relativistic quantum field theory, determining the wavefunction of a
hadron is an extraordinarily complex nonperturbative relativistic many-body problem.
In principle it is possible to compute the light-cone wavefunctions by diagonalizing the
QCD light-cone Hamiltonian on the free Hamiltonian basis. In the case of QCD in one
space and one time dimensions, the application of discretized light-cone quantization
(DLCQ)2 provides complete solutions of the theory, including the entire spectrum of
mesons, baryons, and nuclei, and their wavefunctions 3, 4. In the DLCQ method, one
simply diagonalizes the light-cone Hamiltonian for QCD on a discretized Fock state
basis. The DLCQ solutions can be obtained for arbitrary parameters including the
number of flavors and colors and quark masses. More recently, DLCQ has been applied
to new variants of QCD1+! with quarks in the adjoint representation, thus obtaining
color-singlet eigenstates analogous to gluonium states 5
The extension of this program to physical theories in 3+ 1 dimensions is a formidable
computational task because of the much larger number of degrees of freedom; howev-
er, progress is being made. Analyses of the spectrum and light-cone wavefunctions of
positronium in QED3+! are given elsewhere6 Hiller, Okamoto and 17 have been pursu-
ing a nonperturbative calculation of the lepton anomalous moment in QED using the
DLCQ method. Burkardt has recently solved scalar theories with transverse dimension-
s by combining a Monte Carlo lattice method with DLCQ8. Also of interest is recent
work of Hollenberg and Witte9 , who have shown how Lanczos tri-diagonalization can be
combined with a plaquette expansion to obtain an analytic extrapolation of a physical
system to infinite volume.
There has also been considerable work on the truncations required to reduce the
space of states to a manageable level lO, 11, 12. The natural language for this discussion
is that of the renormalization group, with the goal being to understand the kinds of
effective interactions that occur when states are removed, either by cutoffs of some
kind or by an explicit Tamm-Dancoff truncation. Solutions of the resulting effective
156
Hamiltonians can then be obtained by various means, for example using DLCQ or
basis function techniques. Some calculations of the spectrum of heavy quarkonia in
this approach have recently been reported 13
One of the remarkable simplicities of the LC formalism is the fact that one can write
down exact expressions for the spacelike electromagnetic form factors (P + QIJ+IP) of
any hadrons for any initial or final state helicity. At a fixed light-cone time, the exact
Heisenberg current can be identified with the free current j+. It is convenient to choose
the frame in which q+ = 0 so that qi is Q2 = -q~. Since the quark current j+ has
simple matrix elements between free Fock states, each form factor for a given helicity
transition A -+ ).' can be evaluated from simple overlap integrals of the light-cone
wavefuncti ons 14, 15:
F)..I,)..(Q2) = L JIT d2kLi JIT dXi1f)n,)..I(Xi, k~i' Ai)1/Jn,)..(Xi, kLi' Ai) ,

n
(7)
where the integrations are over the unconstrained relative coordinates. The internal
transverse momenta of the final state wavefunction are k~ = kL + (1 - x )qL for the
struck quark and k~ = kL - xqL for the spectator quarks. Thus given the light-
cone wavefunctions {1/Jn(Xi, k L;> Ai)} one can compute the electromagnetic and weak
form factors from a simple overlap of light-cone wavefunctions, summed over all Fock
states 14, 15. For spacelike momentum transfer only diagonal matrix elements in particle
number n' = n are needed. In contrast, in the equal-time theory one must also consider
off-diagonal matrix elements and fluctuations due to particle creation and annihilation
in the vacuum. In the nonrelativistic limit one can make contact with the usual formulae
for form factors in Schrodinger many-body theory.
The structure functions of a hadron can be computed from the square integral of
its LC wavefunctions 16 . For example, the quark distribution measured in deep inelastic
scattering at a given resolution Q2 is
(8)
where the struck quark is evaluated with its light-cone fraction equal to the Bjorken
variable: Xq = XBj = Q2/2p . q. A summation over all contributing Fock states is
required to evaluate the form factors and structure functions. Thus the hadron and
nuclear structure functions are the probability distributions constructed from integrals
over the absolute squares l1/JnI 2 , summed over n. In the far off-shell domain of large
parton virtuality, one can use perturbative QCD to derive the asymptotic fall-off of
the Fock amplitudes, which then in turn leads to the QCD evolution equations for
distribution amplitudes and structure functions. More generally, one can prove fac-
torization theorems for exclusive and inclusive reactions which separate the hard and
soft momentum transfer regimes, thus obtaining rigorous predictions for the leading
power behavior contributions to large momentum transfer cross sections. One can also
compute the far off-shell amplitudes within the light-cone wavefunctions where heavy
quark pairs appear in the Fock states. Such states persist over a time T ~ P+ / M2 until
they are materialized in the hadron collisions. As we shall discuss below, this leads to
a number of novel effects in the hadroproduction of heavy quark hadronic states 17.
Although we are still far from solving QCD explicitly, a number of properties
of the light-cone wavefunctions of the hadrons are known from both phenomenology
and the basic properties of QCD. For example, the endpoint behavior of light-cone
wavefunctions and structure functions can be determined from perturbative arguments
157
and Regge arguments. Applications are presented elsewhere 18 . There are also corre-
spondence principles. For example, for heavy quarks in the nonrelativistic limit, the
light-cone formalism reduces to conventional many-body Schrodinger theory. On the
other hand, one can also build effective three-quark models which encode the static
properties of relativistic baryons.
SOLVING NONPERTURBATIVE QUANTUM FIELD THEORY USING

LCQ
A large number of studies have been performed of model field theories in the LC
framework. This approach has been remarkably successful in a range of toy models in
1+ 1 dimensions: Yukawa theory19, the Schwinger model (for both massless and massive
fermions )20, 21, rjJ4 theory22, QCD with various types of matter3, 4, 5, 23, 24, and the sine-
Gordon model 25 It has also been applied with promising results to theories in 3+ 1
dimensions, in particular QED 6 and Yukawa theory26. In all cases agreement was found
between the LC calculations and results obtained by more conventional approaches, for
example, lattice gauge theory. In many cases the physics of spontaneous symmetry
breaking and vacuum structure of the equal-time theory is represented by the physics
of zero modes in LCQ27.
QCD 1 +1 with Fundamental Matter
This theory was originally considered by 't Hooft in the limit of large Nc 28. Later
Burkardt 3, and Hornbostel, Pauli and 1\ gave essentially complete numerical solutions
of the theory for finite N c , obtaining the spectra of baryons, mesons, and nucleons and
their wavefunctions. The results are consistent with the few other calculations available
for comparison, and are generally much more efficiently obtained. In particular, the
mass of the lowest meson agrees to within numerical accuracy with lattice Hamiltonian
results 29 . For Nc = 4 this mass is close to that obtained by 't Hooft in the Nc ----t 00
limit 28 . Finally, the ratio of baryon to meson mass as a function of Nc agrees with the
strong-coupling results of Date, Frishman and Sonnenschein30 .
In addition to the spectrum, of course, one obtains the wavefunctions. These
allow direct computation of, e.g., structure functions. As an example, Fig. 1 shows the
valence contribution to the structure function for an SU(3) baryon, for two values of the
dimensionless coupling m/ g. As expected, for weak coupling the distribution is peaked
near x = 1/3, reflecting that the baryon momentum is shared essentially equally among
its constituents. For comparison, the contributions from Fock states with one and two
additional qq pairs are shown in Fig. 2. Note that the amplitudes for these higher Fock
components are quite small relative to the valence configuration. The lightest hadrons
are nearly always dominated by the valence Fock state in these super-renormalizable
models; higher Fock wavefunctions are typically suppressed by factors of 100 or more.
Thus the light-cone quarks are much more like constituent quarks in these theories than
equal-time quarks would be. As discussed above, in an equal-time formulation even the
vacuum state would be an infinite superposition of Fock states. Identifying constituents
in this case, three of which could account for most of the structure of a baryon, would
be quite difficult.
158
SU(3) Baryon
1.5
mIg
o 0.1
......... 1.6
.c 1.0
~
+-~
.c
..........
0.5
0
0 0.2 0.4 0.6 0.8 1.0
1185 x= klK ......
Figure 1: Valence contribution to the baryon structure function in QCD1+l, as a func-
tion of the light-cone longitudinal momentum fraction. The gauge group is SU(3), m
is the quark mass, and 9 is the gauge coupling4.
I I I I I I I I
1.5 - -
3-9- SU(3) Baryon - .-, SU(3) Baryon
; \. O
; \
o mIg = 1 .6 (x 103) o mig = 1.6 (x 107)
mIg =0.1 (x 102) mig =0.1 (x 1(4) _
I \
I
2 jI '. - 1.0 -!
.0
""
.............
.
I
\
\
~
.c I
I
b
b -J:J.::t!
-.0""
""'"-"" i / .....
I \
I
"""-"'"
1
J
!.;..... ". "
- OS). .......\... -
/ ......::-::::.:~o.,. (a) . :! ".... (b)
". "0 f "O:::'.:..~~o,.,
1 1 .............. ""'0. J 1 I __ 1 ___
0e-----~----~--~~-4~-.-- __ O*---~-----L~~
',..1.. ~~- ~
o 0.2 0.4 0.6 0.8 1.0 o 0.2 0.4 0.6 0.8 1.0
11-95 x = kIK x= kIK 8084Al
Figure 2: Contributions to the baryon structure function from higher Fock components:
(a) valence plus one additional qq pair; (b) valence plus two additional qq pairs4.
159
Collinear QCD
QCD can be simplified in a dramatic way by eliminating all interactions which

involve nonzero transverse momentum. The trigluon interaction is eliminated but the
four-gluon and helicity flip qqg vertices still survive. In this simplified "reduced" or
"collinear" theory, one still has all of the degrees of freedom of QCD(3+ 1) including
transversely polarized color adjoint gluons, but the theory is effectively a one-space,
one-time theory which can be solved using discretized light-cone quantization. Recent-
ly Antonuccio and Dalley 24 have presented a comprehensive DLCQ analysis of collinear
QCD, obtaining the full physical spectrum of both quarkonium and gluonium states.
One also obtains the complete LC Fock wavefunctions for each state of the spectrum.
An important feature of this analysis is the restoration of complete rotational sym-
metry through the degeneracy of states of the rest frame angular momentum. In fact
as emphasized by Burkardt 31 , parity and rotational invariance can be restored if one
separately renormalizes the mass that appears in the helicity-flip qqg vertices and the
light-cone kinetic energy.
Antonuccio and Dalley 24 have also derived ladder relations which connect the end-
point Xq ---+ 0 behavior of Fock states with n gluons to the Fock state wavefunction with
n - 1 gluons, relations which follow most by imposing the condition that the k+ = 0
mode of the constraint equations vanishes on physical states. An important condition
for a bound state wavefunction is that gauge invariant quanta have should finite kinet-
ic energy in a bound state, just as the square of the "mechanical velocity" operator
iJ2 = (ji - eA')2 has finite expectation value in nonrelativistic electrodynamics. Such
a condition automatically connects Fock states of different particle number. Thus the
ladder relations should be generalizable to the full 3 + 1 theory by requiring that the
gauge-extended light-cone kinetic energy operator have finite expectation value.
EXCLUSIVE PROCESSES AND LIGHT-CONE QUANTIZATION
A central focus of future QCD studies will be hadron physics at the amplitude
level. Exclusive reactions such as pion electroproduction ,'p ---+ np are more subtle to
analyze than deep inelastic lepton scattering and other leading-twist inclusive reactions
since they require the consideration of coherent QCD effects. Nevertheless, there is an
extraordinary simplification: In any exclusive reaction where the hadrons are forced to
absorb large momentum transfer Q, one can isolate the nonperturbative long-distance
physics associated with hadron structure from the short-distance quark-gluon hard scat-
tering amplitudes responsible for the dynamical reaction. In essence, to leading order
in l/Q, each exclusive reaction AB ---+ CD factorizes in the form:
TAB -+ CD = Jo[1 IIdxicPD(Xi,

t t
Q)cPdXi, Q)cPA(Xi, Q)cPB(Xi, Q)Tquark , (9)
where cPA(Xi,Q) = Jki<Q2ITJ2k.LiITdxivalence(Xi,k.Li,'\;) is the process-independent

distribution amplitude-the light-cone wavefunction which describes the coupling of
hadron A to its valence quark with longitudinal light-cone momentum fractions 0 <
Xi < 1 at impact separation b = O(l/Q)-and Tquark is the amplitude describing the
hard scattering of the quarks collinear with the hadrons in the initial state to the
quarks which are collinear with the hadrons in the final state. Since the propagators
and loop momenta in the hard scattering amplitude Tquark are of order Q, it can be
computed perturbatively in QCD. The dimensional counting rules 32 for form factors
160
and fixed CM scattering angle processes follow from the nominal power-law fall off
of Tquark ' The scattering of the quarks all occurs at short distances; thus the hard
scattering amplitude only couples to the valence-quarks the hadrons when they are
at small relative impact parameter. Remarkably, there are no initial state or final
state interaction corrections to factorization to leading order in 1/ Q because of color
coherence; final state color interactions are suppressed. This feature not only insures
the validity of the factorization theorem for exclusive processes in QCD, but it also
leads to the novel effect of "color transparency" in quasi-elastic nuclear reactions 33 , 34.
An essential element of the factorization of high momentum transfer exclusive reac-
tions is universality, i.e., the distribution amplitudes <PA(Xi, Q) are unique wavefunctions
specific to each hadron 35 . The distribution amplitudes obey evolution equations and
renormalization group equations 16 which can be derived through the light-cone equa-
tions of motion or the operator product expansion. Thus the same wavefunction that
controls the meson form factors also controls the formation of the mesons in exclusive
decay amplitudes of B mesons such as B -+ 7r7r at the comparable momenta.
THE EFFECTIVE CHARGE av(Q2) AND LIGHT-CONE QUANTIZA-

TION
The heavy quark potential plays a central role in QCD, not only in determining
the spectrum and wavefunctions of heavy quarkonium, but also in providing a phys-
ical definition of the running coupling for QCD. The heavy quark potential V( Q2) is
defined as the two-particle irreducible amplitude controlling the scattering of two in-
finitely heavy test quarks QQ in an overall color-singlet state. Here Q2 = _q2 = if
is the momentum transfer. The effective charge av (Q2) is then defined through the
relation V(Q2) = -47rCFav(Q2)/Q2 where CF = (N; - 1)/2Nc = 4/3. The running
coupling av( Q2) satisfies the usual renormalization group equation, where the first t-
wo terms (30 and (31 in the perturbation series are universal coefficients independent of
the renormalization scheme or choice of effective charge. Thus av provides a physical
expansion parameter for perturbative expansions in PQCD.
By definition, all quark and gluon vacuum polarization contributions are summed
into av; the scale Q of av(Q2) that appears in perturbative expansions is thus fixed
by the requirement that no terms involving the QCD (3-function appear in the coef-
ficients. Thus expansions in av are identical to that of conformally invariant QCD.
This argument is the basis for BLM scale-fixing36 and commensurate scale relations 37 ,
which relate physical observables together without renormalization scale, renormaliza-
tion scheme, or other ambiguities arising from theoretical conventions.
There has recently been remarkable progress 38 in determining the running coupling
av( Q2) from heavy quark lattice gauge theory using as input a measured level splitting
in the T spectrum. The heavy quark potential can also be determined in a direct
way from experiment by measuring e+e- -+ cc and e+e- -+ bb at threshold 39 . The
cross section at threshold is strongly modified by the QCD Sommerfeld rescattering
of the heavy quarks through their Coulombic gluon interactions. The amplitude near
threshold is modified by a factor S((3, Q2) = x/(1 - exp( -x )), where x = CFav( Q2)/ (3
and (3 = /1 - 4m~) / s is the relative velocity between the produced quark and heavy
quark. The scale Q reflects the mean exchanged momentum transfer in the Coulomb
rescattering. For example, the angular distribution for e+ e- -+ QQ has the form
1 + A((3) cos 2 Bern. The anisotropy predicted in QCD for small (3 is then A = A/(1 + A),
161
where
A= {32 8({3,4m~{32/e) 1- :av(m~exp7/6)
(10)
2 8({3, 4m~{32) 1 - ~!av(m~ exp 3/4)'
The last factor is due to hard virtual radiative corrections. The anisotropy in e+e- ---+
QQ will be reflected in the angular distribution of the heavy mesons produced in the
corresponding exclusive channels.
The renormalization scheme corresponding to the choice of av as the coupling is
the natural one for analyzing QCD in the light-cone formalism, since it automatically
sums all vacuum polarization contributions into the coupling. For example, once one
knows the form of av( Q2), it can be used directly in the light-cone formalism as a means
to compute the wavefunctions and spectrum of heavy quark systems. The effects of the
light quarks and higher Fock state gluons that renormalize the coupling are already
contained in av.
The same coupling can also be used for computing the hard scattering amplitudes
that control large momentum transfer exclusive reactions and heavy hadron weak de-
cays. Thus when evaluating Tquark the scale appropriate' for each appearance of the
running coupling av is the momentum transfer of the corresponding exchanged glu-
on 40 This prescription agrees with the BLM procedure. The connection between av
and the usual a MS scheme is described elsewhere37
THE PHYSICS OF LIGHT-CONE FOCK STATES
The light-cone formalism provides the theoretical framework which allows for a
hadron to exist in various Fock configurations. For example, quarkonium states not only
have valence QQ components but they also contain QQg and QQgg states in which the
quark pair is in a color-octet configuration. Similarly, nuclear LC wave functions contain
components in which the quarks are not in color-singlet nucleon sub-clusters. In some
processes, such as large momentum transfer exclusive reactions, only the valence color-
singlet Fock state of the scattering hadrons with small inter-quark impact separation
b.L = O(l/Q) can couple to the hard scattering amplitude. In reactions in which large
numbers of particles are produced, the higher Fock components of the LC wavefunction
will be emphasized. The higher particle number Fock states of a hadron containing
heavy quarks can be diffractively excited, leading to heavy hadron production in the
high momentum fragmentation region of the projectile. In some cases the projectile's
valence quarks can coalesce with quarks produced in the collision, producing unusual
leading-particle correlations. Thus the multi-particle nature of the LC wavefunction
can manifest itself in a number of novel ways. For example:
Color Transparency
QCD predicts that the Fock components of a hadron with a small color dipole
moment can pass through nuclear matter without interactions33, 34. Thus in the case of
large momentum transfer reactions, where only small-size valence Fock state configura-
tions enter the hard scattering amplitude, both the initial and final state interactions
of the hadron states become negligible.
Color Transparency can be measured though the nuclear dependence of totally d-
iffractive vector meson production du/dt(,* A ---+ V A). For large photon virtualities (or
for heavy vector quarkonium), the small color dipole moment of the vector system im-
plies minimal absorption. Thus, remarkably, QCD predicts that the forward amplitude
162
,*A --+ V A at t --+ 0 is nearly linear in A. One is also sensitive to corrections from the
nonlinear A-dependence of the nearly forward matrix element that couples two gluons
to the nucleus, which is closely related to the nuclear dependence of the gluon structure
function of the nucleus 41 .
The integral of the diffractive cross section over the forward peak is thus predicted
,*
to scale approximately as A2 / R~ '" A 4 / 3 Evidence for color transparency in quasi-
elastic p leptoproduction A --+ pO N(A - 1) has recently been reported by the E665
experiment at Fermilab42 for both nuclear coherent and incoherent reactions. A test
could also be carried out at very small tmin at HERA, and would provide a striking test
of QCD in exclusive nuclear reactions. There is also evidence for QCD "color trans-
parency" in quasi-elastic pp scattering in nuclei 43 . In contrast to color transparency,
Fock states with large-scale color configurations interact strongly and with high particle
number production 44 .
Hidden Color
The deuteron form factor at high Q2 is sensitive to wavefunction configurations

where all six quarks overlap within an impact separation hi < O(l/Q); the leading
power-law fall off predicted by QCD is Fd(Q2) = f(a s (Q2))/(Q2)5, where, asymp-
totically, f(a s (Q2)) ex: a s (Q2)S+2'Y.45 The derivation of the evolution equation for the
deuteron distribution amplitude and its leading anomalous dimension I is given else-
where46 . In general, the six-quark wavefunction of a deuteron is a mixture of five
different color-singlet states. The dominant color configuration at large distances cor-
responds to the usual proton-neutron bound state. However at small impact space
separation, all five Fock color-singlet components eventually acquire equal weight, i.e.,
the deuteron wavefunction evolves to 80% "hidden color." The relatively large normal-
ization of the deuteron form factor observed at large Q2 points to sizable hidden color
contributions 47 .
Spin-Spin Correlations in Nucleon-Nucleon Scattering and the Charm Thresh-
old
One of the most striking anomalies in elastic proton-proton scattering is the large
spin correlation ANN observed at large angles 48 . At vis ~ 5 GeV, the rate for scattering
with incident proton spins parallel and normal to the scattering plane is four times
larger than that for scattering with anti-parallel polarization. This strong polarization
correlation can be attributed to the onset of charm production in the intermediate
state at this energy49. The intermediate state luuduudcc) has odd intrinsic parity
and couples to the J = S = 1 initial state, thus strongly enhancing scattering when
the incident projectile and target protons have their spins parallel and normal to the
scattering plane. The charm threshold can also explain the anomalous change in color
transparency observed at the same energy in quasi-elastic pp scattering. A crucial test
is the observation of open charm production near threshold with a cross section of order
of Ipb.
Anomalous Decays of the J / 'IjJ
The dominant two-body hadronic decay channel of the J / 'IjJ is J / 'IjJ --+ p7r, even
though such vector-pseudoscalar final states are forbidden in leading order by helicity
conservation in perturbative QCD 50 . The 'IjJ', on the other hand, appears to respect
163
PQCD. The J/1/J anomaly may signal mixing with vector gluonia or other exotica50 .
The QCD Van Der Waals Potential and Nuclear Bound Quarkonium
The simplest manifestation of the nuclear force is the interaction between two heavy
quarkonium states, such as the Y(bb) and the J/1/J(cc). Since there are no valence quarks
in common, the dominant color-singlet interaction arises simply from the exchange of
two or more gluons. In principle, one could measure the interactions of such systems by
producing pairs of quarkonia in high energy hadron collisions. The same fundamental
QCD van der Waals potential also dominates the interactions of heavy quarkonia with
ordinary hadrons and nuclei. The small size of the QQ bound state relative to the much
larger hadron allows a systematic expansion of the gluonic potential using the operator
product expansion51. The coupling of the scalar part of the interaction to large-size
hadrons is rigorously normalized to the mass of the state via the trace anomaly. This
scalar attractive potential dominates the interactions at low relative velocity. In this
way one establishes that the nuclear force between heavy quarkonia and ordinary nuclei
is attractive and sufficiently strong to produce nuclear-bound quarkonium 51 , 52.
Anomalous Quarkonium Production at the Tevatron
Strong discrepancies between conventional QCD predictions and experiment of a

factor of 30 or more have recently been observed for 1/J, 1/J', and Y production at large PT
in high energy pp collisions at the Tevatron 53 . Braaten and Fleming 54 have suggested
that the surplus of charmonium production is due to the enhanced fragmentation of
gluon jets coupling to the octet cc components in higher Fock states iccgg) of the
charmonium wavefunction. Such Fock states are required for a consistent treatment of
the radiative corrections to the hadronic decay of P-waves in QCD55.
INTRINSIC HEAVY QUARK CONTRIBUTIONS IN HADRONIC WAVE-

FUNCTIONS
It is important to distinguish two distinct types of quark and gluon contributions

to the nucleon sea measured in deep inelastic lepton-nucleon scattering: "extrinsic" and
"intrinsic" 56. The extrinsic sea quarks and gluons are created as part of the lepton-
scattering interaction and thus exist over a very short time ~T '" I/Q. These factor-
izable contributions can be systematically derived from the QCD hard bremsstrahlung
and pair-production (gluon-splitting) subprocesses characteristic of leading twist per-
turbative QCD evolution. In contrast, the intrinsic sea quarks and gluons are multi con-
nected to the valence quarks and exist over a relatively long lifetime within the nucleon
bound state. Thus the intrinsic qq pairs can arrange themselves together with the va-
lence quarks of the target nucleon into the most energetically-favored meson-baryon
fluctuations.
In conventional studies of the "sea" quark distributions, it is usually assumed that,
aside from the effects due to antisymmetrization, the quark and anti quark sea contri-
butions have the same momentum and helicity distributions. However, the ansatz of
identical quark and antiquark sea contributions has never been justified, either theo-
retically or empirically. Obviously the sea distributions which arise directly from gluon
splitting in leading twist are necessarily CP-invariant; i.e., they are symmetric under
quark and antiquark interchange. However, the initial distributions which provide the
164
boundary conditions for QCD evolution need not be symmetric since the nucleon state
is itself not CP-invariant. Only the global quantum numbers of the nucleon must be
conserved. The intrinsic sources of strange (and charm) quarks reflect the wavefunc-
tion structure of the bound state itself; accordingly, such distributions would not be
expected to be CP symmetric. Thus the strange/anti-strange asymmetry of nucleon
structure functions provides a direct window into the quantum bound-state structure
of hadronic wavefunctions.
It is also possible to consider the nucleon wavefunction at low resolution as a
fluctuating system coupling to intermediate hadronic Fock states such as non-interacting
meson-baryon pairs. The most important fluctuations are most likely to be those closest
to the energy shell and thus have minimal invariant mass. For example, the coupling of
a proton to a virtual J(+ A pair provides a specific source of intrinsic strange quarks and
anti quarks in the proton. Since the sand 5 quarks appear in different configurations
in the lowest-lying hadronic pair states, their helicity and momentum distributions are
distinct.
Recently Bo-Qiang Ma and I have investigated the quark and anti quark asymmetry
in the nucleon sea which is implied by a light-cone meson-baryon fluctuation model
of intrinsic qq pairs 57. We utilize a boost-invariant light-cone Fock state description
of the hadron wavefunction which emphasizes multi-parton configurations of minimal
invariant mass. We find that such fluctuations predict a striking sea quark and antiquark
asymmetry in the corresponding momentum and helicity distributions in the nucleon
structure functions. In particular, the strange and anti-strange distributions in the
nucleon generally have completely different momentum and spin characteristics. For
example, the model predicts that the intrinsic d and s quarks in the proton sea are
negatively polarized, whereas the intrinsic d and 5 anti quarks provide zero contributions
to the proton spin. We also predict that the intrinsic charm and anticharm helicity and
momentum distributions are not strictly identical. We show that the above picture
of quark and anti quark asymmetry in the momentum and helicity distributions of the
nucleon sea quarks has support from a number of experimental observations, and we
suggest processes to test and measure this quark and antiquark asymmetry in the
nucleon sea.
Consequences of Intrinsic Charm and Bottom
Microscopically, the intrinsic heavy-quark Fock component in the 7r- wavefunction,

lU:dQQ), is generated by virtual interactions such as 99 - t QQ where the gluons couple
to two or more projectile valence quarks. The probability for QQ fluctuations to exist
in a light hadron thus scales as o:~(m~)/m~ relative to leading-twist production 58 .
This contribution is therefore higher twist, and power-law suppressed compared to
sea quark contributions generated by gluon splitting. When the projectile scatters
in the target, the coherence of the Fock components is broken and its fluctuations can
hadronize, forming new hadronic systems from the fluctuations 17 . For example, intrinsic
cc fluctuations can be liberated provided the system is probed during the characteristic
time 6.t = 2P1ab/ M;c that such fluctuations exist. For soft interactions at momentum
scale fl, the intrinsic heavy quark cross section is suppressed by an additional resolving
factor ex fl2 / m~ 59. The nuclear dependence arising from the manifestation of intrinsic
charm is expected to be ITA ~ ITN A 2 / 3 , characteristic of soft interactions.
In general, the dominant Fock state configurations are not far off shell and thus
have minimal invariant mass M2 = Li m},dxi where mT,i is the transverse mass of the
ith particle in the configuration. Intrinsic QQ Fock components with minimum invariant
165
mass correspond to configurations with equal-rapidity constituents. Thus, unlike sea
quarks generated from a single parton, intrinsic heavy quarks tend to carry a larger
fraction of the parent momentum than do the light quarks 56 . In fact, if the intrinsic
QQ pair coalesces into a quarkonium state, the momentum of the two heavy quarks is
combined so that the quarkonium state will carry a significant fraction of the projectile
momentum.
There is substantial evidence for the existence of intrinsic cc fluctuations in the
wavefunctions of light hadrons. For example, the charm structure function of the proton
measured by EMC is significantly larger than that predicted by photon-gluon fusion
at large XB/o. Leading charm production in 7r Nand hyperon-N collisions also re-
quires a charm source beyond leading twist58, 61. The N A3 experiment has also shown
that the single J /1/J cross section at large XF is greater than expected from gg and qlj
production 62 The nuclear dependence of this forward component is diffractive-like, as
expected from the BHMT mechanism. In addition, intrinsic charm may account for the
anomalous longitudinal polarization of the J /1/J at large x F seen in 7r N --. J/ 1/JX inter-
actions 63 . Further theoretical work is needed to establish that the data on direct J/ 1/J
and Xl production can be described using the higher-twist intrinsic charm mechanism 17 .
A recent analysis by Harris, Smith and Vogt 64 of the excessively large charm struc-
ture function of the proton at large x as measured by the EMC collaboration at CERN
yields an estimate that the probability Pee that the proton contains intrinsic charm Fock
states is of the order of 0.6% 0.3%. In the case of intrinsic bottom, PQCD scaling
predicts
(11)
more than an order of magnitude smaller. If super-partners of the quarks or gluons exist
they must also appear in higher Foc~ states of the proton, such as luud gluino gluino).
At sufficiently high energies, the diffractive excitation of the proton will produce these
intrinsic quarks and gluinos in the proton fragmentation region. Such supersymmetric
particles can bind with the valence quarks to produce highly unusual color-singlet hy-
brid supersymmetric states such as luud gluino) at high XF. The probability that the
proton contains intrinsic gluinos or squarks scales with the appropriate color factor and
inversely with the heavy particle mass squared relative to the intrinsic charm and bot-
tom probabilities. This probability is directly reflected in the production rate when the
hadron is probed at a hard scale Q which is large compared to the virtual mass M of
the Fock state. At low virtualities, the rate is suppressed by an extra factor of Q2 / M2.
The forward proton fragmentation regime is a challenge to instrument at HERA, but
it may be feasible to tag special channels involving neutral hadrons or muons. In the
case of the gas jet fixed-target ep collisions such as at HERMES, the target fragments
emerge at low velocity and large backward angles, and thus may be accessible to precise
measurement.
Double Quarkonium Hadroproduction
It is quite rare for two charmonium states to be produced in the same hadronic
collision. However, the NA3 collaboration has measured a double J/1/J production
rate significantly above background in multi-muon events with 7r- beams at laboratory
momentum 150 and 280 GeV /c and a 400 GeV /c proton beam65 . The relative double
to single rate, (11/;1/;/(11/;, is (3 1) x 10- 4 for pion-induced production, where (11/; is the
integrated single 1/J production cross section. A particularly surprising feature of the
NA3 7r- N --. 1/J1/JX events is that the laboratory fraction of the projectile momentum
166
carried by the 1/J1/J pair is always very large, X,p,p ~ 0.6 at 150 GeV I c and X,p,p ~ 0.4
at 280 GeV Ic. In some events, nearly all of the projectile momentum is carried by
the 1/J1/J system! In contrast, perturbative 99 and qq fusion processes are expected to
produce central 1/J1/J pairs, centered around the mean value, (x,p,p) ~ 0.4-0.5, in the
laboratory. There have been attempts to explain the NA3 data within conventional
leading-twist QCD. Charmonium pairs can be produced by a variety of QCD processes
including BB production and decay, BB ---+ 1/J1/JX and 0(0:;) 1/J1/J production via 99
fusion and qq annihilation 66 , 67. Li and Liu have also considered the possibility that a
2++cccc resonance is produced, which then decays into correlated 1/J1/J pairs68. All of
these models predict centrally produced 1/J1/J pairs69, 67, in contradiction to the 7r- data.
Over a sufficiently short time, the pion can contain Fock states of arbitrary com-
plexity. For example, two intrinsic cc pairs may appear simultaneously in the quantum
fluctuations of the projectile wavefunction and then, freed in an energetic interaction,
coalesce to form a pair of 1/J's. In the simplest analysis, one assumes the light-cone
Fock state wavefunction is approximately constant up to the energy denominator58.
The predicted 1/J1/J pair distributions from the intrinsic charm model provide a natural
explanation of the strong forward production of double J 11/J hadroproduction, and thus
gives strong phenomenological support for the presence of intrinsic heavy quark states
in hadrons.
It is clearly important for the double J 11/J measurements to be repeated with higher
statistics and at higher energies. The same intrinsic Fock states will also lead to the
production of multi-charmed baryons in the proton fragmentation region. The intrin-
sic heavy quark model can also be used to predict the features of heavier quarkonium
hadroproduction, such as YY, Y 1/J, and (cb) (cb) pairs. It is also interesting to study
the correlations of the heavy quarkonium pairs to search for possible new four-quark
bound states and final state interactions generated by multiple gluon exchange68 , since
the QCD Van der Waals interactions could be anomalously strong at low relative ra-
pidit y 51, 52.
Leading Particle Effect in Open Charm Production
According to PQCD factorization, the fragmentation of a heavy quark jet is inde-

pendent of the production process. However, there are strong correlations between the
quantum numbers of D mesons and the charge of the incident pion beam in 7r N ---+ DX
reactions. This effect can be explained as being due to the coalescence of the produced
intrinsic charm quark with co-moving valence quarks. The same higher-twist recombi-
nation effect can also account for the suppression of J 11/J and Y production in nuclear
collisions in regions of phase space with high particle density58.
There are other ways in which the intrinsic heavy quark content of light hadrons can
be tested. More measurements of the charm and bottom structure functions at large x F
are needed to confirm the EMC data60 . Charm production in the proton fragmentation
region in deep inelastic lepton-proton scattering is sensitive to the hidden charm in the
proton wavefunction. The presence of intrinsic heavy quarks in the hadron wavefunction
also enhances heavy flavor production in hadronic interactions near threshold. More
generally, the intrinsic heavy quark model leads to enhanced open and hidden heavy
quark production and leading particle correlations at high XF in hadron collisions,
with a distinctive strongly shadowed nuclear dependence characteristic of soft hadronic
collisions.
It is of particular interest to examine the fragmentation of the proton when the
electron strikes a light quark and the interacting Fock component is the luudcc) or
167
luudbb) state. These Fock components correspond to intrinsic charm or intrinsic bottom
quarks in the proton wavefunction. Since the heavy quarks in the proton bound state
have roughly the same rapidity as the proton itself, the intrinsic heavy quarks will
appear at large XF. One expects heavy quarkonium and also heavy hadrons to be
formed from the coalescence of the heavy quark with the valence u and d quarks,
since they have nearly the same rapidity. Since the heavy and valence quark momenta
combine, these states are preferentially produced with large longitudinal momentum
fractions
The role of intrinsic charm becomes dominant over leading-twist fusion processes
near threshold, since the multi-connected intrinsic charm configurations in the higher
light-cone Fock state of the proton are more efficient that gluon splitting in producing
charm. The heavy c and c will be produced at low velocities relative to each other
and with the spectator quarks from the proton and virtual photon. As is the case of
e+e- -+ cc near threshold, the QCD Coulomb rescattering will give Sommerfeld cor-
rection factors 5((3, Q2) which strongly distort the Born predictions for the production
amplitudes.
THE FORM FACTORS OF ELEMENTARY AND COMPOSITE SYSTEMS
In this section I will review the light-cone formalism for both elementary and
composite systems 70, 71, 15. We choose light-cone coordinates with the incident lepton
directed along the z direction 72 (p == pO p3):
P = (+
J1. _ -
P ,p ,p
-t )_(
1 - p, P+'
+ M2 0) .l ,
2q p
q = ( O'y'
-t)
q.l , (12)
where q2 = -2q . p = -ql and M = mi is the mass of the composite system. The
Dirac and Pauli form factors can be identified71 from the spin-conserving and spin-flip
current matrix elements (J+ = J O+ J3):
Mij \P+q,i IJ;iO)lp,i} =2Fl(l) , (13)
Mil = \P + q, iIJ;iO) I i} =
p, -2(ql - iq2) F~~) , (14)
where i corresponds to positive spin projection Sz = +~ along the axis. z

Each Fock-state wave function In) of the incident lepton is represented by the
functions ~~~lJxi' I:.i' Si), where
specifies the light-cone momentum coordinates of each constituent i = 1, ... ,n, and 5 i
specifies its spin projection 5;. Momentum observation on the light cone requires
n
LXi = 1,
i=l

and thus < Xi < 1. The amplitude to find n (on-mass-shell) constituents in the lepton
is then ~(n) multiplied by the spinor factors usi (k i )/(kn- 1 / 2 or vsi (k;)/(kn 1 / 2 for each
168
constituent fermion or anti-fermion 73 The Fock state is off the "energy shell":
---4 2 2
The quantity (k 1.; + mf)/xi is the relativistic analog of the kinetic energy Pi /2mi in
the Schrodinger formalism.
The wave function for the lepton directed along the final direction p + q in the
current matrix element is then
where 14
for the struck constituent and
p + q direction with
---4
for each spectator (i -I- j). The k .i are transverse to the
Lk
n ---4'
1.i = 0.
i=1
The interaction of the current )+(0) conserves the spin projection of the struck
constituent fermion (Us,,+us)/k+ = 28sS I. Thus from Eqs. (13) and (14)
(15)
and
_ (q1 - i q2) F ( 2) = !M+

2M 1 q 2 11
= 21ejj[dX] [d2~] ~;~nj'1(X,k:,S) ~~~i(x,~,s) , (16)
ej
where is the fractional charge of each constituent. [A summation of all possible Fock
states (n) and spins (S) is assumed.] The phase-space integration is
n
[dx] == 8(1 - LXi) II dXi , (17)
;=1
and
(18)
Equation (15) evaluated at q2 = 0 with F1(0) = 1 is equivalent to wave-function

normalization. The anomalous moment a = F2 (0)/ F1 (0) can be determined from the
coefficient linear in q1 - iq2 from the coefficient linear in q1 - iq2 from ~;+q in Eq. (16).
In fact, since74
[)
[)---4
.1.*
'f'p+q
-"
= - L.J Xi--=:;-
[) .1.*
'f'p+q (19)
q 1. i#j [)k 1.i
169
(summed over spectators), we can, after integration by parts, write explicitly
~ = - ~ej j[dx]j [lfkl-] ~t/J;TXi (O~li +io~2J t/Jp!. (20)
The wave function normalization is

(21)
A sum over all contributing Fock states is assumed in Eqs. (20) and (21).
We thus can express the anomalous moment in terms of a local matrix element at
zero momentum transfer. It should be emphasized that Eq. (20) is exact; it is valid for
the anomalous element of any spin-~ system.
As an example, in the case of the electron's anomalous moment to order Q' in
QED,15 the contributinq intermediate Fock states are the electron-photon states with
spins 1- 1)
~, and ~, I -1 ;:
V2 (kl ~ik21 (I-~) --t I-~, 1))
(22)
V2 M(II~2-;;> (I-~) --t 1~,-1))
and
* e/.fi { -V2 M(\~2-;;> (1-~,1) I~)) --t
(23)
t/JpT =
M2-~-~
k 2 +A2 k 2 +;;>2 X 2 (kl-xik21 (112,-1 )
_ 10
yE.
11))
2 . --t
The quantities to the left of the curly bracket in Eqs. (22) and (23) are the matrix
elements of
u * U and u U
(p+ _ k+ )1/2 , . E (p+ )1/2 (p+F/2'E(p+-k+)1/2 '
respectively, where t = tlW = (l/V2)(x iii), E k = 0, f+ = 0 in the light-cone
gauge for vector spin projection Sz = 1 1o,11. For the sake of generality, we let the
intermediate lepton and vector boson have mass mand ), respectively.
Substituting (22) and (23) into Eq. (20), one finds that only the I-~, interme- 1)
diate state actually contributes to a, since terms which involve differentiation of the
denominator of t/Jp! cancel. We thus have15
d2kl- {I [m - (1 - x)M] /x(1- x)
a = 4M e2 j 167r 3 10 dx -[M-2---(.!..ki-+-m"-2-)-/(--'l'------'-'x )---'---(-ki----<-+-)-2-)/-x-;;"]2 ,
(24)
or
Q't M[m-M(l-x)]x(l-x)
a = -; 10 dx m2x + ,),2(1 _ x) - M2x(1- x) , (25)
which, in the case of QED (m = M,). = 0) gives the Schwinger results a = Q'/27r.
The general result (20) can also be written in matrix form:
2~ = - ~ej J[dxJ[ d2k l-l t/J+

J
s:. L:t/J, (26)
--+
where S is the spin operator for the total system and L l- is the generator of "Galiean"
--+ --+
transverse boosts10,11 on the light cone, i.e., S l- . Ll- = (S+L_ + S_L+)/2 where
S = (SI iS2 ) is the spin-ladder operator and
L = LXi
i'i
(oZ. l-.
=f i o~ .)
2.
(27)
170
(summed over spectators) in the analog of the angular momentum operator p x r+.
Equation (20) can also be written simply as an expectation value in impact space.
The results given in Eqs. (15), (16), and (20) may also be convenient for calculating
the anomalous moments and form factors of hadrons in quantum chromo dynamics
. --+
directly from the quark and gluon wave functions 1/J( k.1, X, S). These wave functions
can also be used to construct the structure functions and distribution amplitudes which
control large momentum transfer inclusive and exclusive processes 71 , 76. The charge
radius of a composite system can also be written in the form of a local, forward matrix
element 77 :
r
We thus find that, in general, any Fock state In) which couples to both 1/J and 1/J!
will give a contribution to the anomalous moment. Notice that because of rotational
symmetry in the X, fj direction, the contribution to a = F2 (0) in Eq. (20) always involves
the form (a, b = 1, ... , n)
(29)
compared to the integral (21) for wave-function normalization which has terms of order
and
(30)
here p is a rotationally invariant function of the transverse momenta and f-t is a constant
with dimensions of mass. Thus, in order of magnitude
(31)
summed and weighted over the Fock states. In the case of a renormalizable theory,
the only parameters f-t with the dimension of mass are fermion masses. In super-
renormalizable theories, p can be proportional to a coupling constant g with dimension
of mass 78 .
In the case where all the mass-scale parameters of the composite state are of the
same order of magnitude, we obtain a = O(M R) as in Eqs. (17) and (18), where
R = (ki) -1/2 is the characteristic size79 of the Fock State. On the other hand, in
theories where f-t2 ~ (kl), we obtain the quadratic relation a = O(f-tMR2).
Thus composite models for leptons can avoid conflict with the high-precision QED
measurements in several ways .
There can be strong cancellations between the contribution of different Fock

states.
171
The parameter fJ can be minimized. For example, in a renormalizable theory
this can be accomplished by having the bound state of light fermions and heavy
bosons. Since fJ 2: M, we then have a 2: O(M2 R2) .
If the parameter fJ is of the same order s the other mass scales in the composite
state, then we have a linear condition a = O(MR).
MOMENTS OF NUCLEONS AND NUCLEI IN THE LIGHT-CONE FOR-

MALISM
The use of covariant kinematics leads to a number of striking conclusions for the
electromagnetic and weak moments of nucleons and nuclei. For example, magnetic
moments cannot be written as the naive sum 71 = L 71 i of the magnetic moments
of the constituents, except in the nonrelativistic limit where the radius of the bound
state is much larger than its Compton scale: RAMA ~ 1. The deuteron quadrupole
moment is in general nonzero even if the nucleon-nucleon bound state has no D-wave
component So Such effects are due to the fact that even "static" moments must be
computed as transitions between states of different momentum pi" and pI' + ql', with
ql' -+ O. Thus one must construct current matrix elements between boosted states.
The Wigner boost generates nontrivial correCtions to the current interactions of bound
systemsS1 Remarkably, in the case of the deuteron, both the quadrupole and magnetic
moments become equal to that of the Standard Model in the limit MdRd -+ O. In this
limit, the three form factors of the deuteron have the same ratios as do those of the W
boson in the Standard Modelso .
One can also use light-cone methods to show that the proton's magnetic moment
fJp and its axial-vector coupling 9A have a relationship independent of the specific form
of the light-cone wavefunctions2 At the physical value of the proton radius computed
from the slope of the Dirac form factor, Rl = 0.76 fm, one obtains the experimental
values for both fJp and gA; the helicity carried by the valence u and d quarks are
each reduced by a factor ~ 0.75 relative to their nonrelativistic values. At infinitely
small radius RpMp -+ 0, fJp becomes equal to the Dirac moment, as demanded by the
Drell-Hearn-Gerasimov sum rules3, S4. Another surprising fact is that as Rl -+ 0 the
constituent quark helicities become completely disoriented and 9A -+ O.
One can understand the origins of the above universal features even in an effective
three-quark light-cone Fock description of the nucleon. In such a model, one assumes
that additional degrees of freedom (including zero modes) can be parameterized through
an effective potentiaP6. After truncation, one could in principle obtain the mass M and
light-cone wavefunction of the three-quark bound-states by solving the Hamiltonian
eigenvalue problem. It is reasonable to assume that adding more quark and gluonic
excitations will only refine this initial approximation lO In such a theory the constituent
quarks will also acquire effective masses and form factors.
Since we do not have an explicit representation for the effective potential in the
light-cone Hamiltonian Pefr for three quarks, we shall proceed by making an Ansatz for
the momentum-space structure of the wavefunction Ill. Even without explicit solutions
of the Hamiltonian eigenvalue problem, one knows that the helicity and flavor struc-
ture of the baryon eigenfunctions will reflect the assumed global SU(6) symmetry and
Lorentz invariance of the theory. As we will show below, for a given size of the proton
the predictions and interrelations between observables at Q2 = 0, such as the proton
172
magnetic moment IIp and its axial coupling gA, turn out to be essentially independent
of the shape of the wavefunction 82 .
The light-cone model given by Ma85 and by Schlump86 provides a framework for
representing the general structure of the effective three-quark wavefunctions for baryons.
The wavefunction \Ii is constructed as the product of a momentum wavefunction, which
is spherically symmetric and invariant under permutations, and a spin-isospin wave
function, which is uniquely determined by SU(6)-symmetry requirements. A Wigner-
Melosh rotation 87 , 88 is applied to the spinors, so that the wavefunction of the proton
is an eigenfunction of J and Jz in its rest frame 89, 90, 91. To represent the range of
uncertainty in the possible form of the momentum wavefunction, one can choose two
simple functions of the invariant mass M of the quarks:
"pH.o.(M 2 ) N H . O . exp( _M2 /2f32), (32)

"pPower ( M 2) NPower(l + M2 / f32)-p , (33)
where f3 sets the characteristic internal momentum scale. Perturbative QeD predicts a
nominal power-law fall off at large k1. corresponding to p = 3.516 . The Melosh rotation
insures that the nucleon has j = ~ in its rest system. It has the matrix representation 88
R M ( x.,. k. )_
1.., m -
m + Xi M - iif . (it X i )
--r=========--'-
k (34)
v(m + xiM)2 + ki;
with ii = (0,0,1), and it becomes the unit matrix if the quarks are collinear, RM(Xi, 0, m) =
1. Thus the internal transverse momentum dependence of the light-cone wavefunctions
also affects its helicity structure81 .
The Dirac and Pauli form factors F1 (Q2) and F2( Q2) of the nucleons are given
by the spin-conserving and the spin-flip matrix elements of the vector current Jt (at
Q2 = _q2)15
F 1 (Q2) (p + q, t IJtlp, t), (35)

(Q1 - iQ2)F2( Q2) -2M(p + q, t IJtlp, 1) . (36)
We then can calculate the anomalous magnetic moment a = limQ2 ..... o F 2 ( Q2). t The
same parameters as given by Schlump86 are chosen, namely m = 0.263 GeV (0.26
GeV) for the up (down) quark masses, f3 = 0.607 GeV (0.55 GeV) for "pPower ("pH.D.),
and p = 3.5. The quark currents are taken as elementary currents with Dirac moments
f;;;. All of the baryon moments are well-fit if one takes the strange quark mass as 0.38
GeV. With the above values, the proton magnetic moment is 2.81 nuclear magnetons,
and the neutron magnetic moment is -1.66 nuclear magnetons. (The neutron value
can be improved by relaxing the assumption of isospin symmetry.) The radius of the
proton is 0.76 fm, i.e., MpR1 = 3.63.
In Fig. 3(a) we show the functional relationship between the anomalous moment
ap and its Dirac radius predicted by the three-quark light-cone model. The value of
R2 = _6 dF1 (Q2) I (37)

1 dQ2 Q2=0
is varied by changing f3 in the light-cone wavefunction while keeping the quark mass
m fixed. The prediction for the power-law wavefunction "pPower is given by the broken
tThe total proton magnetic moment is /lp = 2~ (1 + ap ).
173
line; the continuous line represents 'l/JH.O .. Figure 3(a) shows that when one plots the
dimensionless observable ap against the dimensionless observable M Rl the prediction is
essentially independent of the assumed power-law or Gaussian form of the three-quark
light-cone wavefunction. Different values of p > 2 also do not affect the function-
al dependence of ap(MpRl) shown in Fig. 3(a). In this sense the predictions of the
three-quark light-cone model relating the Q2 - t 0 observables are essentially model-
independent. The only parameter controlling the relation between the dimensionless
observables in the light-cone three-quark model is m/ Mp which is set to 0.28. For the
physical proton radius MpRl = 3.63 one obtains the empirical value for ap = 1.79
(indicated by the dotted lines in Fig. 3(a)).
The prediction for the anomalous moment a can be written analytically as a =
bv )a NR , where aNR = 2Mp/3m is the nonrelativistic (R - t 00) value and 'v is given
as 92
(38)
The expectation value bv) is evaluated as*
(39)
Let us now take a closer look at the two limits R - t 00 and R - t O. In the
nonrelativistic limit we let f3 - t 0 and keep the quark mass m and the proton mass
Mp fixed. In this limit the proton radius Rl - t 00 and ap - t 2Mp/3m = 2.38, since
(,v) -t. 1. t Thus the physical value of the anomalous magnetic moment at the empirical
proton radius MpRl = 3.63 is reduced by 25% from its nonrelativistic value due to
relativistic recoil and nonzero k.l. t
To obtain the ultra-relativistic limit we let f3 - t 00 while keeping m fixed. In this
limit the proton becomes pointlike, MpRl - t 0, and the internal transverse momenta
k.l - t 00. The anomalous magnetic momentum of the proton goes linearly to zero
as a = 0.43Mp R l since bv) - t O. Indeed, the Drell- Hearn-Gerasimov sum rule83 ,84
demands that the proton magnetic moment become equal to the Dirac moment at
small radius. For a spin-! system
a2 = -M2
2-
27r a
1 00
8th
ds
-
S
[O"p(s) - O"A(S)] , (40)
where O"p(A) is the total photoabsorption cross section with parallel (anti-parallel) pho-
ton and target spins. If we take the point-like limit, such that the threshold for inelastic
excitation becomes infinite while the mass of the system is kept finite, the integral over
the photoabsorption cross section vanishes and a = 0.15 In contrast, the anomalous
magnetic moment of the proton does not vanish in the nonrelativistic quark model as
R - t o. The nonrelativistic quark model does not reflect the fact that the magnetic
moment of a baryon is derived from lepton scattering at nonzero momentum transfer,
*Here [J3kj == dk1dk2dk38(kl + k2 + k3). The third component of k is defined as k3i == ~(xiM -
k1i ).
m'+M
3;,
This measure differs from the usual one used 16 by the Jacobian ~dk
x,
n
.. which can be absorbed
into the wavefunction.
tThis differs slightly from the usual nonrelativistic formula 1 + a = 6 ~~ e mq
"q
due to the non-
vanishing binding energy which results in Mp i= 3mq.
+The nonrelativistic value of the neutron magnetic moment is reduced by 31%.
174
Figure 3: (a). The anomalous magnetic moment of the proton a p = F2 (0) as a function
of its Dirac radius MpRl in Compton units. (b). The axial vector coupling of the
neutron to proton beta-decay as a function of MpRl' In each figure, the broken line is
computed from a wavefunction with power-law fall off and the solid curve is computed
from a Gaussian wavefunction. The experimental values at the physical proton Dirac
radius are indicated by the dotted line82
i.e., the calculation of a magnetic moment requires knowledge of the boosted wavefunc-
tion. The Melosh transformation is also essential for deriving the DHG sum rule and
low-energy theorems of composite systems81 .
A similar analysis can be performed for the axial-vector coupling measured in
neutron decay. The coupling 9A is given by the spin-conserving axial current Jt matrix
element
9A(0) = (p, i IJtlp, j) . (41)
The value for 9A can be written as 9A = ({A)9~R, with 9~R being the nonrelativistic
value of 9A and with 'YA given by92, 93
(m + x3M)2 - ki3
'YA(Xi, k.li' m) = (m + X3 M)2 + k21.3 . (42)
In Fig. 3(b) the axial-vector coupling is plotted against the proton radius MpRl' The
same parameters and the same line representation as in Fig. 3(a) are used. The func-
tional dependence of 9A(MpRt} is also found to be independent of the assumed wave-
function. At the physical proton radius MpRl = 3.63, one predicts the value 9A = 1.25
(indicated by the dotted lines in Fig. 3(b, since ((A) = 0.75. The measured value
is 9A = 1.2573 0.002894 This is a 25% reduction compared to the nonrelativistic
SU(6) value 9A = 5/3, which is only valid for a proton with large radius Rl ~ l/Mp.
The Melosh rotation generated by the internal transverse momentum93 spoils the usual
identification of the 'Y+'Y5 quark current matrix element with the total rest-frame spin
projection sz, thus resulting in a reduction of 9A.
Thus, given the empirical values for the proton's anomalous moment a p and radius
MpRl, its axial-vector coupling is automatically fixed at the value 9A = 1.25. This is
an essentially model-independent prediction of the three-quark structure of the proton
in QCD. The Melosh rotation of the light-cone wavefunction is crucial for reducing
the value of the axial coupling from its nonrelativistic value 5/3 to its empirical value.
The near equality of the ratios 9A/9A(R1 - t 00) and ap/ap(Rl - t 00) as a function
of the proton radius Rl shows the wave-function independence of these quantities.
175
We emphasize that at small proton radius the light-cone model predicts not only a
vanishing anomalous moment but also limR 1 --+o9A(Mp R I ) = O. One can understand
this physically: in the zero radius limit the internal transverse momenta become infinite
and the quark helicities become completely disoriented. This is in contradiction with
chiral models, which suggest that for a zero radius composite baryon one should obtain
the chiral symmetry result 9A = l.
The helicity measures ~u and ~d of the nucleon each experience the same reduc-
tion as does 9A due to the Melosh effect. Indeed, the quantity ~q is defined by the
axial current matrix element
(43)
and the value for ~q can be written analytically as ~q = bA)~qNR, with ~qNR being
the nonrelativistic or naive value of ~q and fA given by Eq. (42).
The light-cone model also predicts that the quark helicity sum ~~ = ~u + ~d
vanishes as a function of the proton radius R I . Since ~~ depends on the proton size,
it cannot be identified as the vector sum of the rest-frame constituent spins. The rest-
frame spin sum is not a Lorentz invariant for a composite system93 . Empirically, one
can measure ~q from the first moment of the leading-twist polarized structure function
91(X, Q). In the light-cone and parton model descriptions, ~q = J~ dx[qT(x) - ql(x)],
where qT(x) and q!(x) can be interpreted as the probability for finding a quark or
antiquark with longitudinal momentum fraction x and polarization parallel or anti-
parallel to the proton helicity in the proton's infinite momentum frame l6 . [In the
infinite momentum frame there is no distinction between the quark helicity and its spin
projection sz.] Thus ~q refers to the difference of helicities at fixed light-cone time or
at infinite momentum; it cannot be identified with q(sz = +~) - q(sz = -~), the spin
carried by each quark flavor in the proton rest frame in the equal-time formalism.
Thus the usual SU(6) values ~uNR = 4/3 and ~dNR = -1/3 are only valid pre-
dictions for the proton at large M R I . At the physical radius the quark helicities are
reduced by the same ratio 0.75 as is gAl g~R due to the Melosh rotation. Qualitative
arguments for such a reduction have been given elsewhere95, 96. For MpRI = 3.63, the
three-quark model predicts ~u = 1, ~d = -1/4, and ~~ = ~u+~d = 0.75. Although
the gluon contribution ~G = 0 in our model, the general sum rule97
1 1
-~~+~G+L
2 z
=-
2 (44)
is still satisfied, since the Melosh transformation effectively contributes to L z

Suppose one adds polarized gluons to the three-quark light-cone model. Then
the flavor-singlet quark-loop radiative corrections to the gluon propagator will give an
anomalous contribution b(~q) = -~~G to each light quark helicity98. The predicted
value of 9A = ~u - ~d is of course unchanged. For illustration we shall choose ~; ~G =
0.15. The gluon-enhanced quark model then gives the values in Table 1, which agree
well with the present experimental values. Note that the gluon anomaly contribution
to ~s has probably been overestimated here due to the large strange quark mass. One
could also envision other sources for this shift of ~q such as intrinsic flavor 96 . A specific
model for the gluon helicity distribution in the nucleon bound state is given elsewhere l8 .
In the above analysis of the singlet moments, it is assumed that all contributions to
the sea quark moments derive from the gluon anomaly contribution b(~q) = -~;~G.
In this case the strange and anti-strange quark distributions will be identical. On the
other hand, if the strange quarks derive from the intrinsic structure of the proton,
176
Quantity NR 3q 3q + g Experiment
~u
4
:3 1 0.85 0.83 0.03
~d _1 _1 -0.40 -0.43 0.03
3 4
~s 0 0 -0.15 -0.10 0.03
~~ 1 3
:I 0.30 0.31 0.07
Table 1: Comparison of the quark content of the proton in the nonrelativistic quark
model (NR), in the three-quark model (3q), in a gluon-enhanced three-quark model
(3q + g), and with experiment 99 .
then one would not expect this symmetry. For example, in the intrinsic strangeness
wavefunctions, the dominant fluctuations in the nucleon wavefunction are most likely
dual to intermediate A- J( configurations since they have the lowest off-shell light-cone
energy and invariant mass. In this case s( x) and s( x) will be different.
The light-cone formalism also has interesting consequences for spin correlations in
jet fragmentation. In LEP or SLC one produces sand s quarks with opposite helicity.
This produces a correlation of the spins of the A and X, each produced with large z in
the fragmentation of their respective jet. The A spin essentially follows the spin of the
strange quark since the ud has J = O. However, this cannot be a 100% correlation since
the A generally is produced with some transverse momentum relative to the s jet. In
fact, from the light-cone analysis of the proton spin, we would expect no more than a
75% correlation since the A and proton radius should be almost the same. On the other
hand if z = EA/ Es - t 1, there can be no wasted energy in transverse momentum. At
this point one could have 100% polarization. In fact, the nonvalence Fock states will be ,
suppressed at the extreme kinematics, so there is even more reason to expect complete
helicity correlation in the endpoint region.
We can also apply a similar idea to the study of the fragmentation of strange
quarks to As produced in deep inelastic lepton scattering on a proton. One can use the
correlation between the spin of the target proton and the spin of the A to directly mea-
sure the strange polarization ~s. It is conceivable that any differences between ~s and
~s in the nucleon wavefunction could be distinguished by measuring the correlations
between the target polarization and the A and A polarization in deep inelastic lepton
proton collisions or in the target polarization region in hadron-proton collisions.
In summary, we have shown that relativistic effects are crucial for understanding
the spin structure of nucleons. By plotting dimensionless observables against dimen-
sionless observables, we obtain relations that are independent of the momentum-space
form of the three-quark light-cone wavefunctions. For example, the value of gA ~ 1.25
is correctly predicted from the empirical value of the proton's anomalous moment. For
the physical proton radius MpRl = 3.63, the inclusion of the Wigner-Melosh rotation
due to the finite relative transverse momenta of the three quarks results in a rv 25%
reduction of the nonrelativistic predictions for the anomalous magnetic moment, the
axial vector coupling, and the quark helicity content of the proton. At zero radius, the
quark helicities become completely disoriented because of the large internal momenta,
resulting in the vanishing of gA and the total quark helicity ~~.
177
CONCLUSIONS
One of the central problems in particle physics is to determine the structure of

hadrons in terms of their fundamental QCD quark and gluon degrees of freedom. As
I have outlined in this talk, the light-cone Fock representation of quantum chromo-
dynamics provides both a tool and a language for representing hadrons as fluctuating
composites of fundamental quark and gluon degrees of freedom. Light-cone quantiza-
tion provides an attractive method to compute this structure from first principles in
QCD. However, much more progress in theory and in experiment will be needed to
fulfill this promise.
Acknowledgements
Much of this talk is based on an earlier review written in collaboration with Dave
Robertson and on collaborations with Matthias Burkardt, Sid Drell, John Hiller, Kent
Hornbostel, Paul Hoyer, Hung Jung Lu, Bo-Qiang Ma. Al Mueller, Chris Pauli, Steve
Pinsky, Felix Schlumpf, Ivan Schmidt, Wai-Keung Tang, and Ramona Vogt. I also
thank Simon Dalley and Francesco Antonuccio for helpful discussions.
References
1. P. A. M. Dirac, Rev. Mod. Phys. 21,392 (1949); S. Weinberg, Phys. Rev. 150, 1313
(1966).
2. S. J. Brodsky and H.-C. Pauli, in Recent Aspects of Quantum Fields, H. Mitter and
H. Gausterer, Eds., Lecture Notes in Physics, Vol. 396 (Springer-Verlag, 1991), and
references therein.
3. M. Burkardt, Nucl. Phys. A 504, 762 (1989).
4. K. Hornbostel, S. J. Brodsky, and H.-C. Pauli, Phys. Rev. D 41, 3814 (1990).
5. K. Demeterfi, I. R. Klebanov, and G. Bhanot, Nucl. Phys. B 418, 15 (1994).
6. M. Krautgiirtner, H.-C. Pauli, and F. W61z, Phys. Rev. D 45, 3755 (1992); M. Kaluza
and H.-C. Pauli, Phys. Rev. D 45, 2968 (1992).
7. J. R. Hiller, S. J. Brodsky, and Y. Okamoto, in progress (1995); J. R. Hiller, in
Theory of Hadrons and Light-Front QeD, edited by St. D. Glazek (World Scientif-
ic, 1995), and talk presented at the 5th Meeting on Light-Cone Quantization and
Nonperturbative QCD, Regensburg, Germany, June 1995.
8. M. Burkardt, Phys. Rev. D 49, 5446 (1994).
9. L.C.L. Hollenberg and N. S. Witte, Phys. Rev. D 50,3382 (1994).
10. R. J. Perry, A. Harindranath and K. G. Wilson, Phys. Rev. Lett. 65, 2959 (1990).
11. R. J. Perry, Ann. Phys. 232, 116 (1994).
12. K. G. Wilson, T. S. Walhout, A. Harindranath, W.-M. Zhang, R. J. Perry, and
St. D. Glazek, Phys. Rev. D 49, 6720 (1994).
13. M. Brisudova, talk presented at the 5th Workshop on Light-Cone QCD, Telluride,
CO, August 1995; M. Brisudova and R. J. Perry, manuscript in preparation (1995).
14. S. D. Drell and T. M. Van, Phys. Rev. Lett. 24, 181 (1970).
15. S. J. Brodsky and S. D. Drell, Phys. Rev. D 22, 2236 (1980).
16. G. P. Lepage and S. J. Brodsky, Phys. Rev. D 22, 2157 (1980).
178
17. S. J. Brodsky, P. Hoyer, A. H. Mueller, W.-K. Tang, Nucl. Phys. B369, 519 (1992).
18. S. J. Brodsky, M. Burkardt, and I. A. Schmidt, Nucl. Phys. B 441, 197 (1994).
19. H.-C. Pauli and S. J. Brodsky, Phys. Rev. D 32, 1993, 2001 (1985).
20. H. Bergknoff, Nucl. Phys. B 122, 215 (1977); T. Eller, H.-C. Pauli, and S. J. Brod-
sky, Phys. Rev. D 35, 1493 (1987); Y. Mo and R. J. Perry, J. Compo Phys. 108,159
(1993); K. Harada, A. Okazaki, and M. Taniguchi, Phys. Rev. D 52, 2429 (1995)
and KYUSHU-HET 28, hep-th/9509136.
21. G. McCartor, Z. Phys. C 52,611 (1991); ibid. 64, 349 (1994).
22. A. Harindranath and J. P. Vary, Phys. Rev. D 36, 1064, 1141 (1987); ibid. 37,3010
(1988).
23. S. Dalley and I. R. Klebanov, Phys. Rev. D 47, 2517 (1993); G. Bhanot, K. Deme-
terfi, and I. R. Klebanov, Phys. Rev. D 48, 4980 (1993);
24. F. Antonuccio and S. Dalley, OUTP-9518P, hep-lat/9505009, and OUTP-9524P,
hep-ph/9506456.
25. M. Burkardt, Phys. Rev. D 47, 4628 (1993).
26. St. D. Glazek, A. Harindranath, S. Pinsky, J. Shigemitsu, and K. G. Wilson, Phys.
Rev. D 47, 1599 (1993); P. M. Wort, Phys. Rev. D 47, 608 (1993).
27. S. J. Brodsky and D. G. Robertson, SLAC-PUB-95-7056, OSU-NT-95- 06.
28. G. 't Hooft, Nucl. Phys. B 75, 461 (1974).
29. C. J. Hamer, Nucl. Phys. B 195, 503 (1982).
30. G. D. Date, Y. Frishman and J. Sonnenschein, Nucl. Phys. B 283, 365 (1987).
31. M. Burkardt, New Mexico State University Preprint (1996), hep-ph/9601289.
5(1994); A. C. Kalloniatis and D. G. Robertson, Phys. Rev. D 550, 5262 (1994).
32. S. J. Brodsky and G. R. Farrar, Phys. Rev. D 11, 1309 (1975).
33. G. Bertsch, S. J. Brodsky, A. S. Goldhaber, and J.F. Gunion, Phys. Rev. Lett. 47,
297 (1981).
34. S. J. Brodsky and A. H. Mueller, Phys. Lett. B 206, 685 (1988).
35. S. J. Brodsky and G. P. Lepage, in Perturbative Quantum Chromodynamics, edited
by A. Mueller (World Scientific, Singapore, 1989).
36. S. J. Brodsky, G. P. Lepage, and P. Mackenzie, Phys. Rev. D28, 228 (1983).
37. S. J. Brodsky and H. J. Lu, Phys. Rev. D 51, 3652 (1995).
38. C. T. H. Davies, K. Hornbostel, G. P. Lepage, A. Lidsey, J. Shigemitsu, and J. Sloan,
Phys. Lett. B 345, 42 (1995).
39. S. J. Brodsky, A. H. Hoang, J. H. Kuhn, and T. Teubner, Phys. Lett. B 359, 355
(1995).
40. S. J. Brodsky, C. R. Ji, H. J. Lu, and A. Pang (in preparation).
41. S. J. Brodsky, L. Frankfurt, J. F. Gunion, A. H. Mueller, and M. Strikman, Phys.
Rev. D 50, 3134 (1994).
42. Adams, et at., Phys. Rev. Lett. 74, 1525 (1995).
43. S. Heppelmann, Nucl. Phys. B (Proc. Suppl.) 12, 159 (1990), and references therein.
44. B. Blaettel, G. Baym, 1.1. Frankfurt, H. Heiselberg, and M. Strikman, Phys. Rev.
D 47, 2761 (1993).
45. S. J. Brodsky and B. T. Chertok, Phys. Rev. D 14, 3003 (1976).
46. S. J. Brodsky, C.-R. Ji, and G. P. Lepage, Phys. Rev. Lett. 51, 83 (1983).
47. G. R. Farrar, K. Huleihel, and H. Zhang, Phys. Rev. Lett. 74,650 (1995).
48. A. D. Krisch, Nucl. Phys. B (Proc. Suppl.) 25, 285 (1992).
49. S. J. Brodsky and G. F. de Teramond, Phys. Rev. Lett. 60, 1924 (1988).
50. S. J. Brodsky, G. P. Lepage, and S. F. Tuan, Phys. Rev. Lett. 59, 621 (1987).
51. M. Luke, A. V. Manohar and M. J. Savage, Phys. Lett. B 288, 355 (1992).
52. S. J. Brodsky, G. F. de Teramond, and I. A. Schmidt, Phys. Rev. Lett. 64, 1011
(1990).
53. CDF Collaboration (F. Abe et at.), FERMILAB-PUB-95-271-E, Aug 1995.
54. E. Braaten and S. Fleming, Phys. Rev. Lett. 74,3327 (1995).
55. G. T. Bodwin, E. Braaten, and G. P. Lepage, Phys. Rev. D 51, 1125 (1995).
179
56. S. J. Brodsky, P. Hoyer, C. Peterson, and N. Sakai, Phys. Lett. B 93, 451 (1980);
S. J. Brodsky, C. Peterson, and N. Sakai, Phys. Rev. D 232745 (1981) .
57. S. J. Brodsky and Bo-Qiang Ma, SLAC-PUB 7126 (1996).
58. R. Vogt and S. J. Brodsky, Nucl. Phys. B 438, 261 (1995).
59. S. J. Brodsky, J. C. Collins, S. D. Ellis, J. F. Gunion, and A. H. Mueller, Snowmass
Summer Study 1984:227 (QCD184:S7:1984).
60. J. J. Aubert, et al., Nucl. Phys. B 123, 1 (1983).
61. S. J. Brodsky, W.-K. Tang, and P. Hoyer, Phys. Rev. D52, 6285 (1995).
62. J. Badier, et al., Z. Phys. C 20, 1010 (1983).
63. C. Biino, et al., Phys. Rev. Lett. 58,2523 (1987).
64. B. W. Harris, J. Smith, and R. Vogt, LBL-37266, (1995), hep-ph/9508403.
65. J. Badier, et al., Phys. Lett B 114,457 (1982), ibid. 158, 85 (1985).
66. R. E. Ecclestone and D. M. Scott, Phys. Lett. B 120, 237 (1983).
67. V. G. Kartvelishvili and S. E'sakiya, Sov. J. Nucl. Phys. 38(3),430 (1983) [Yad.
Fiz. 38, 772 (1983)].
68. B. A. Li and K. F. Liu, Phys. Rev. D 29,426 (1984).
69. V. Barger, F. Halzen, and W. Y. Keung, Phys. Lett. B 119, 453 (1983).
70. See J. D. Bjorken, J. B. Kogut, and D. E. Soper, Phys. Rev. D3, 1382 (1971), and
references therein.
71. A summary of light-cone perturbative-theory calculation rules for gauge theories is
given by Lepage and Brodsky16. We follow the notation of this reference here.
72. This is the light-cone analog of the infinite-momentum frame introduced in S. D.
Drell. D. J. Levy, and T. M. Van, Phys. Rev. Lett. 22, 744 (1969). See also S. J.
Brodsky, F. E. Close, and J. F. Gunion, Phys. Rev. D6, 177 (1972).
73. The polarization of each vector-boson constituent is specified by the helicity index
Sj in ' as in Eq. (22).
~
74. We use momentum conservation to eliminate the dependence of ' on k J.j, where
j is the struck quark. Note that the results (15), (16), and (20) are independent of
the charge of the constituents.
75. Related calculations in the infinite-momentum frame are given in S. J. Chang and
s. K. Ma, Phys. Rev. 180, 1506 (1969); Bjorken, et al., Ref.1D; D. Foerster, Ph.D.
Thesis, University of Sussex, 1972; and S. J. Brodsky, R. Roskies, and R. Suaya,
Phys. Rev. D8, 4574 (1973). The infinite-momentum-frame calculation of the order-
0;2 contribution to the anomalous moment of the electron is also given in the last
reference.
76. S. J. Brodsky, T. Huang, and G. P. Lepage, SLAC-PUB-2540 (unpublished).
77. Parton-model expression for other definitions of the charge radius are given in F.
E. Close, F. Halzen, and D. M. Scott, Phys. Lett. 68B, 447 (1977).
78. For example, the contribution to the nucleon anomalous moment in the quark model
gives a contribution Sa ex gMN/ (ki) if there is a g(j>3 trilinear coupling of scalars.
It is thus possible to obtain a contribution to the anomalous moment of a fermion
which is linear in its mass even if all of its consituent fermions are massless.
79. A better estimate is R2 = (Srt, where S - Li':l[(ki + m2 )/x]i.
80. S. J. Brodsky and J. R. Hiller, Phys. Rev. C 28, 475 (1983).
81. S. J. Brodsky and J. R. Primack, Ann. Phys. 52 315 (1969); Phys. Rev. 174,2071
(1968).
82. S. J. Brodsky and F. Schlumpf, Phys. Lett. B 329, 111 (1994).
83. S. B. Gerasimov, Yad. Fiz. 2, 598 (1965) [Sov. J. Nucl. Phys. 2, 430 (1966)].
84. S. D. Drell and A. C. Hearn, Phys. Rev. Lett. 16, 908 (1966).
85. B.-Q. Ma, Phys. Rev. C 432821 (1991); Int. J. Mod. Phys. E 1809 (1992) .
86. F. Schlumpf, Phys. Rev. D 47, 4114 (1993); Mod. Phys. Lett. A 8, 2135 (1993);
Phys. Rev. D 48,4478 (1993); J. Phys. G 20,237 (1994).
87. E. Wigner, Ann. Math. 40, 149 (1939).
ISO
88. H. J. Melosh, Phys. Rev. D 9, 1095 (1974); L. A. Kondratyuk and M. V. Terent'ev,
Yad. Fiz. 31, 1087 (1980) [Sov. J. Nucl. Phys. 31, 561 (1980)]; D. V. Ahluwalia and
M. Sawicki, Phys. Rev. D 47, 5161 (1993).
89. L. L. Frankfurt and M. 1. Strikman, Nucl. Phys. B 148, 107 (1979), Phys. Rep. 76,
215 (1981); 1. A. Kondratyuk and M. 1. Strikman, Nucl. Phys. A 426, 575 (1984);
L. 1. Frankfurt, T. Frederico, and M. Strikman, Phys. Rev. C 48, 2182 (1993).
90. F. Coester and W. N. Polyzou, Phys. Rev. D 26, 1349 (1982); P. L. Chung, F. Co-
ester, B. D. Keister and W. N. Polyzou, Phys. Rev. C 37, 2000 (1988).
91. H. Leutwyler and J. Stern, Ann. Phys. 112, 94 (1978).
92. P.1. Chung and F. Coester, Phys. Rev. D 44, 229 (1991).
93. B.-Q. Ma, J. Phys. G. 17, L53 (1991); B.-Q. Ma and Qi-Ren Zhang, Z. Phys. C 58,
479 (1993).
94. Particle Data Group, Phys. Rev. D 45 (Part 2), 1 (1992).
95. G. Karl, Phys. Rev. D 45, 247 (1992).
96. H. Fritzsch, Mod. Phys. Lett. A 5, 625 (1990).
97. R. 1. Jaffe and A. Manohar, Nucl. Phys. B 337, 509 (1990).
98. A. V. Efremov and O. V. Teryaev, Proceedings of the International Symposium on
Hadron Interactions (Bechyne), eds. J. Fischer, P. Kolar and V. Kundrat (Prague),
302 (1988); G. Altarelli and G. G. Ross, Phys. Lett. B 212, 391 (1988); R. D. Carlitz,
J. C. Collins and A. H. Mueller, Phys. Lett. B 214, 229 (1988).
99. J. Ellis and M. Karliner, Phys. Lett. B 341, 397 (1995).
181
DISCRETIZED LIGHT-CONE QUANTIZATION
Hans-Christian Pauli
Max-Planck-Institut fur Kernphysik

D-69117 Heidelberg
ABSTRACT
The method of Discretized Light-Cone Quantization is reviewed in simple terms.

Emphasis is put on how one should define a Hamiltonian, and on periodic boundary
conditions. Some numerical results for one and for three space dimensions are compiled.
The challenges and the virtues of the method are discussed in short.
INTRODUCTION
One of the most important tasks in hadron physics is to calculate the spectrum and
the wavefunctions of the physical particles from a covariant gauge theory. The method
of 'Discretized Light-Cone Quantization' (DLCQ) [55,56] has precisely this goal and has
three important aspects: (1) Rejuvenation of the 'old-fashioned' Hamiltonian approach;
(2) Discretized plane waves, or periodic boundary conditions; (3) Quantization at equal
light-cone time T = Z + t, rather than at equal usual time T = t. It was reviewed in [9]
and faces many advantages and challenges [10]. Similar ideas have been advanced also
by [63, 75].
But here is a problem: The diagonalization of the Hamiltonian in conventional
many-body theory is bestowed with difficulties. How can one dear to address to a field
theory, where not even the particle number is conserved? Let us review the difficulties
in short for a conventional non-relativistic many-body problem. One starts out with a
Hamiltonian H = T +U. The kinetic energy T is a one-body operator and thus simple.
The potential energy U is at least a two-body operator and thus complicated. The
problem is solved if one has found one or several eigenvalues and eigenfunctions of the
Hamiltonian eigenvalue equation HiV = EiV. For to achieve this one can expand the
eigenstates in terms of products of single particle states (xlm) , which usually belong to
a complete set of ortho-normal functions of position X, labeled by a quantum number
m. When anti symmetrized, one refers to them as 'Slater-determinants'. All Slater-
determinants with a fixed particle number form a complete set.
One might proceed thus as follows [53, 54]. In the first step one chooses a complete
set of single particle wave functions, which solve an arbitrary 'single particle Hamilto-
183
o 1 5 2 4 6 8 10 12
o
E/hw
~
0
~
~ 6
~ 2 5.5
~ -r+--r~~~~~----~----~
"0
..,
..~.
~
V35
~
4
3.5
\
~ I.
1.5 , ....-~~-
5 o
o 2 6 12 20 30 47 M 0 2 6 12 20 m
Figure 1: Non-relativistic many-body theory.
nian'. The choice of the latter is a science of its own. In the second step, one defines one
and only one reference state. Every Slater determinant can be classified relative to this
reference state as a n-particle-n-hole (n-ph) state, but for given A particles they cannot
be more than A-ph excitations. Within the so defined Hilbert space one calculates in
a third step the Hamiltonian matrix. In the last and fourth step one diagonalizes this
matrix by numerical means.
In Figure 1, the Hamiltonian matrix for a two-body interaction is displayed schemat-
ically. Most of the matrix-elements vanish, since a 2-body Hamiltonian changes the
state of at most 2 particles. The structure of the Hamiltonian is therefore a penta-
diagonal bloc matrix with a finite number of blocs. Within a bloc, however, the num-
ber of states is infinite. It is made finite by an artificial cut-off, for example on the
kinetic energy or on the single particle quantum numbers m. Since a finite matrix can
be diagonalized on a computer, the problem becomes numerically soluble. One must
verify at the end, of course, that the physical results are insensitive to the cut-offs.
This procedure was actually carried out in one space dimension [54], with the two
sets of single-particle functions
They are the suitably normalized eigenfunctions of the harmonic oscillator with its
Hermite polynomials Hm , and of a free particle with periodic boundary conditions, re-
spectively. Both depend parametrically on a characteristic length parameter L, which
in the oscillator case is L == h/rnw. The calculations are particularly easy for particle
number A = 2, and for a harmonic two-body interaction. The respective results are
displayed in Figure 1. They are surprisingly different. For the discretized plane waves,
the results converge very rapidly to the exact eigenvalues E = ~,~, , ... ,
as shown
in the right part of the figure. Opposed to this, the results with the oscillator states
converge only slowly. Obviously, the larger part of the Slater determinants with the
localized Hermite polynomials is wasted for building up the translationally invariant
solutions, as they are required by Galilean invariance. It is obvious, that the plane
waves are superior, since they have that symmetry implemented from the outset. The
approach with discretized plane waves was succesfully applied to get the exact eigen-
184
values and eigenfunctions for up to 30 particles in one space dimension [54]. From these
calculations, one may conclude: Discretized plane waves
are a manifestly useful tool for many-body problems;
are a complete and denumerable set of states;
allow to construct a Hamiltonian matrix in momentum space;
generate good wavefunctions even for a confining potential.
These features make them an almost ideal tool for the application in gauge field theory.
HOW TO CONSTRUCT THE HAMILTONIAN
In non-relativistic quantum mechanics, the Hamiltonian is that operator which

propagates the system in time, i.e. iftl<I = HI<I. In a covariant theory, however, the
concept of time must be generalized, since the space-time parametrization is arbitrary.
According to Dirac [20], see also [21, 22], one can define the generalized Hamitonian as
that operator which propagates the system in generalized time xo, i.e. ifxol<I = Pol<I.
There are no more than three standard forms how to choose generalized time and the
corresponding generalized coordinates: the 'instant', the 'front' and the 'point' form.
Their essentials are collected in Fig.2. All other parametrizations can be reduced by a
Lorentz transform to one of them. As a historical irony, the action-oriented Feynman
Approach was formulated at about the same time as Dirac's Forms of Hamiltonian
Dynamics. After that, Hamiltonian approaches became old-fashioned and have been
abandoned. Dirac's approach was almost forgotten. The front form was rediscovered
independently several times, and nowadays is known under such different names as
'null-plane quantization', 'infinite-momentum frame', or 'light-cone quantization', see
also [9].
How can one find such a generalized Hamiltonian? The Hamiltonian is one of the
constants of motion as it emerges from the variation of the Lagrangian density operator
in canonical field theory [5]. The Lagrangian density for example for Quantum Elec-
trodynamics (QED) is given by
(2)
The covariant derivative D" == 0" + igA" and F"V == 0" AV - oVA", the electromagnetic
field tensors, are used as abbreviations. As usual, one works in units where "h = c = l.
The coupling constant 9 is dimensionless, the fermion mass is m. Here and in the
sequel, the raising and lowering of the Lorentz indices by means of the metric tensor
will be applied consistently, i.e. x" = gw'x''', The four Dirac components of the fermion
fields w"', their hermitean conjugates wl and the four Lorentz components of the vector
potentials A" are denoted collectively by cPr (with r = 1,2, ... ,12). All of them are
operators, one but their operator properties need be not known at this point. When
they are varied independently from the derivatives o"cPr, one gets the operator equations
of motion,
o"F"V = gr and (i'y"D" - m)w = 0, (3)
i.e. the Maxwell and the Dirac equations, respectively. The current J>' = bC/bA>..
becomes J>" == w'yAw. The variational calculus applied to the Lagrangian generates
also the energy-momentum stress tensor
185
y
The instant form The front form The point form
zO = cI zO = cI + > zO = -r , d=Tcoshw
%1 = .z: i 1 =% i 1 ;;;;;; W
1 Z = .,.sinhwsin~cos
%2 = Y %2 = Y %2 = 8, Y = ~sinhwsin8sinq.
%,3 =z i 3 = ct. - z z:3 = , z = 'TsinhwC0:I59
_
g"" =
00
( 0
0- 1 00 ~)
0 -1 0
0 _= (~ _~2
g"" 00 0
~
_ ~2sinh2w
~)
0
~ 0 0 0 o 0 _T 2 sinh2 '" sin 2 8
Figure 2: The three forms of Hamiltonian dynamics.
Manifest gauge-invariance is obtained with an operator identity obtained from the

Maxwell equations [58], namely FIJ.K.8VAK. = FIJ.K. FVK. + JIJ. AV + 8K.[FIJ.K. AV]. Its last term
can be dropped when inserted into Eq.(4), since total derivatives do not matter in
a variational calculus. The vanishing four-divergence of the energy-momentum stress
tensor is closely related to the existence of the ten operator constants of motion pv
and JIJ.V of the Lorentz group [5, 14, 15]. In the sequel we shall deal only with the
energy-momentum four-vector pv.

Since one has to integrate 81J. TlJ.v = over the space-like coordinates to obtain the
time-derivative of energy momentum, one cannot specify pv without explicit reference
to the generalized coordinates. For the instant form, the space-like coordinates are the
usual coordinates i = (x, y, z) and the time-like coordinate is the usual time Xo = t.
Energy-momentum is thus pv = fnd3i TDv. The normalization volume n refers to
the area of integration and to the boundary conditions, see also below. Its space-like
components P do not depend on the interaction. In Dirac's terminology [20] they are
simple operators and thus momenta. Its time-like component pO = Po depends on the
interaction. In Dirac's terminology it is a complicated operator and thus a Hamiltonian.
It propagates the system in time, i.e. i8ol<I = Pol<I, and is that operator which we
were looking for above.
In the front form, the time-like coordinate is chosen as Xo == x+ = t + z, and
the space-like coordinates as i = (x, y, t - z) == (iJ., x-). The four-vector of energy-
momentum on the light cone becomes then
It is independent of x+. The factor ~ is emerging from the Jacobian. Its space-like
components PJ. = (PI, p 2 ) and P+ do not depend on the interaction. In Dirac's termi-
nology there are momenta. Its time-like component P- depends on the interaction, is
complicated and thus a Hamiltonian. It propagates initial data in the light-cone time
x+, i.e. i8+1<I = P+I<I. One notes that the instant and the front form are quite
similar from the formal point of view. Why the one should be superior to the other
cannot be seen in general and has to be discussed below.
186
Quantum Chromodynamics (QCD) proceeds analogously. One gets QCD by re-
placing the QED vector potentials AJ.' by 3 x 3 hermitean and traceless matrices Ai'" Any
such matrix can be expressed as a sum over the eight generating matrices T~d and the
operator functions A~, i.e. A~cl = L:a T:dA~. These matrices obey the commutations
relations [Ta, T b ] =ijabcTc wi~h the str~cture constants rbc [9]. Correspondingly, the
Dirac spinors get an additional color index c, i.e. \lia,c with c = 1,2,3. As usual dou-
bly occuring color or glue indices without emphasis on reference to raised or lowered
positions will be summed over. Imposing gauge invariance takes care of the rest and
the Lagrangian density becomes for QCD
c = -2"TrFJ.'
1 vF J.'v + 2"1 [-
\Ii(i,J.'DJ.' - mF)\Ii + h.c. ] (6)
Both the convective derivative DJ.' == OJ.' + igAJ.' and the color-electromagnetic field
FJ.'V = oJ.' AV - OV AJ.' + ig [AJ.', AV] are now matrices. The color-Maxwell and the color-
Dirac equations become
(7)
with the current J>. = 8C/8A~ = WTa,>.\Ii + rbco>.A~A~. Correspondingly one gets for
energy-momentum on the light cone
THE HAMILTONIAN IN DISCRETIZED MOMENTUM SPACE
Energy-momentum pv is a constant of motion only for properly defined boundary

conditions. The requirement that all fields vanish at infinity is insufficiently concise,
and difficult to realize with the plane wave states into which one usually expands the
fields. Possible terms from the boundary vanish however strictly for periodic bound-
ary conditions. They can be realized by periodic boundary conditions on the vector
potentials AJ.' and anti-periodic boundary conditions on the spinor fields, since C is
bilinear in the \Ii a. A plane wave state eip,.x" is characterized by the single particle
four-momentum pJ.' = (p+,iIL,p-) with p- = (m 2 + pfJ/p+. Each single particle is
on its mass-shell p2 = m 2 . The boundary conditions are thus satisfied by discretized
momenta, hence Discretized Light-Cone Quantization, with
{ ~ nil, with nil = ~, ~, ... ,00 for fermions,

"Lnll, with nil = 1,2, ... ,00 for bosons,
andtiJ.. ;.L n.L, with nx,ny = O,1,2, ... ,oo for both. (9)
The two artificial length parameters Land L.L also define the normalization volume
n == 2L(2L.L)2. More explicitely, the free fields are expanded as
~a(X) = ~ ~ ~ (bqua(p, ),)eiPX + d!va(p, ),)e-;px) ,

and AJ.'(x) = .In ~ ~ (aqfJ.'(p, ).)eipX + a~f~(p, ).)e- iPX ) , (10)
187
0 1 2 3 S 4 8 9 7 10 14
n Sector 9 qq gg qqg ggg qlIqq qtlgg gggg ~qqg qIIgg, ggggl
0 9 t -<-< ~'1(
1 qq
>- I I -< t;~
2 gg
-< IX -<-< ~'1(
3 qqg ;r >-->- I I -<-< ~t(
5 ggg
T >- IX -<-< ~'1(
4 qqqQ ;t >- I I -<
8 qIIgg ;r;r >->- III -<-<
9 gggg
T >- IX .. -<-<
7 qqqqg ;r >-->- I I
10 qqggg
77 >-->- III
14 ggggg
T >- IX
Figure 3: The Hamiltonian matrix for a meson. Allowing for a maximum parton
number 5, the Fock space can be divided into 11 sectors. Within each sector there are
many Fock states l<1>i). The matrix elements are represented by diagrams, which are
characteristic for each bloc. Note that the figure mixes apects of QeD where the single
gluon is absent and of QED which has no three-photon vertices.
particularly for the two transversal vector potentials Ai == A1, (i = 1,2). The light-cone
gauge and the light-cone Gauss equation, i.e. A+ = 0 and A- = (i~~)2 J+ - (i5+) iOjAi,
respectively, complete the specification of the vector potentials N'. The subtlety of the
missing zero-mode nil = 0 in the expansion of the A-L will be discussed below.
Each plane wave state "q" = (nil, n x , ny, >., c, f) is specified by six quantum num-
bers: the three discrete space-like momenta ni, the helicity A, the color c, and the
flavor f of the particle. The creation and destruction operators like a~ and a q create
and destroy single particle states q, or 'partons', and obey (anti-) commutation rela-
tions like [a q, a!,] = {bq, b!,} = {dq, d~} = Dq,q" They specify the quantum properties of
the theory. In the discretized theory one has simple Kronecker symbols. The spinors
Uo: and Vo:, and the transversal polarization vectors ~ are the usual ones, and can be
found elsewhere [9]. Finally, after expressing all fields in terms of the free field ex-
pansions of Eq.(lO) and performing the space-like integrations, one ends up with the
energy-momenta P" = P"(a q, a~, bq, bt, dq, dt). They are operators acting in Fock space.
Details are given elsewhere [9].
The Hilbert space for the single particle creation and destruction operators is the
Fock space, i. e. the complete set all possible Fock states
l<1>i) = Ni bt bt2 ... btN dt dt2 ... dt,,,, at at2 ... atNIO) , (11)
in analogy to the set of all possible Slater-determinants of section 1. Like there, one
here has one and only one reference state, the Fock-space vacuum 10), annihilated
188
by all destruction operators. As consequence of discretization, the Fock states are
denumerable by a running index i = 1,2, ... and orthonormal, i.e. (ilj) = Dij, with
a suitable normalization constant N i . Note that they can be made color-singlets.
As illustrated in Figure 3 for the case of total charge zero, the Fock space can
be structured into classes according to the parton number N + N + N. Within each
class one can further divide it into sectors according to the number of qq-pairs. The
sectors are denumerated by another running number n in an arbitrary way. Since all
components of the energy momentum commute with each other, and since the space-
like momenta are diagonal in momentum representation, all Fock states must have the
same value of P+ = LvP; and A = LvCPl.)v, with the sums running over all partons
1/ E n in a particular Fock-space sector. Since P+ is a positive operator with a positive
eigenvalue it is convenient [55, 56] to introduce the harmonic resolution K = 2LP+ /,rr.
For any fixed p+ and thus for any fixed resolution, the number of Fock-space sectors is
limited since each parton has a lowest possible value of p+, i. e. 7r / L for quarks and 27r / L
for gluons. As a consequence, any Hamiltonian matrix in DLCQ has a finite number of
blocs as illustrated in Figure 3. However, within each bloc the number of Fock states is
unlimited, since the transversal momenta can combine in infinitly many ways to satisfy
a given total transversal momentum. Therefore the space must be regularized.
Since the Fock states are denumerable, one can associate a matrix with the light-
cone energy operator P- (a q , a~, bq , b~, dq , d~). At each entry of rows and columns sits a
number, the matrix element, which is either zero, or real, or complex. These matrix
elements are compiled elsewhere [9] and correspond to different parts of the interac-
tion. In Figure 3 the interactions are representated by graphs, which are energy but
not Feynman diagrams. The interaction conserves three- but not four-momentum.
The vertex interaction connects states which differ by one parton. The remainder are
the instantaneous seagull and the fork interactions. They are gauge-artefacts. The
seagull interaction acts only between states with the same parton number, and the
fork interaction only between states which differ by two partons. When the parton
number differs by more than two, the matrix elements are strictly zero. From the
outset, the Hamiltonian matrix has thus a penta-diagonal bloc structure similar to a
non-relativistic Hamiltonian with two-body interactions, see Figure 1. The latter is a
remarkable simplification as compared to the conventional quantization. In the instant
form, the Hamiltonian changes the states of up to four particles. Consequently, the
Hamiltonian matrix has there a nano-diagonal bloc structure but with infinitely many
blocs.
As a component of a four-vector, the light-cone energy P- changes from one
Lorentz frame to another. The contraction of the four-momentum PI' PI" however,
is a Lorentz scalar. It will be referred to somewhat improperly as the 'light-cone
Hamiltonian' operator H. Its matrix elements and eigenvalues have the dimension
of an invariant mass squared. Writing it out in components, H = P+ P- - Pi, one
realizes that the diagonalization of P- and of H are completely equivalent since P+
and Pl. are diagonal and can be replaced by their eigenvalues. Based on the boost
properties of light-cone variables [7, 8, 50, 51]' one preferably introduces intrinsic par-
ticle momenta x and kl.. They are frame-independent and defined by Xv = p; / P+ and
(Pl.)v = (kl.)v+xvA, subject to the constraints LvXv = 1 and Lv (kl.) v = O. The latter
sums run over all single particlre states in a Fock state. Expressed in these intrinsic
variables, the light-cone hamiltonian becomes simply H = P+ P-.
For the free theory (g = 0), the total four-momentum is diagonal, P; = LvrJ:. Its
contraction (PI )2 is the free invariant mass-squared of a Fock state, which plays the
same role in DLCQ as the kinetic energy in non-relativistic quantum mechanics. This
analogy applies also in regulating the Fock space. One admits a Fock state only when
189
its kinetic energy is below a certain cut-off [9], that is if
L
vEn
(m 2
X
-2
+ kJ..) :s A~
v
. (12)
Apart from a sector-dependent mass scale An this is is nothing but the familiar Brodsky-
Lepage cut-off [7, 8, 50, 51]. Since only Lorentz scalars appear, the regularization is
Lorentz-invariant. Now, finally, the Hamiltonian matrix is finite, but possibly large. A
finite matrix can be submitted to a computer and diagonalized numerically.
THE EFFECTIVE HAMILTONIAN
The goal of DLCQ is to find the eigenstates and the eigenvalues of the light-cone
Hamiltonian,
HIll!) = EIIl!) (13)
In terms of the sectors In) like above, or 'projectors' In) (nl, this can be written
n
L (iIHnlJ) (jIll!) = E(illl!) , for i = 1,2, ... ,n . (14)
j=l
More explicitly one faces a set of N coupled bloc matrix equations
(
(lIHnll)
(2I HnI 1)
(1I HnI2)
(2I H nI 2) (lI Hnln)
(2IHnln) J (.(1(211l!)
11l!) J ((1(2 Ill!)
Ill!) J
(15)
... =E ...

(nIHnI 1) (nIHnln) (nlll!) (nlll!)
The index keeps track notationally of the n blocs in the Fock space. In the continuum
limit this set of coupled matrix equations converts to a set of infinitly many coupled
integral equations, which are rather cumbersome to write down.
In simple cases, particularly in 1+ 1 dimension, one can diagonalize straightfor-
wardly the above matrix, even for reasonably large values of the harmonic resolution
K. The eigenvalues do not change dramatically when one truncates the Fock-space.
But truncation cannot be the general procedure to cope with large matrix dimensions
since in principle this violates gauge invariance. Although chemists diagonalize nowa-
days matrices with dimensions 50 millions or more routinely, DLCQ applied to gauge
theory faces a formidable matrix diagonalization problem. Here is the bottle-neck of
the method. One has to develop new tools by matter of principle.
Intuitively one aims at something like an effective interaction between a quark
and an antiquark, similar to the effective interaction between a negative and a positive
point charge. Effective interactions are a well known tool in many-body physics [52].
Formally, one defines say "an interesting part" and a "non-interesting part" of the
Hilbert space, refered to as the P- and the Q-space, respectively. The division is
though completely arbitrary. Accounting for the impact of the latter on the former
gives rise to an effective interaction in the P-space alone. Among field theorists the
method is better known as the TD-Approach, since Tamm [72] and Dancoff [18] were
the first ones to apply it to a field theory. It deserves a closer look.
The problem of diagonalizing a finite matrix can always be written in terms of bloc
matrices like
( (PIHIP) (PIHIQ)) ((PIIl!)) ((PIIl!)) (16)

(QIHIP) (QIHIQ) (QIIl!) = E (QIIl!) .
190
The second of these coupled equation can be rewritten as (QIE - HIQ)(QI'lI) =
(QIHIP) (PI'lI) If one could invert the matrix in the Q-sector, (QIE - HIQ), one
could express the Q-space wavefunction in terms of the P-space wavefunction. Since
the Q-space matrix is quadratic, finite and hermitian, this is no particular problem on
a computer. But here is the problem: E is the eigenvalue. This number is the goal of
the calculation and unknown at this point. One therefore solves first an other problem.
One defines
1
(QI'lI)w = Gq(w)(QIHIP) (PI'lI) , with GQ(w) = IQ) w _ H (QI , (17)
introducing Gq(w) , the resolvent of the Q-space Hamiltonian matrix. The starting
point energy w is at disposal as a free parameter. Inserting this into the first of the
coupled equations Eq.(16), one obtains
(PIHeff(w)IP) = (PIHIP) + (PIHIQ)GQ(w)(QIHIP) . (18)
The effective interaction in the P-space is thus the original matrix plus a part where
the system is scattered virtually into the Q-space, propagating there by impact of the
true interaction in the resolvent, and finally scattered back into the P-space. Since
every value of w defines a different Hamiltonian one gets an w-dependent spectrum:
(19)
By varying w one can generate a set of energy functions Ek(w). Whenever one finds a
solution to the fixpoint equation
(20)
one has found one of the true eigenvalues of H, by construction. Because of the poles
in Ek(w) one can find this way all eigenvalues of H, irrespective of how small the P-
space is choosen. We shall give explicit examples for that, below. This looks like a
complicated procedure, but working with an effective interaction has the advent age
that the resolvent can be approximated systematically. The two resolvents
1
and Go(w) = IQ) w _ T(QI , (21)
are related by the well known identity
GQ(w) = Go+GoUGq(w) = Go + GoUGo+ GoUGoUGo+ GoUGoUGoUGo+ ... , (22)

in practice that is by the infinite series of perturbation theory. Note that the matrix
w - T is diagonal and easily invertible.
The Tamm-Dancoff Approach. Following the original TD-Approach [72, 18] one
identifies the P-space with the qq states and the Q-space with the rest. Moreover, one
restricts to the lowest non-trivial order of perturbation theory. The effective Hamilto-
nian of Eq.(19) becomes then
(23)
In addition, one usually truncates the Hilbert space. One restricts the Q-space to the
lowest sector, thus IQ) = Iqqg). TDA is thus based on several uncontrolled assumptions.
The effective interaction in Eq.(23) has a non-integrable singularity [49, 63, 75], which
is the point of orign for a renormalization procedure in the work of the Wilson group
[63,75].
191
The Modified Tamm-Dancoff Approach. In [49] the simple TDA was modified.
Adding and subtracting an average potential energy (U) in the denominator of Eq.(17),
i.e.
1
GQ(w) = IQ) W _ T _ (U) _ 6U(QI, with 6U =U- (U) , (24)
and restricting to the lowest order in 6U, yields
(25)
With a suitable adjusted w* = w - (U), one removes the singularity [49].

The Method of Iterated Resolvents. The above theory of effective interactions al-
lows however also an other procedure [58, 62]. One can identify the Q-space with the
last sector, thus IQ) = In), and the P-space with the rest. The effective interaction
according to Eq.(I9) acts then in the space of n - 1 dimensions. Keeping track of
the bloc matrix dimension notationally by an index as in Eq.(15), i.e. Hn = Hand
H n- 1 = Heff(W), Eq.(I9) is rewritten as
1
with Gn(w) = In) w _ Hn (nl . (26)
This holds for all (bloc) matrix elements (iIHnlj) and thus as an operator. With this
notation, the effective interaction is in the form of a recursion relation. It applies also
for the reduction from n - 1 to n - 2, and so on, until one arrives at the effective
interaction in the II)-space. A systematic application of this recursion relation, called
the method of iterated resolvents [62], avoids perturbation theory and makes use of the
fact that most of the bloc matrix elements of the Hamiltonian matrix vanish identically,
see Figure 1. Ultimately [62], the effective interaction in the qq space can be written
as a sum of three terms
(27)
which are represented diagrammatically in Figure 4. The vertex couplings are actually
coupling functions which are related eventually to a running coupling constant. The
three terms respectively correspond to a one-gluon exchange-interaction like in Eq.(23) ,
a self-mass term which potentially is related to the constituent quark masses, and a
r."K:II:'f:1
U 1,1 U 1,2
I~.:t1L: I
o
I I I
U 1,3 o 5 (;V 10
Figure 4: The effective interaction in the qq- Figure 5: The energy function E(w).
space.
192
two-gluon-annihilation-interaction, which potentially couples quarks of different flavors.
Much work remains however to be done to clear the details.
The power of the method can be demostranted at hand of a simple and explicit
example. A matrix of dimension 4 can not be diagonalized analytically, contrary to
a 2x2-matrix, for example. Interpreting it as a 4x4-bloc-matrix in the above sense,
with each bloc having subdimension 1, the energy function of the 11}-space according
to Eq.(19) becomes identically
1
H~ (~
~)
11 2 3
E(w) = for (28)
33 3 4
w-2-
55 5
w-4---
w-6
and similar expressions for other matrices. As illustrated in Fig 5, the solutions of the
fixpoint equation E(w) = w agree with the four eigenvalues,...., -1.87, -0.0152, 3.33, and
10.6. Note that a form like Eq.(28) could possibly be useful to diagonalize a tridiagonal
matrix of arbitraly large dimension.
SUCCESFUL APPLICATIONS IN 1+1 AND 3+1 DIMENSIONS
DLCQ was first developed and applied [55, 56] to models in one-space and one-
time dimension. Here one can study the solutions to the eigenvalue equation explicitly
as function of the harmonic resolution K. Numerically they are rather stable, and
...;;
Meson Mass ..,;:~'"
.;;,...::"......
3 ...;,;;,......
.. -;;.~"'"
...;;;::,........."."
2 .. ~.
.~."
.~~
..;;f.,..'
. ...... SU(4)
- - SU(3)
.",.
.~. ._.- SU(2)
/.~
t.ii' Hamer: SU(2) Lattice
1.0 mig 1.5
SU(3) Baryon
,,

\
10 \
qqq
o 0.2 0.4 0.6 0. ). 1.0 0 0.2 0.4 0.6 0.8).. 1.0 qqq qq (10 3 )
5
- ~l{t~~~~~!gm~ ~1~!~fiWJ~~ ~1r~~~~~f~J~ -

00
L . -_ _ _ _ _ _ _ _ _ _----'
0.4 0.6 x 0.8 to
Figure 6: Spectra and wavefunctions in 1+1 dimension.
193
reasonably low values of K lead to comparatively accurate results. Particularly simple
are gauge theories [23,24,38,42]. Neither are there dynamical photons (or gluons), nor
do the fermions carry spin. Only the instantaneous interactions are present. Meanwhile
many other results [4, 12, 17, 19, 31, 32, 38, 39, 70] are available.
By reasons of shortness, Figure 6 collects only a few aspects. On the left the
results of Eller [23] for periodic boundary conditions on the fermion fields are shown, a
work which was repeated recently for anti-periodic boundary conditions [25]. The figure
shows the full mass spectrum of QED in the charge zero sector for K = 16 and all values
of the coupling constant and the fermion mass, parametrized by >.. = (1 + ?r(m/ g)2)-~.
It includes the free case at >.. = 0 (g = 0) and the Schwinger model at >.. = 1 (m = 0).
The eigenvalues Mi are plotted in units where the mass of the lowest 'positronium' has
the numerical value 1. All states with M > 2 are unbound.
Calculations for QCD have been done by Hornbostel [42], see also [37]. The results
are displayed in the right part of Figure 6. The figure also shows how easily one can do
calculations for almost any SU (N) because the Fock states can be made color neutral.
Lattice gauge calculations have thus far only been done for SU(2) [16,30]. Whenever
available, they agree with DLCQ. Finally, as one has the wave functions explicitly, one
can calculate straightforwardly structure functions and the like. On the right of the
figure they are displayed for an SU(3) baryon [42], including its content from the higher
Fock states like qqq qq.
QED and QCD in 1+1 dimension are confining theories. They are therefore taken
often as models for mesons or baryons. Confinement in 1+1 dimension is due to a
linearly rising potential energy. A single particle in such a potential always has a
discrete spectrum. The discreteness comes out most cleanly, when one restricts the
Fock-space by hand to the qq-space, see Figure 6. The 'simple states' can be followed
up also in the left part of the figure inspite of being immersed in a quasi-continuum.
As the figure shows, a confining theory also has a continuum spectrum. With a grain
of salt, this continuum can be interpreted as mesons in relative motion.
Calculations like these could be taken as raw material for models which aim at
describing the decay of a meson into two or more mesons within a well defined theory.
One can imagine to create a meson as a 'simple state' in high excitation. The simple
state mixes with the other eigenstates of the full theory due to a residual interaction,
4
M2
3.98
..... '"1'
'1';'
...
"'I
.jl
'Pi'
"'I'
Ip:' 'PI 'Pi'
3.98 '1';' 'PI 'Pi'
...
'PI
3.94
3.95
3.92
3.9 'St'
',.
'ar
3.9
3.88 L...Jl...-----I_---'-_---1._--1._-L_--'--'
5 10 15 20 N -3 -2 -I 0 2 J 3
z
Figure 7: Stability of positronium spec- Figure 8: Positronium spectrum for
trum for Jz = o. -3 ::; Jz ::; 3.
194
.I .l .I
.I /'
0
"-
<::>.
S
;:l
\
.
s0
\.; ......\ ...... \.: ..... ~\ ......\ .. .
\
...,
~
'00
.'. . 0
0.
'. ......0
... ............... . \
rJ)
s::0
t
Il') Il')
Il')
N ...... N
o ci :;3
C)
s::
o .2<lJ
~t '{ ........;........ ';" .... -
~
...,~
Q)
bo
s::
f .~ ........~ ........j: ..... . U3
,po Oi
0. <lJ
..... ~
;:l
ci b.O
~
w
ci
.... >'>........ \ .......

. \ ....... ~\... ..... .
......>......... \........:-:, .......:.~ ... ..
o o
195
and spreads its strength. This looks as if new qq-pairs are popping out of the vacuum,
as illustrated in the lower left part of Figure 6.
Also other aspects ofreaction theory can be studied now. Hiller [38], for example,
has calculated the total annihilation cross section Ree in 1+1 dimension, with success.
Summarizing the work in 1+ 1 dimensions, one could overstate that there are no
numerical or analytical results which could not be reproduced by DLCQ. In fact, in the
mean time the method has become the standard tool for solving field or string theory
problems in two d;mensions.
Calculations in 3+ 1 dimensions proceed in principle like in 1+ 1 dimension. One
chooses the Langrangian parameters 9 and m, sets a resolution K and a cut-off A,
calculates the Hamiltonian matrix, and diagonalizes it by numerical means. Tang [71]
and later Kaluca [48] have truncated the Fock space to the qq and qq 9 states. Similar
cut-offs have also been used in other 3+1 calculations [1,28,41,76]. Such truncations,
however, violate gauge-invariance and a number of additional measures have to be
taken to render the approach 'perhaps-approximately-gauge-invariant' [71]. Wilson et
al [75] have advocated the TD-approach as displayed in Eq.(23). In the continuum limit,
however, the corresponding integral equation has an essential non-integrable singularity,
which is supposed to be taken care of by a renormalization procedure.
For testing DLCQ with a non-trivial example [49], the spectrum and the wavefunc-
tions of positronium have been calculated for a large coupling constant. Positronium is
usually dealt with by the non-relativistic Coulomb potential. The non-relativistic wave
functions serve as a base for calculating the hyperfine structure perturbatively [6]. One
can not do much else since the hyperfine interaction is so singular, particularly one
can not solve the problem by solving a Hamiltonian. The calculation of positronium
including the hyperfine interaction is thus far from being a trivial problem. Rather can
it be considered as a paradigm for a relativistically correct calculation in 3+1 dimen-
sions. DLCQ avoids the usual problems with the recoil or the reduced mass because
one works in a momentum representation.
The non-integrable singularity is removed by the modified TDA as given in Eq.(25),
yielding an effective light-cone Hamiltonian H = T + U. The effective interaction U,
see also Eq.(24) and Fig.4, scatters an electron with on-shell four-momentum ke and
helicity Se into a state with k~ and s~, and correspondingly the positron with ke and
Se. In the continuum limit it becomes the kernel of an integral equation, which turns
out to be
(29)
The colliniar singularity has cancelled, only the usual integrable Coulomb singularity
remains. It resides in the inverse of the average four-momentum transfer Q2 = -Hl~ +
m. In the present Hamiltonian approach, the four-momentum transfer of the electron
l~ = (k~ - ke)JJ. is in general different from the four-momentum transfer of the positron
l~ = (ke - k~)JJ.. The two become equal for the scattering of free particles, where Q2 is
identical with the Feynman four-momentum transfer q~. The above effective interaction
is manifestly boost- and gauge-invariant.
The calculations [49] for the very large coupling constant a = 0.3 and total angular
momentum projection Jz = 0 have recently been repeated by Ttittmann [73], but with
very much improved technology. The integral equation is approximated by Gaussian
quadratures and the results are studied as function of the number of integration points
N. Figure 7 demonstrates how fast the invariant mass-squared eigenvalues stabilize as
function of N. Plotting the first 100 eigenvalues one recognizes the ionisation threshold
at M2 ,...., 4m 2 , the Bohr spectrum, and even more importantly, the fine and hyperfine
196
structure like the singlet and the triplet. The agreement with Fermi's estimate is
quantitative, particularly for the physical value of the fine structure constant ex =
1/137. In order to verfy this agreement, one needs numerical stability and a precision
of roughly 1O-1l. Compare this precision for example with the accuracy of lattice gauge
calculations.
The effective interaction in Eq. (29) is not manifestly invariant under rotations.
In the front form, one has a different classification scheme than angular momentum
or parity, by necessity, see for example [14, 15]. Nevertheless, by physical reasons, one
expects that the multiplet structure of the eigenvalues known for ex = 1/137 should pre-
vail for larger coupling constants. Does it? Trittmann [73] has solved the positronium
problem also for non-zero values of J z , with the results shown in Figure 8: The eigenval-
ues are degenerate with a high numerical precision. The non-relativistic spectroscopic
identification of the terms is inserted in the figure.
The stability with respect to the cut-off A has also been checked [49, 73], of couse.
In Figure 9 the singlet wavefunction is shown as function of x and Ik.Ll.
Phenomenological work in 3+ 1 dimensions has been the stepchild of light-cone
quantization, despite the fact that the approach is motivated by the empirical needs.
Particularly one lacks the contact to phenomenology beyond the perturbative regime
Table 1: Masses of qq-hadrons compared with all 30 scalar and vector mesons. The
flavor quark masses (in Mev) obtained from the fit are given in column 2. The numbers
give the calculated (measured) mass in MeV.
pO p+ K*+ D*u B*+

768(768) 773(768) 910(892) 2110(2007) 5712(5325)
u 222.8
714(135)
11"0
w K*o D* B*o
782(782) 914(896) 2109(2010) 5709(5325)
d 236.2
658(140) 668(549)
-
11" TJ
cP D*s B*u
s
1019(1019) 2156(2110) 5735( - )
s 427.2
825(494) 831(498) 953(958)
K- K TJ
I
IN B*+
e
3097(3097) 6502( - )
c 1701.3
2079(1865) 2078(1869) 2131 (1969) 3082(2980)
DO D+ D+s TJe
y
9460(9460)
b 5328.2
5701(5278) 5698(5279) 5726(5375) 6495( - ) 9455( - )
B- If Ifs B-e TJb
197
[9]. One needs more work like [11] or [67] which relates formalism and actual experi-
mental numbers, even at the expense of not being rigorous. Since Brodsky will report
on this, I will discuss some other recent work. If one repeats the calculation which
leads to Eq.(29) but for QCD, one gets essentially the same effective interaction except
that the fine structure constant a is replaced by 4a 8 /3 due to the color factors. For
a while it was believed that confinement is hidden in the zero modes to be discussed
below, but this hope has not materialized. The effective interaction U1,1 in Eq.(27) and
Fig. 4 suggests a running coupling constant, which however has not yet been worked
out. Instead of, to reconcile asymptotic freedom with a confining potential, one can
try [61] to parametrize it like a 8 ( Q2) = 12n /27In(1 + Q2/1\:2), and to solve the effective
Hamiltonian almost analytically by variational methods. The parameter I\: = 193 MeV
is determined by the experimental value of as at the Z-mass, and the five flavor quark
masses are fitted to the five vector mesons qq. They exhaust all freedom to calculate
the masses of the remaining 25 vector and scalar mesons. They are compiled in Table 1
and compared with the empirical values [61]. The agreement is remarkable, but as
usual one fails to describe the pions.
ZERO MODES AND COLLINEAR MODELS
The field expansions of Eq.(lO) have no zero modes, no components with nil = 0,
since they are omitted by hand in Eq.(9). Contrary to the original expectations, the
zero momentum modes of the Lagrangian field operators have proved far more than just
a "set of measure zero". Indeed, in the 4 theory in 1+ 1 dimensions they are crucial
in reproducing the vacuum properties of the theory [33, 34, 66], namely spontaneous
symmetry breaking and a vacuum condensate [3, 40, 64]. In the DLCQ-framework this
is achieved by the I. roperty of the zero modes to be no dynamical degrees of freedom,
but subjects to satisfy a constraint equation of all other modes. This preserves the
simplicity of the vacuum and thus the partonic picture of light-cone field theory. The
symmetry is broken explicitly only, when the solution of the constraint equation is
substituted into the Hamiltonian [3, 64, 40].
The four-point coupling of gluons suggests that at least some aspect of this feature
prevails also in full QCD. In a sequence of papers [43, 44, 45, 46, 47, 59] the constraint
problem was disentangled from that of the gauge-symmetry of the non-Abelian theory.
Similar problems occur even in the Abelian case [43, 44, 45]. Not all of the various
zero modes are constrained. Some of them are intimately related to the topology
of the hyper-torus implicit in DLCQ and become important in the presence of non-
Abelian gauge groups [46]. DLCQ is particularly suited for systematically exploring
lower dimensional manifolds embedded in higher dimensional theories because of the
periodic boundary conditions: zero and non-zero momentum modes can be cleanly
distinguished. One thus obtains effective theories in lower space-time dimensions that
are not identical to the original theory. In the essence, this is dimensional reduction
[68,69]. Dimensionally reduced models have been introduced to DLCQ by Klebanov et
al [4,17, 19] but without zero modes and assuming only color singlet string states. More
on all of that can be found in the contributions to these proceedings by Kalloniatis,
Robertson and Vandesande.
In [46], a 1+ 1 dimensional pure SU (2) gauge theory coupled to external sources
was examined. Suppressing all momentum excitations, one obtains a 0+ 1 dimensional
thus quantum mechanical theory of a single gluon mode. Formally, it is one of the
components of A +, which arises in the light-cone Coulomb gauge {LA + = 0 and which
likes to appear in the dimensionless combination z = At g L /n. It corresponds to a
198
0.4
0.2
20 40 60 K 80
Figure lO: Glueball Spectrum.
2 I I I I
,
M
GeV
.. - .... ....
--
1.5 - -r-ao fo
. i ... 1T --71 ..... l.. ...
1 - _ ao - fa
0.5 ~
- -"I
0
--1T
0 0
0-+ 0++ JPC 0-+ 0++ JPC
Figure 11: 16 Mesons in the collinear model.
quantized flux loop around the circle defining space. The mode is of purely topological
origin, and thus this model is manifestly isomorphic to a quantization in the instant
form defined on the analogous topology [35, 36J. In [59], one proceeds with a pure
SU(2) gauge theory in 2+1 dimensions. Its dimensionally reduced version looks like a
1+1 dimensional SU (2) gauge theory coupled to adjoint matter. The topological mode
appears here again, but now coupled nontrially to the Pock modes of the transverse
gluon field, structures that were foreseen already by Franke et al [26, 27]. These equa-
tions [59J are very difficult to solve. One therefore proceeds stepwise. On the one hand,
one can restrict to the topologinal mode and omit the Fock excitations. This leads
to a Schrodinger equation, (-d? /dz 2 + Vo(z)) fn(z) = fnfn(z) , where "the potential
energy" Vo(z) is the Fock space vacuum expectation value of P-, and where fn(z) can
be viewed as the wave functions of the vacuum sector. On the other hand, one can
freeze the topological mode at z = ~ and diagonalize the Fock space part of P- [60J.
Truncating to the two gluon sector, one gets a discrete glue ball spectrum for P+ p-
as displayed in Fig. 10. Sine the coupling constant g appears only in the combination
9 = g/2LJ.., the glue ball masses are proportional to mo = 259.)2/rr [60]. The true
eigenstates of the model [59] could then be obtained by diagonalizing P- in the product
space of such states, but preliminary estimates lead to small changes of the spectrum,
only.
199
The low excitations of the glueball spectrum of Fig. 10 are almost identical with
the results of [13], despite working in the light-cone gauge A + = 0 for else the same
model as [59]. They disregard the topological mode right from the outset. It is thus
questionable whether the topological mode has strong impact on the spectral aspects
of the theory. This paves way to the collinear model in which full QCD in the light-
cone gauge is dimensionally reduced to 1+1 dimensions. Formally, one gets it by
requiring the transverse derivatives of all field components to vanish, i.e. (hA~ = 0
and 8.L'l/Jo = O. The field lines are then all parallel, and the model becomes a field
theoretical realization of the phenological flux tube models with built-in confinement.
Technically, one gets the model by setting k.L = 0 in the matrix elements of p- compiled
in [9]. The model is sufficiently simple to include flavor explicitly in a DLCQ calculation.
Fixing two free parameters m = mu = md and 9 by the masses of the ry and the ry', and
truncating Fock space to qij and 99, one gets [2] for the eight low lying scalar mesons
the results displayed in Figure 11. Despite the calculation being preliminary, one notes
that the pions drop out again.
CONCLUSIONS
It remains notoriously difficult to understand the low-energy regime of Quantum

Chromodynamics in terms of the simplistic but otherwise successful constituent quark
picture. Over the years, the light-cone community [29] has made much progress. De-
tailed calculations display agreement with other methods particularly lattice gauge
theory, to the extent that they are available. Zero modes should not be ignored, since
they can be important carriers of quantum structure. The vacuum is simple, or at
least simplier than in the instant form, which in turn implies that the groundstate of
the free theory is also the grounds tate of the interacting theory. Calculations in one-
space and one-time dimension, like the collinear, dimensionally reduced tube models
have an interest of their own, and are a wonderful playground for answering questions
which otherwise can not be obtained. We have learned that there are physical and non-
physical gauges, even something like a wave function of the vacuum. Renormalization
theory remains a difficult problem within a Hamiltonian approach and worth the effort
of the best. There is an impressive bulk of numerical results, and the contact with
phenomenology through QCD-inspired light-cone approaches is one the rise. The list
of challenges and open problems has diminuished during the years, but a few stumble
stones remain, among them to find
an efficient effective interaction and its renormalization;
the mechanism of chiral symmetry breaking;
the mechanism of confinement.
On the other hand, one has a long list of adventages:
Explicit Heisenberg matrix representation of a gauge-field theory.
Exact symmetries of the Lagrangian and frame-independence.
A trivial vacuum in the light-cone gauge A+ = o.
All particles are physical; no ghosts.
The fermions are treated exactly; no fermion doubling.
Covariant cut-offs; unlike lattice gauge theory.
Full spectrum and eigenfunctions of the theory.
Exact calculation of structure functions and form factors.
Economy: Seconds versus hours and days of computer time.
Thus far one has not met an unsurmountable problem. The prominent goal deserves
another few years of effort. DLCQ is taylored to associate numbers with abstract gauge
200
field theories in a comparatively simple way, and this will contribute undoubtedly to
the progress in understanding these theories in comparing to experiment.
ACKNOWLEDGMENTS
It is a pleasure to thank the organizers of ORBIS SCIENTIAE 1996 for providing

such a stimulating and enjoyable atmosphere at the conference. I also thank Rolf Bayer
and Uwe Trittmann for preparing for me the results presented in Figs. 5, 7, 8, and 11
prior to publication.
*
References
[1] A.M. Annenkova, E.V. Prokhatilov and V.A. Franke, Phys.At.Nucl. 56,813 (1993).
[2] R. Bayer, and H.C. Pauli, work in progress, 1996.

[3] C.M. Bender, S.S. Pinsky, B. van de Sande, Phys.Rev. D48, 816 (1993).
[4] G. Bhanot, K. Demeterfi, and I.R. Klebanov, Phys. Rev. D48, 4980 (1993).
[5] J.D. Bjorken and S.D. Drell, Relativistic Quantum Mechanics (McGraw-Hill, New
York, 1964); J.D.Bjorken and S. D. Drell, Relativistic Quantum Fields (McGraw-
Hill ,NewYork, 1965).
[6] S.J. Brodsky, and J.R. Primack, Phys. (N.Y.) 52, 315 (1969).
[7] S.J. Brodsky, R. Roskies and R. Suaya, Phys. Rev. D8, 4574 (1973).
[8] S. J. Brodsky and G. P. Lepage, in Perturbative Quantum Chromodynamics, A.
H. Mueller, Ed., (World Scientific, Singapore, 1989).
[9] S.J. Brodsky and H.C. Pauli, published in Recent Aspect of Quantum Fields, H.
Mitter and H. Gausterer, Eds.; Lecture Notes in Physics, Vol. 396, Springer-Verlag,
Berlin, Heidelberg, 1991.
[10] S.J. Brodsky, G. McCartor, H.C. Pauli, and S.S. Pinsky, Particle World 3, 109
(1993).
[11] S.J. Brodsky and F. Schlumpf, Phys.Lett. B329, 111 (1994).
[12] M. Burkardt, Nucl. Phys. A504, 762 (1989).
[13] M. Burkardt and B. Vandesande, preprint MPI H-V38-1995 (hepth/9510104).
[14] P.L. Chung, W.N. Polyzou, F. Coester and B.D. Keister, Phys. Rev. C37 (1988)
2000.
[15] F. Coester, Null plane Dynamics of Particles and fields, Prog. Nuc. Part. Phys.
29, 1 (1992).
[16] D.P. Crewther and C.J. Hamer, Nucl. Phys. B170, 353 (1980).
[17] S. Dalley and I.R. Klebanov, Phys. Rev D47, 2517 (1993).
201
[18] S. M. Dancoff, Phys. Rev. 78,382 (1950).
[19] K. Demeterfi, LR Klebanov, and G. Bhanot, Nucl. Phys. B418, 15 (1994).
[20] P.A.M. Dirac, Rev. Mod. Phys. 21, 392 (1949).

[21] P.A.M. Dirac, The Principles of Quantum Mechanics, 4th ed., Oxford Univ. Press,
Oxford (1958).
[22] P.A.M. Dirac, in Perturbative Quantum Chromodynamics, D.W. Duke and J.F.
Owens, Eds. (Am. lnst. Phys., NewYork , 1981).
[23] T. Eller, H.C. Pauli, and S.J. Brodsky, Phys. Rev. D35, 1493 (1987).
[24] T. Eller, and H.-C. Pauli, Z. Physik C42, 59 (1989).
[25] S. Elser, Diplomarbeit, U. Heidelberg 1994.

[26] V.A. Franke, Yu.A. Novozhilov, E.V. Prokhvatilov, Lett.Math.Phys. 5, 239,437
(1981).
[27] V.A. Franke, Yu.A. Novozhilov, and E.V. Prokhvatilov, in Dynamical Systems and
Microphysics, Academic Press, 1982, p.389-400.
[28] S. Glazek, A. Harindranath, S. Pinsky, J. Shigemitsu, and K. Wilson, Phys. Rev.

D47, 1599 (1993).
[29] Theory of Hadrons and Light-front QCD, S.D. Glazek, Ed., (World Scientific Pub-
lishing Co., Singapore, 1995).
[30] C.J. Hamer, Nucl. Phys. B121, 159 (1977).

[31J A. Harindranath and J. P. Vary, Phys. Rev. D36, 1141 (1987).
[32] A. Harindranath and J. P. Vary, Phys. Rev. D37,1076 (1988).
[33] T. Heinzl, S. Krusche and E. Werner, Phys. Lett. B256, 55 (1991).
[34] T. Heinzl, S. Krusche, S. Simburgen, and E. Werner, Z. Phys. C56, 415 (1992).
[35] J.E. Hetrick, Nucl.Phys. B30 (1993) 228.
[36] J.E. Hetrick, Int. J. Mod. Phys. A9 (1994) 3153.
[37] M. Heyssler, and A.C. Kalloniatis, Phys. Lett. B354, 453 (1995).
[38] J.R Hiller, Phys. Rev. D43, 2418 (1991).
[39] J.R Hiller, Phys. Rev. D44, 2504 (1991).
[40] J. Hiller, S.S. Pinsky and B. Vandesande, Phys. Rev. D51, 726 (1995).
[41] L.C.L. Hollenberg, K. Higashijima, RC. Warner, and B.H.J. McKellar, Prog.
Theor. Phys. 87, 3411 (1991).
[42] K. Hornbostel, S. J. Brodsky, and H. C. Pauli, Phys. Rev. D41, 3814 (1990).
[43] A.C. Kalloniatis, and H.C.Pauli, Z.Phys. C60, 255 (1993).
202
[44] A.C. Kalloniatis, and H.C.Pauli, Z.Phys. C63, 161 (1994).
[45] A.C. Kalloniatis, and D.G. Robertson, Phys. Rev. D50, 5262 (1994).
[46] A.C. Kalloniatis, H.C.Pauli, and S.S. Pinsky, Phys. Rev. D50, 6633 (1994).
[47] A.C. Kalloniatis, Heidelberg preprint MPIH-V29-1995.

[48] M. Kaluea and H.C. Pauli, Phys. Rev. D45, 2968 (1992).
[49] M. Krautgartner, H.C. Pauli and F. W61z, Phys. Rev. D45 (1992) 3755.
[50] G.P. Lepage and S.J. Brodsky, Phys. Rev. D22, 2157 (1980).
[51] G. P. Lepage, S. J. Brodsky, T. Huang and P.B.Mackenzie, in Particles and Fields
2, A. Z. Capri and A.N.Kamal, Eds. (Plenum Press, New York, 1983).
[52] P.M. Morse and H. Feshbach, Methods in Theoretical Physics, 2 Vols, Mc Graw-
Hill, New York, N.Y., 1953.
[53] H. C. Pauli, Nucl. Phys. A369, 413 (1981).
[54J H.C. Pauli, Z. Phys. A319 (1984) 303.
[55J H.C. Pauli and S.J. Brodsky, Phys. Rev. D32 (1985) 1993.
[56J H. C. Pauli and S. J. Brodsky, Phys. Rev. D32, 2001 (1985).
[57] H.C. Pauli, Nucl. Phys. A560, 501 (1993).
[58J H.C. Pauli, The Challenge of Discretized Light-Cone Quantization, in: "Quantum
Field Theoretical Aspects of High Energy Physics" , B. Geyer and E. M. Ilgenfritz,
Eds., NaturwissenschaftlichTheoretisches Zentrum der Universitiit Leipzig, 1993.
[59J H.C. Pauli, A.C. Kalloniatis, and S.S. Pinsky, Phys. Rev. D52, 1176 (1995).
[60J H.C. Pauli, and R Bayer, Phys.Rev. D53, 939 (1996).

[61] H.C. Pauli and J. Merkel, Heidelberg preprint MPIH-V45-1995; hep-ph 9512257.
[62] H.C. Pauli, Solving Quantum Chromodynamics by Discretized Light-Cone Quanti-
zation, Heidelberg preprint MPIH-Vl-1996.
[63J RJ. Perry, A. Harindranath, and K.G. Wilson, Phys. Rev. Lett. 65, 2959 (1990).
[64] S.S. Pinsky, and B. Vandesande, Phys. Rev. D49, 2001 (1994).
[65J S.S.Pinsky, and A.C. Kalloniatis, Phys. Lett. B. 365, 225 (1996).
[66J D.G. Robertson, Phys. Rev. D47, 2549 (1993).
[67J F. Schlumpf, J. Phys. G20, 237 (1994); and references therein.
[68J W. Siegel, Phys. Lett. 84B (1979) 193.
[69J W. Siegel, Phys. Lett. 94B (1980) 37.
[70] J.B. Swenson, and J. R Hiller, Phys. Rev. D48, 1774 (1993).
203
[71] A.C. Tang, S.J. Brodsky, and H.C. Pauli, Phys. Rev. D44, 1842 (1991).
[72] I. Tamm, J. Phys. (USSR) 9, 449 (1945).
[73] U. Trittmann and H.C. Pauli, work in progress, 1996.
[74] B. Vandesande and S. Pinsky, Phys. Rev. D46, 5479 (1992).
[75] K.G. Wilson, T. Walhout, A. Harindranath, W.M. Zhang, R.J. Perry, and
S.D. Glazek, Phys. Rev. D49, 6720 (1994); and references therein.
[76] J.J. Wivoda and J.R. Hiller, Phys. Rev. D47, 4647 (1993).
204
POSSIBLE MECHANISM FOR VACUUM DEGENERACY IN YM 2 IN
DLCQ
Alex C. Kalloniatis
Institute for Theoretical Physics III

University of Erlangen-Nuremberg
D-91058 Erlangen, Germany
ABSTRACT
SU(2) Yang-Mills Theory coupled to massive adjoint scalar matter is studied in

(1 +1) dimensions using Discretised Light-Cone Quantisation. The gauge fixing and
quantisation of the theory is briefly discussed. The most difficult problem is the presence
of a constrained zero mode of a matter field component, analogous to structures which
in other theories have given rise to the correct vacuum structure in those contexts. Here
the novelty is the coupling of this problem with a dynamical zero mode coming from
the gluon part of the theory. I develop a. diagrammatic method to solve the constraint
equation. The implementation of the solution in the potential governing the vacuum
dynamics of the independent zero mode leads to a centrifugal barrier. This double well
structure seems consistent with the two-fold vacuum degeneracy to be expected from a
standard 'instanton' analysis. However, the barrier height seems too low still to bring
states into approximate degeneracy. It is suggested that an improper treatment of the
renormalisation is the cause of this.
INTRODUCTION
As has already been discussed in previous lectures in this session, light-cone field
theoretic methods (Weinberg, 1966; Pauli and Brodsky, 1985; Perryet aI., 1990) carry
some promise for solving the problem of the hadron spectrum from Quantum Chromo-
dynamics. The fundamental advantage of Dirac's 'front form' Hamiltonian framework
(Dirac, 1949) is the simple vacuum structure and the (related) positiyity of the mo-
mentum operator. With these one can foresee a picture of hadrons emerging from a
diagonalisation of the light-cone Hamiltonian consistent within the intuitive picture of
the constituent quark model).
However, two issues have hindered the progress of this program: renormalisation
and the vacuum problem. My main concern in the present lecture is the latter problem,
although it will become clear at the end that the two may not be in the end inseparable
205
problems. Why the simple vacuum structure of the Front form should also be a curse
has been discussed in Robertson's contribution to the present volume: quite simply,
the low energy properties of the strong interaction demand some structure in the QCD
vacuum.
A scenario in which vacuum structure and the advantages of the front-form could
sensibly co-exist was recently put forward by Robertson (1993) in the context of some
scalar field theories. In the infra-red regularisation achieved by Discretised Light-Cone
Quantisation (DLCQ) of a given field theory, some zero mode field operators are not
dynamical field quanta, thus preserving vacuum simplicity, but rather satisfy constraint
equations (Maskawa and Yamawaki, 1976). In some models which are known to exhibit
spontaneous symmetry breaking these constraints happen to possess multiple solutions.
The light-cone symmetry breaking paradigm thus associates each of these solutions with
a particular vacuum choice in the instant-form treatment of the same theory.
Turning to QCD, the paradigm appears difficult to apply. One expects B-vacua or
their analogue and spontaneous breaking of chiral symmetry leading to the appearance
of condensates in the nonperturbative vacuum as a consequence of non-trivial gluonic
configurations in the ground state. Constrained zero modes of the gluons indeed occur
in non-Abelian gauge theory in (3+1) dimensions in DLCQ as shown by Franke et al.
(1981) for SU(2) pure glue theory, in a modified light-cone gauge consistent with the
space compactification,
(1)
An additional colour rotation rendering the zero mode of A + diagonal in colour space
was also be performed. It turns out that a constrained zero mode appears only in
the corresponding diagonal component of the transverse gluons, satisfying a linear
constraint equation for the zero mode of the field Ai. Within the paradigm, then, as
it has been developed up till now, 'multiple vacua' still seem difficult torecover.
There is however a new feature which gauge theory exhibits, which does not appear
in the theories in which the paradigm was originally developed: this is the presence
of a dynamical, zero mode. In the case of Franke et al., (1981) it manifests itself as
the surviving piece of the A+ gauge potential. Dynamical zero modes in general mean
the naive argument for vacuum triviality in the front form does not apply. Is it then a
combination of this dynamical with the above constrained zero mode which generates
the desired vacuum structure?
This is what I will give some evidence for in the following, which was originally
presented in (Kalloniatis, 1995). I study SU(2) Yang-Mills theory coupled to scalar
adjoint matter in (1 + 1) dimensions. As shown elsewhere, this theory can be regarded
as the dimensional reduction of pure gauge theory in (2+1) dimensions (Demeterfi et
al., 1993). The approach for the dynamical, or 'gauge" mode which I will take here is
akin to an adiabatic approximation. The dynamics of the gauge mode is then solved for
in the vacuum of the particle sector. In this theory, there is no Z-vacuum degeneracy
due to the Jacobian coming from the nontrivial gauge fixing. Rather, one expects a
two-fold Z2 degeneracy as evident from the global symmetries of the adiabatic potential
which controls the dynamics of the gauge zero mode. I will show that it is through an
interplay between all the lowest modes of the various sectors of the theory that give
a hint to this two-fold vacuum degeneracy. The outstanding problem however will be
aspects of the renormalisation, which I will touch upon at the end.
SCALAR ADJOINT MATTER AND SU(2) YANG-MILLS
The following section is a review of the results of (Pauli et al., 1995). First some
notation: my light-cone conventions will be those of (Kogut and Soper, 1971): x ==
206
(X O x 1 )/V'i. The theory I consider is (1+1) dimensional non-Abelian gauge theory
covariantly coupled to massive scalar adjoint matter
C = Tr( -~F",,8F ",,8 + D"'D", - JL~2) . (2)

The field strength tensor and covariant derivative 0", are respectively defined by
D'" = 8'" + ig[A"',] and F"',8 = 8'" A,8 - 8,8 A '" + ig[A "', A,8] . (3)
As in (Pauli et al., 1995) I represent field matrices in a colour helicity basis:
 = T3cp3 + T+CP+ + T-CP_ . (4)
The equations of motion are
D,8F,8", = gJ'" ,with J" = -i[,D"'], and (0"0" + JL~) = 0. (5)
Note that the 'matter current' JD: is not conserved, 8",J'" -; 0, whereas the total 'gluon
current' J~ = J'" - i [F"',8, A,8] is conserved. I use the light-cone Coulomb gauge fL A + =
0, which preserves the zero mode of A +. Then a single rotation in colour space suffices
to diagonalise the SU(2) colour matrix A + = AjT3. Finally, the diagonal zero mode
of A -(xci) can be gauged away (Kalloniatis et al., 1994) at some fixed light-cone time
xci. For writing the Hamiltonian, it is convenient to choose this time as xci = 0,
the null-plane initial value surface on which we specify the independent fields. The
quantum mode At has a conjugate momentump == 8L/8(At) = 2L8+At and satisfies
the commutation relation
(6)
I shall work in Schrodinger representation for this quantum mechanical degree of free-
dom. In the following it will be useful to invoke the dimensionless combination
z == gA+L
__3_ (7)
There are additional global symmetries which can be seen in terms of this mode. First,
Gribov copies (Gribov, 1978; Singer, 1978) correspond to shifts z --+ z + 2n, nEZ, see
the Appendix. Shifts z --+ z + (2n + 1), n E Z are 'copies' generated by the group of
centre conjugations of SU(2), namely Z2 symmetry. The finite interval 0 < z < 1 is
called the fundamental modular domain, see for example (van Baal, 1992). Two further
symmetries are important: Weyl reflection, z --+ -z and the composition of a reflection
and centre transformation leading to a symmetry of the theory under z --+ (1 - z ). After
selection of the fundamental domain, this last is the only remnant symmetry leading
to symmetry under reflection about z = 1.
It will later be convenient to switch to a
variable ( = (z -1).
The diagonal component of the hermitian colour matrix is CP3. The quantisation,
with the exception of the zero mode /P3 = ao/.;;r:;r, is straightforward. At x+ = 0, I
expand in momentum modes
CP3(X-) = ao
r:o= + 1 ~(
r:o= ~ al WI e-'"I"
LX
-
+ att W/ e +"/"-)
'Lx (8)
V 41r V 41r /=1
where WI = 1/0. Note that I will reserve [, [', [1, ... for the nonzero integer valued
momenta of the real scalar field. The momentum field conjugate to CP3 is 1r 3 = 8-CP3.
The quantum commutation relation at equal x+ for the normal modes is well known
(Maskawa and Yamawaki, 1976). It leads to the Fock commutator [at, at,] = 01,1' (1, l' >
0). Sometimes it will be convenient to write the Kronecker 0/,1' as 8f'. As the zero mode
207
of 71"3 vanishes, the zero mode of 'P is not a degree of freedom, but will turn out to
satisfy the constraint equation hinted at in the introduction.
The off-diagonal components of CP are complex fields with 'P+(x-) = 'P~(x-). The
momentum conjugate to 'P- is 71"- = (8_ + igv )'P+. The other conjugate pair is obtained
simply by hermitian conjugation. In contrast to the method used in (Pinsky and
Kalloniatis, 1996), a way of quantizing this field which is not complicated by the large
gauge transformations (Pauli et al., 1995) is given as follows. The momenta are taken
over half-integers, while maintaining consistency with periodic boundary conditions,
namely
e+imofx-
L ( brn Urn
00 .
v'47f +
11' _ 11' _
'P-(x) = e- myX d~ Vm e+''''Yx ) . (9)

471" 1
m=2'
where um(z) = l/Jm + ( and vm(z) = l/Jm - (. The objects mo and (are functions
of z, defined by mo(z) = (integerpartofz) - !,((z) = z - 1no(z). They satisfy the
relations
mo(z+l) = mo(z)+l, mo(-z) = -mo(z), ((z+l) = ((z), ((-z) = -((z). (10)
The domain interval is now -! !

< ((z) < for all values of z. For the fundamental
domain mo = -!,
but the specific choice no longer matters. The Fock modes then
obey boson commutation relations
(11 )
Finally one notes that a large gauge transformation z -? z + 1 produces only mo -?

mo+ 1 and thus only a change of the overall phase in Eq.(9). Most importantly the Fock
vacuum defined with respect to b", and dm is invariant under these transformations.
As usual for a Coulomb gauge, the A-fields are redundant variables obtained from
implementing Gauss' law strongly. Explicitly, Gauss' law reads
(12)
and the hermitian conjugate of the latter with (Jt)t == J~. The currents are, explicitly,
( 13)
The index s indicates noncommuting operators in this product which must be sym-
metrised in order to preserve hermiticity. The Eqs.(12) are trivially soluble for the
fields A- subject to the following exception. The first of Eq.(12) can be solved only if
the zero mode (Jt)o == Qo on the r.h.s vanishes. This cannot be satisfied as an operator,
but must be used to select out physical states, i.E. Qolphys) == O. In second-quantised
form this gives
00
Qolphys) = L (b~bm - d~dm) Iphys) = 0. (14)

m=1
Thus physical states have equal numbers of "b" and "d" particles.
A DIAGRAMMAR FOR THE CONSTRAINT EQUATION
The origin of the constraint equation for ao has already been discussed in (Pauli et
al., 1995): the zero mode of the diagonal part of Eq.(5), 871"Tr (r 3 (D"'D" + J1~)CP)o/g2 =
208
O. After some algebra this gives
z.( 'P+ 1. J + - 'P- 1. J+ ) - --p~a a = 0 . (15)

(8_ - zgv) - (8_ + zgv) + a,s g2..;;r;
To proceed here I first introduce the dimensionless ratio of boson mass and coupling
_ 47rp~
Pa = -2-' (16)
g
Next I introduce the following notation:

1
2(m + ()
r~mn(() -Vn + -Vn)
[( WI
WI
'U m + -
3 ( WI
U
+ -'Urn)
WI
3]
vn
m
r~mn(() - [( -
WI
-
'U m ) 3
- Un + ( -WI + -'Un) 'Urn3 1 . (17)
'Urn WI 'Un WI
The constraint equation turns out to be

00
L [~n(()(b~bnaa)s + ~n( -()(d~dnaa)s] + Paaa

n=!
L L
00 00
[8~+nr~mn(()(arbmdn + alb~d~) +
1=1 m,n=~
81+mr~mn(()(arb~bn + albmb~) + 81+mr~mn( -()(ard~,dn + aldmd~)l. (18)
One observes that, the dynamically independent composite operators on the left-hand
side of the equation are two-body operators - propagators - while on the right-hand side
they are three-body operators, namely interact.ion vertices. This suggest.s diagrammatic
rules for the various quantities in this expression which are represented in Fig.(I).
Additional rules are as follows:
1. Matrix elements are composed by attaching legs to the right and left of the blob
representing the operator.
2. Closed loops represent a sum over all positive integer or ha.lf-integer moment.a.
3. Detached diagrams represent multiplication of the corresponding expressions.
4. Hermitian conjugation is achieved by reflection of t.he diagram while keeping the

sense of arrows the same.
5. Charge conjugation is achieved by reversing the direction of the arrows.
Evidently, these are quite similar to Feynman rules but describe the building of more
than just S-matrix elements. Here I will use them to rela.te matrix elements of the
constrained zero mode as determined by the constraint equation.
To illustrate the use of the rules, the object
(19)
describes a matrix element of aa multiplied by a propagator factor. Diagrammatically,
this is represented by Fig.(2).
209
Graph I Diagram Rules I
. _,.... . ~ . 6. m {() , b!nID)
'-<---- '
)--<- ' 6. n { - ( ) , d~ ID}
r' ariD)
'~:
r~mn{()
<,
m
n r\"'''(()
"~~ r(mn( _()
0 ao
Figure 1: Diagrammatic Rules for the objects appearing in the constraint equation
Eq.(18). Hermitian conjugation involves reflection through the vertical plane without
changing senses of arrows. Further rules are explained in the text.
Figure 2: Example of use of the diagrammatic rules for the product of a matrix element
and a propagator: ~m()(Dlbmdn ao ariD}.
Next, one can represent matrix elements of the constraint equation in a particu-
lar sector. For example, taking (Dial on the left and btnd~ID) on the right and using
commutators where possible, one obtains an equation which can be represented dia-
grammatically as shown in Fig.(3). The mass term is evidently the fifth term. Similar
diagrammatic equations can be generated by sandwiching the constraint between higher
particle states, and the intuitive principal for their construction is straightforward:
1. On the right hand side, join up the incoming and outgoing lines with the three
available vertices in all possible permutations, with the remaining lines running
through as 'spectators'.
210
2. On the left hand side, assign propagator legs once only to all 'incoming/outgoing'
b- or d-boson legs, attach all possible permutations of tadpoles and loops.
The following features can be observed in Fig.(3) and the original equation Eq.(18).
Firstly, the equation involves matrix elements of the constrained mode in the 5-
body sector (two particles 'in', three 'out'). To get at these one takes matrix elements
of the constraint equation in this higher sector. This in turn involves matrix elements
from the 7-body sector (three 'in', four 'out'). This sequence carries on ad infinitum.
Thus all the particle sectors are related to each other via the constrained mode which is
playing the role of an effective interaction. How complicated is this coupling of sectors?
Firstly, the Fock vacuum does not couple into the hierarchy: the VEV (OlaoIO) = 0
because under charge conjugation aD -+ -aD while the vacuum is even.
Also, the following commutation relations for aD must be satisfied:
[aD, Qol 0
[ao,P+] o.
These can be used to further reduce the number of non-zero matrix elements.
Momentum conservation through the vertices will mean that for many sectors the
right-hand side of the hierarchy of coupled equations will vanish. It is difficult to prove
generally, but the system of linear coupled equations can be shown to be nonsingular
for simple examples. Thus matrix elements of aD which cannot be described in terms
of the basic vertices can be argued to vanish.
Renormalisation of the Constraint
The graphs in Fig.(3) with detached bubbles are logarithmically divergent. This
is the first place renormalisation hits us, though as a warning of what will follow, it
is not the only place. With a cutoff regulator A in the momentum sums these terms
become
A
L (6. n ( 0 + 6. n ( -0 lao '" (In A + finite lao . (20)
n=1/2
Figure 3: The constraint equation Eq.( 18) evaluated between (Olal and bt dtlO) depicted
diagrammatically. Tn 71
211
Figure 4: Identification of terms that connect different particle sectors of the constraint
with an effective operator with insertion ~((). The analogous diagram can be drawn
for the anti clockwise flow of momentum in the loop on the left hand side but where the
insertion is made on the lower d-leg. Namely ~ appears with argument -C.
This divergence can be absorbed into the mass term by a subtraction of the sum for
( = 0, namely with physical mass P apparently to be given by
A
P po + 4 2: ~m(O)
m=~
1
Po + 4[2(JE + InA) + In2 + O(ljA)]. (21)
One can verify that this is the same renormalisation used to remove all cutoff depen-
dence in the two-particle bound state equation in, for example, (Pauli and Bayer, 1996).
We shall see that this will not suffice: that in fact p must still be taken to have further
cutoff dependence. For the moment, we ignore this so that the left hand side of the
original constraint now takes the following form
A
2: [~m(()(b!"bmao+b!"aobm+aob!"bm+(bmaob!,,-ao))+(b ---> d,( ---> -()]+p(()ao (22)
m=~
with
A
p(() = p + 2 L (~m(() + ~m( -() - 2~m(O)) (23)
m=~
being a convergent function in A. In this way, matrix elements of the zero mode ao are
rendered cutoff independent.
Iteration to all orders
One can attempt solving the equation iteratively ~ essentially a weak coupling
expansion. One would observe that to any order in the iteration the three separate
contributions coming from respectively fo(() and f 1 (() always decouple from each
other. Mixing occurs solely via contractions between respectively b with bt and d with
dt . Diagrammatically, one can picture this in terms of precisely the terms in Fig.(3)
that make the constraint operatorially non-trivial, namely the third and fourth terms
which connect the different particle sectors. The way I proceed is to reexpress such
terms as an effective three-body operator with an insertion, ~, on the leg of the same
colour as that in the loop. This is illustrated in Fig.(4). Evidently, no mixing can occur
212
between band d modes once the basic three-point vertices ha,ve been extracted. Thus
the insertion can only arise as the sum to all orders of products of the basic propagator
for a given value of ( as follows:
(24)
Observe that for all but the lowest mode m = 1/2, the propagator satisfies l.6. m (() I < 1,
while for m = 1/2 the propagator will in general be 'large" with the effect of 2: 1 / 2 (()
still larger.
Solution of the Constraint
The above considerations motivate the following operator ansatz for the solution
Thus ao is essentially a three-body operator, where the dressing of the 'bare' vertices will
be absorbed into the, as yet, arbitrary coefficients c!mn((). I insert Eq.(2.5) into Eq.(18)
with the identification of Fig.(4) implemented. It suffices to consider the constraint
in the three-particle space in order to solve for the coefficients. once one builds in
the iteration to all orders in the insertion via Eq.(24). I represent the constraint as
L[aol = R, where Land R denote the left- and right-hand sides of Eq.(18). It suffices
to take matrix elements
(Dial L[aol b~d~ID) (Olal R b~d~ID) ,

(Olb n L[aol b~,arIO) (Olb n R b~,arIO) . (26)
The equations decouple and one can determine the Co and C1 coefficients in terms of
the basic vertices and propagator
,
rf~;(() 'D~!n(() == .6. m(() + .6. n( -() + 2:",(() + 2: n( -() + p((),
'Dm,n(()
r~:;(() ,'D~;n(() == .6. m(() + .6. n(() + 2: m(() + ~n(() + p((). (27)
'Dm,n(() ,
One can now directly give the lowest particle matrix elements as
(OlaL ao b~d~IO)
(OlbMl ao b~2alI0)
(OldNl ao d~2alI0) (28)
Eqs.(25), (27) and (28) are not to be regarded as the final solution. Rather, the method
thus outlined can be the basis for further refinement of the solution by considering
additional types of graph in this procedure.
THE POTENTIAL
The Hamiltonian is the Poincare generator P-, given in the Appendix with other
213
details from (Pauli et al., 1995). After removing a dimensionful factor L(g/47r)2 one
obtains the rescaled Hamiltonian, H. One seeks the ground state of the theory. If not
for the gauge mode (, the naive arguments leading to the trivial Fock vacuum being the
true ground state would hold rigorously. Now in fact the ground state is some infinite
superposition of ( modes, suggesting that an oscillator basis is not so convenient. As
mentioned, I use instead a Schrodinger representation within the adiabatic approach.
The gauge mode ( is frozen for the purposes of computing the ground state in the
'particle sector'. It is here the advantages of the light-cone become significant: the
ground state is now the Fock vacuum. However, the tadpole terms that arise as one
normal orders the Hamiltonian with respect to the trivial vacuum generate (-dependent
structures which become operators once the gauge mode is unfrozen,
H = :H[(]: +V[(]. (29)
Equivalently, V[(] == (OIH[(]IO). As shown in (Pauli et al., 1995) and the Appendix,
there is also a kinetic term for ( surviving in : H:. Thus, projecting in the Fock vacuum
sector, we obtain a Hamiltonian
(30)
For later purposes it is convenient to break V into three terms, coming respectively
from the mass term, the constrained mode ao dependent term and the ao independent
term
V[(] = v~[(] + Va [(] + vo[(] . (31)
Solving the Schrodinger equation corresponding to H( generates the ground state wave
functional Wo( () allowing the introduction of the 'true' ground state as the tensor
product state
In) == Wo[(] 10) . (32)
Vacuum 'Degeneracy'
The only residue of the large gauge symmetries, vVeyl reflections and central con-
jugations in the fundamental modular domain is the symmetry ( -+ -(, which is the
basic symmetry of the potential. Thus there is an associated quantum number which
labels the wavefunctions, namely symmetric W~+)[(] and anti symmetric W~-)[(] states.
Thus identical copies of the particle spectrum can be built on either the vacuum de-
scribed by W~+)[(] or W~-)[(] differing only by a fixed shift in the energy. The shift will
be finite as the longitudinal interval length L (factored out of H( at this point) is kept
finite. This is how the Z2 analogue of {}-vacua arise in this theory. The finiteness of the
shift as L -+ 00 depends on the specific structure of V. In particular, the presence of
an impenetrable barrier in the centre of the fundamental domain would bring the two
lowest energies into exact degeneracy for any interval size L.
Renormalisation of Vp + Vo
The term Va was computed in (Pauli et al., 1995). Here I quote the relevant result:
(33)
214
where A is the (half-integer valued) cutoff in ultra-violet momenta. Using u n ( 0 =
l/Jn + (, vn(O = 1/vn=7" and Wn = l/fo one can check this has the symmetry
( -+ -C. Substituting these coefficients, the expression is
Vo[(] = tt
-1 -1
n- 2 m- 2
((m+O-(n-())2
(m + n)2(m + ()(n - ()
+ ~
1=1
t(
m=~ l(m
(m -I + ~y (m -1- ()2
+ ()(l + m + ()2 + l(m - ()(I + m - ()2
) (34)
Closer inspection reveals the sums to be logarithmic in the cutoff with a coefficient
that depends on (. It is this dependence I seek to extract. A useful trick is to add
and subtract each of the three terms in Eq.(34) evaluated at ( = O. It is then a
straightforward though tedious task to show that the expression can be reorganised as
A 1
Vo[(] = 4G[(] L -m + Convergent[(] + const (35 )
7n=~
with
One recognises in G the convergent functional jJ( () - p, Eq.(23). One sees it comes in
as the operator dependence of the logarithmic divergence.
The contributions of the mass term are more straightforward. After normal order-
ing, one extracts the structure
(37)
where Po was the bare (dimensionless) mass. In order to render the constraint equation
and particle sector convergent and cutoff independent, the renormalised mass p =
po + 2L:;;'=1/2 ~ was introduced, Eq.(21). Substituting this into Eq.(37) gives
A 1
Vp [(] = -4G[(] L:: - + pG[(] + const.
1 lTI
(38)
m=2'
The (-dependent but logarithmic divergent terms in Eq.(35) and Eq.(38) are equal
but opposite in sign. Thus the same renormalisation holds respectively in the three
separate cases of the two-particle bound state equation, the constraint equation, and
in the Vo + Vp parts of the potential.
The shape of this part of the potential is flat at the centre of the fundamental
domain, and divergent at the domain boundaries ( = ~. A careful study of this
singularity as (-+ ~ suggests it is not as strong as 1/((~)Z, thus the boundaries are
penetrable. However, as argued first by Luscher (1983), the wavefunctions should be
multiplied by the Faddeev-Popov determinant for the present gauge which causes them
to vanish rigorously at the boundaries. This suppresses in fact the usual () dependence.
In the absence of the constrained mode there is no structure within the centre of the
fundamental modular region.
IMPACT OF THE CONSTRAINED MODE
I now include the effects of ao. There are terms in the Hamiltonian linear and
215
quadratic in ao, which were separated into HConstr = H21nstr + Hglnst>.. They were
respectively
H(I)
Constr
(2)
HConstr = (39)
where the Q arise from the ao independent parts of the currents and are given in
the Appendix. The other operators are just Bk = (aOb k + bkao)/2 and Dk = (aod k +
dkao)/2 which arise because of symmetrisation of noncommuting operator products. I
correspondingly consider the VEV s of the two terms separately. The linear term leads
to
VI [(, p] (OIHgLstr 10)

~ L O~'+I[ Rlmn( ()( (Olalbmaob~ 10) + (Olbnaoarb~, 10))
l,m,n
+ Rlmn(-O(b-> d)] (40)
with
WI U
Rlmn(o = u~(- - --'-'?:.) (41)
Um WI
which is actually part of the vertex function r 1 (0. The quadratic term gives
(42)
m
One sees that the relevant contributions to the VEV come purely from the lowest
particle sector matrix elements of ao. For V2 this follows after inserting a complete set
of states between the two ao operators.
It is next a tedious task of inserting the solutions as given above. The most
compact expression for the result is
(0 IHConstr 10) = -2i

A+'
I=[((m+02+1(m+l+O)
1=1 m=1 l(m + ()3(m + I + ()4 V~~:m+I(O
((m + O(m + I + O(m -1 + 0 _ 2((m + 02(~ l(m + I + OJ)]

Vm,m+I(O
+[( -+ -(] .
(43)
Evidently, the dominant contributions to the ( dependence here are the lowest modes
with m = !.
The sums are evaluated numerically using Mathematica. The behaviour
of the potential was studied for different values of the physical dimensionless mass p.
As in the part of V independent of ao, the result here has a logarithmic dependence on
the cutoff regulator despite the cutoff independence of ao. At this point I suppress the
divergence by hand, but will return to the nature of this problematic divergence at the
end.
216
Shape of the Constrained Part of the Potential
The final result for the renormalised potential, Va, is represented in Fig. (5) for three
values of the renormalised mass p. In the perturbative region p -+ = the potential
is flat. However, as the p decreases, or coupling increased, the potential develops two
degenerate minima. The barrier height appears to be at its maximum precisely at
p = O. A least square fit gives parabolae for the shape of the individual wells. For
example, for the region 0 < ( < 0.2 the curve can be well described by
v:,(fit)(() = 54.9362 (2 - 11.2670 ( - 0.038.5 . (44)
Taking this together with the coefficient of the kinetic term in Eq.(30), one can estimate
the lowest eigenstate if the given well were the complete potential: Eo '" 14.8 ~ 0.6.
The latter number is the barrier height. Evidently even the kinetic energy of even the
lowest states is larger than the potential barrier height, and the vacuum would have to
be said to be unique.
RENORMALISATION PROBLEM
I now return to the remaining logarithmic divergences which in the previous calcu-
lation were suppressed by hand. The origin of these divergences is simple to see: in the
vertex f 1 (() appears the factor (!!!L
Un
+ :'!:n.).
Wl
This is contracted with the Kronecker delta
function 0;;'+1 in the sums appearing in the potential. Evidently, for fixed momentum
m but [large this factor scales as a constant
. en
11m
1-00
Um+1
( WI
Un
+ un)
WI
-_ ')
~. (45)
Consequently, the large [ behaviour in the sums is not suppressed by a contribution

from this term.
One eventually realises this divergence is the tip of the iceberg when one begins
reinserting the solution to the constraint equation into the Hamiltonian. New many-
body operators will be induced, including two-body operators with coefficients that
diverge logarithmically if the mass p is taken to be cutoff independent. Clearly, the
direction the solution must take is that the constrained zero mode ao should not be
cutoff independent, and indeed there is actually nothing that says it should be: it is
not a 'renormalisation group invariant'. The topological mode ( however is related to
the exponent of the Wilson loop, which presumably is RG invariant. Thus a further
cutoff dependence on p must be invoked, such that the potential is finally rendered
cutoff independent. This work is in progress.
SUMMARY AND CONCLUSIONS
The basic motivation for studying the theory of two-dimensional SU(2) gauge
theory coupled to massive scalar adjoint matter was to understand the relationship
between dynamical and constrained zero modes in a non-Abelian gauge theory and
whether these modes contributed to the expected vacuum structure of that theory.
Based on the homotopy classifications, one expects a two-fold degenerate vacuum for
the theory. There are three conclusions from this study. Firstly, the constrained zero
mode has a structure which can be categorised diagrammatically. This was useful for
217
3
2.5
l.5
0.5
..
-0.5
0 -~
- b
-1
-0.4 -0.2 o 0.2 0.4
Figure 5: Va versus (, namely the contribution of ao to the vacuum potential. The

potential is a functional of ( over the fundamental modular domain -1/2 < ( < 1/2.
The curves are plotted for various values of renormalised mass p: (a) p = 0 (b) p = 20
and (c) p = 100. The potential has two minima whose depth increase with the coupling,
namely with decreasing p. The wells are at their deepest for effectively massless' gluons'.
solving the relevant constraint equation. Secondly, when one takes care to preserve all
the modes of the present theory of massive adjoint scalars coupled to SU(2) glue, there
is some structure in the theory additional to those originally identified by Demeterfi
et al. (1993), consistent with the above homotopy expectations. This I take as a
hint of a mechanism for vacuum structure in light-cone quautisation of non-Abelian
gauge theory. Thirdly, we learn that renormalisation in the light-cone approach is not
separable from the problem of the zero modes.
ACKNOWLEDGEMENTS
Thanks go to the organisers of the ORBIS SCIENTIAE 1996 for providing such a
pleasant atmosphere. This work was supported by a Max-Planck Gesellschaft Stipendium.
APPENDIX
Colour Helicity Basis
The colour helicity basis for SU(2) is defined in terms of the Pauli matrices aa:
3 1 3 _ 1 1 . 2
T =-a T = m(a za). (46)
2 2y2
We can turn this into a vector space by introducing elements x a such that tilde quan-
218
tities are defined with respect to the helicity basis, and untilded the usual Cartesian
basis:
and (47)
The relation between the tilde and untilde basis can be written !J;a = AbXb and x a =
Ab!J;b where A= At. With these elements we can construct the metric in terms of the
tilde basis. Essentially we must demand the invariance of the inner product of any two
vector space elements, xaYa = xaYa . Using the fact that the metric in the a = 1,2 basis
is just the Kronecker delta 8ab and the transformed metric is gab = A~8cdA~. Thus
A-__1 (1
y'2 1 -i
i) and ~ =
gab (0 1)
1 0 . (48)
The metric to raise and lower indices in the helicity basis becomes x = x'f. The colour
algebra looks formally like the Lorentzian structure in light-cone coordinates.
Gribov Copies
Because of the torus geometry of the space and the non-Abelian structure of the
gauge group, there remain large gauge transformations which are still symmetries of
the theory (Gribov, 1978; Singer, 1978) despite the complete fixing of the theory with
respect to small gauge transformations. These are generated by local SU(2) elements
U(x-) = exp( -ino7r x; 73), no an even integer ( 49)
which satisfy periodic boundary conditions. Another symmetry of the theory is Z2

centre symmetry which here means allowing for antiperiodic 11 or alternately no odd.
Both preserve the periodic boundary conditions on the gauge potentials. On the di-
agonal component of A + U generates shifts that are best expressed in terms of the
dimensionless z, namely z -+ z' = z + no. On the scalar adjoint field and its momenta
the effect of the transformation is
i.p3 -+ i.p3 and i.p -+ i.p exp (=FinoIx-)' (50)
7r 3 -+ 7r 3 and 7r -+ 7rexp(inoIx-). (51)
Colour Invariance of z
The gauge mode z can be written in terms of a colour singlet object. Construct
the Wilson line for a contour C along the x direction from - L to L
W = TrPexp(ig r dXJ.lAJ.l) = TrPexp(i9j+L

Jc -L
dxA+). (52)
In the gauge used in this work, this is simply W = Tr exp(2 i z 7r 7 3 ) 2 cos{27r z) .

Thus, modulo the integers, z = 21,,-arcos(.!f) . The integer shifts are the Gribov copies.
Since W is explicitly constructed in terms of a colour trace, z is a colour singlet.
Fourier Transforms and Currents
The Discrete Fourier transforms of the scalar current components are defined by
219
and 1 '"
J+( x-) == - 4L Ok" - -
L.J e-'-YX J;t(k)
ke H
(53)
One can verify that (Ji(k))t = Ji( -k) and (J~(k))t = J:( -k). In the text, as
in (Pauli et al., 1995) the symmetrised operator products Bk == (aob k + bkao)/2 and
Dk == (aOdk + dkao)/2 were introduced. These carry the ao dependence in the currents
J leading to the construction of Q operators
and (54)
The Q-operators are thus ao independent. It suffices here to give the charge operator
Q_ explicitly
n=l m=~
Hamiltonian
The Hamiltonian is obtained from the energy-momentum tensor 81-''' = 2Tr(FI-'I<F1<")-

gl-''' C. At one level it is a simple expression, standard for (1 +1 )-dimensional gauge the-
ories.
to which must be added the mass term fL~ I dx- Tr~2. The abive form merely expresses
the dominance of the Coulomb potential in (1+1) dimensions. However unpacking the
currents J+ leads to nontrivial structure. The length dimension can be factored out
by defining H via P- = L(g/47r rHo In terms of Fourier components, the gauge mode
and Coulomb parts of H can be rewritten as
00
+ L: ut(J!(k)L(k) + L(k)J!(k)) .
k=~
(57)
Relating J operators to the Q operators one can separate the constrained mode ao
parts of H from the Fock sector. The reader is referred to (Pauli et al., 1995) for more
detail. When combined into the invariant mass-squared operator M2 = 2P+ P-, with
P+ the momentum operator, L completely drops out in favour of I< = (L/7r)P+, the
harmonic resolution. The continuum limit L -+ 00 translates into I< -+ 00 which is
easily implement able in computer simulations.
220
REFERENCES
Demeterfi, K., Klebanov, I.R., and Bhanot, G., 1994, Nucl.Phys. B418:15.
Dirac, P.A.M., 1949, Rev.Mod.Phyq. 21:392.
Franke, V.A., Novozhilov, Yu.A., and Prokhvatilov, E.V., 1981, Lett.Math.Phys. 5:239;437.
Gribov, V.N., 1978,Nucl.Phys. B139:1.
Kalloniatis, A.C., Pauli, H.-C., and Pinsky, 8.8., 1994, Phys.Rev. D50:6633.
Kalloniatis, A.C., 1995. On zero modes and the vacuum problem, MPI-H-V29-1995.
Kogut, J.B., and 8oper, D.E., 1971, Phys.Rev. D1:2901.
Liischer, M., 1983, Nucl.Phys. B219:233.
Maskawa, T., and Yamawaki, K., 1976, Prog. Theor.Phys. 56:270.
Pauli, H.-C., and Brodsky, 8.J., 1985, Phys.Rev. D32:1993;2001.
Pauli, H.-C., Kalloniatis, A.C., and Pinsky, S.S., 1995, Phys. Rev. D52:1176.
Pauli, H.-C., and Bayer, R., 1996, Phys.Rev. D53:939.
Perry, R.J., Harindranath A., and Wilson, K.G., 1990, Phys.Rev.Lett. 65:2959.
Pinsky, 8.S., and Kalloniatis, A.C., 1996, Phys.Lett. B365:225.
Robertson, D.G., 1993, Phys.Rev. D47:2549.
Singer, I.M., 1978, Commun.Math.Phys. 60:7.
van Baal, P., 1992, Nucl.Phys. B369:259.
221
THE VACUUM IN LIGHT-CONE FIELD THEORY
David G. Robertson
Columbus, OH 43210
ABSTRACT
This is an overview of the problem of the vacuum in light-cone field theory, stressing
its close connection to other puzzles regarding light-cone quantization. I explain the
sense in which the light-cone vacuum is "trivial," and describe a way of setting up a
quantum field theory on null planes so that it is equivalent to the usual equal-time
formulation. This construction is quite helpful in resolving the puzzling aspects of the
light-cone formalism. It furthermore allows the extraction of effective Hamiltonians
that incorporate vacuum physics, but that act in a Hilbert space in which the vacuum
state is simple. The discussion is fairly informal, and focuses mainly on the conceptual
issues. Additional technical details of the construction described here will appear in a
forthcoming paper written with Kent Hornbostel.
I. INTRODUCTION
The most striking aspect of light-cone field theories is surely the claim that the
vacuum state is simple, or even trivial. The basic argument for this is elementary. We
begins by noting that the longitudinal momentum P+ = po + p3 is positive semidefinite
and conserved in interactions. However, from the free-particle dispersion relation in
light-cone coordinates, 1
p2 +m2
p- = --=.L'--_ _ (1)
p+
it follows that the only finite-energy states other than the bare vacuum that have
p+ = 0 contain massless quanta with Pi. = O-a set of measure zero in more than 1+1
dimensions. Thus if these can be ignored as unimportant in the full theory, then the
1I shall use the convention x == Xo x 3.
223
bare vacuum is the only state in the theory with P+ = 0, and so must be an exact
eigenstate of the full interacting Hamiltonian.
If true, this enormously simplifies any effort to solve field theories nonperturba-
tively using Hamiltonian diagonalization, as attempts to compute the spectrum and
wavefunctions of physical states are not complicated by the need to recreate a ground
state in which processes occur at unrelated locations and energy scales. It furthermore
results in a constituent picture, in which all quanta in, say, a hadron's wavefunction are
specifically related to that hadron. This allows for an unambiguous definition of the
partonic content of hadrons, and makes interpretation of the wavefunctions straight-
forward. For this reason the light-cone framework is the most natural one in which to
encode hadronic properties [1].
It immediately raises the question, however, of whether field theories quantized on
the light cone are equivalent in all respects with the corresponding equal-time theories.
In many cases, notably QeD, there is important physics that in the usual language
is attributed to the properties of the vacuum. The fact that the pion is light on
the hadronic scale is conventionally traced to spontaneous breaking of the axial flavor
symmetries. It is not clear how a trivial vacuum can be a spontaneously broken vacuum
of anything. Furthermore, the QeD ground state is a superposition of "topological"
vacua labeled by a parameter B. This structure is important for understanding why
there is no ninth light pseudoscalar meson (for three light flavors), and for satisfying
cluster decomposition. In addition, if B =I- 0 (and none of the quarks are massless), then
QeD violates P and T, leading for example to a nonzero electric dipole moment for
the neutron. The structure of the vacuum in QeD is clearly connected to observable
consequences of the theory.
One way of reconciling the apparent triviality of the vacuum in light-cone quanti-
zation with the fact that the vacuum has nontrivial physical consequences is based on
the following set of observations [2]. Eqn. (1) implies that particle states that can mix
with the bare vacuum are high-energy states. This suggests that it is natural to think
in terms of effective Hamiltonians. That is, we may imagine introducing an explicit
cutoff on longitudinal momentum for particles
p+ > A . (2)
(For practical calculations it may be preferable to implement the cutoff in some more
sophisticated way. I use this simple scheme to illustrate the point.) This immediately
makes the vacuum trivial and gives the resulting constituent picture. In principle, it
should then be possible to restore the effects of states with P+ < A, including the
physics of the vacuum, by means of effective interactions which, because the states
eliminated have large light-cone energies, will be local in x+. 2 From this point of view,
the statement of the vacuum problem on the light cone is: How can we compute these
effective interactions?
I should perhaps emphasize at this point that the physical vacuum state in light-
cone field theory is not really trivial. It cannot be, if the light-cone theory is to be
equivalent to the corresponding equal-time theory. The potential advantage of the
light-cone approach is that it may allow us to move the effects of the vacuum into the
Hamiltonian and work with a simple vacuum state. The problem is that sorting out
the structure of the vacuum in light-cone quantization is especially difficult. At the
2It is instructive to contrast this with the analogous procedure in equal-time quantization. In this
case there are states that are kinematically allowed to mix with the vacuum at all energy scales, so
that a description of vacuum physics in terms of an effective Hamiltonian is not practicable.
224
kinematical point at which the vacuum lives there is a diverging density of states with
diverging energies, but for which matrix elements of the Hamiltonian also are singular.
This is the loophole in the simple argument for vacuum triviality: states with P+ = 0
do have divergent energies, but they are infinitely strongly coupled to low-energy states.
It therefore does not follow that they decouple from low-energy physics.
Some sort of small-P+ regularization is necessary even without the motivation of
obtaining a constituent picture, and how we approach the problem of computing the
effective interactions will certainly depend on the specific choice of this regulator. For
example, if we think in terms of simple momentum cutoffs then a renormalization group
approach is natural [3J (the effective interactions are the counterterms that remove
dependence on the cutoff A from physical quantities). Another method that has been
studied in some detail involves using discretization as an infrared regulator [4J. In this
case any vacuum structure must be connected with the properties of the zero modes
(in the Fourier sense) of the fields.
Both of these cutoffs, however, force the vacuum to be trivial from the outset, and
we face the problem of how to put the vacuum physics back in. The RG approach
should work in principle-for a particular set of cutoffs there is presumably a single
effective Hamiltonian that gives completely cutoff-independent results and corresponds
to the QCD Hamiltonian. The difficulty here is a practical one. It is not clear whether
this program can really be carried out, particularly since the use of perturbation theory
to construct the RG is probably not sufficient for QCD. The development of nonpertur-
bative RGs for use in light-cone field theory is an extremely interesting and challenging
problem.
Discretized light-cone quantization (DLCQ) is a self-contained formalism with the
inclusion of the zero modes, so the question here is simply whether the va.cuum physics
is present or not at the end of the day. Analysis of simple models suggests that certain
mean-field aspects of the vacuum are in fact captured by the zero modes [5J; however,
not all of the relevant structure is present. In the <l>t+l model, for example, the zero
mode gives rise to the expected strong-coupling phase transition, but with the (incor-
rect) mean-field critical exponent. This suggests that even with zero modes DLCQ
does not contain all of the necessary physics.
The main purpose of this talk is to describe one way of regulating light-cone field
theories that does not force the vacuum to be trivial from the beginning. In fact, the
construction is designed to be completely equivalent to the usual equal-time formalism,
so that there is no confusion about what is or is not present in the theory. I shall also
sketch how it can be used to extract effective light-cone Hamiltonians, which incorporate
the physical effects of the vacuum but act in a space with a trivial ground state.
In order to widen the context of the discussion, however, let me mention a few other
"classic" puzzles regarding light-cone quantization. All of these problems, including
that of the vacuum, are interrelated, and all of them are clarified by the construction
I shall describe.
It has been noted on occasion that a null plane is not a good Cauchy surface for
a relativistic wave equation; in fact, it is a surface of characteristics for such an
equation [6J. Thus initial conditions on one such surface plus a Hamiltonian are
not in general sufficient to determine a general solution to the field equations.
In the quantum field theory this problem is manifested as missing degrees of
freedom. A particularly clear example of this is that of a free massless fermion
in two spacetime dimensions, as first discussed by McC'artor [7]. Here we find,
following the usual light-cone procedure, that the theory contains only right-
225
moving particles; half of the solution space is mIssmg. Additional degrees of
freedom, initialized on a second null plane, are required to properly incorporate
the left-movers. One may have intuitively the idea that this is a problem specific
to massless fields in 1+1 dimensions (or for only those massless particles with
Pl. = 0 in higher dimensions-a set of measure zero), but I shall argue that this
is not the case.
It is not clear what, if any, boundary conditions can be consistently imposed on a
null plane. The problem is that some points on the surface are causally connected,
and requiring fields at such points to be related in some way may result in an
over-determined system. In other words, some boundary conditions may be in
conflict with the dynamics we wish to solve.
There seems to be a mysterious discontinuity in the way massless and massive
fields are treated on the light cone. As mentioned above, for a free massless
fermion in 1+1 dimensions we must include independent degrees of freedom on
two different null planes to recover the correct theory. For nonzero mass, however,
this does not seem to be necessary; the usual light-cone construction gives a theory
that is precisely isomorphic to the equal-time theory of a free massive Fermi field.
Thus the limit is apparently not smooth-it seems that additional degrees of
freedom must be switched on precisely at m = o. We shall argue below that this
apparent discontinuity is an artifact of considering only free field theories. In an
interacting theory a small-k+ regulator must be employed, and this allows for a
completely unified treatment of massive and massless fields. In particular, the
massless limit may be taken smoothly.
Finally let us mention the "No-problem" problem. This is that simply ignoring
most or all of these problems sometimes works. For example, the spectrum of
the Schwinger model can be computed rather well without worrying about any of
the above [8]. This is somewhat astonishing considering that this is a situation
where these problems might be expected to be at their most severe. All the fields
are massless, the free fermion theory is known to be missing half of the necessary
states, and all points on the initial-value surface are in causal contact. There are
many aspects of the Schwinger model that simply cannot be understood without
worrying about these problems-the anomaly, the occurrence of a condensate,
the {;I-vacuum [9]-but the spectrum of physical bosons is reproduced exactly. It
would be nice if this feature were to hold in more complicated examples, that
certain things would be computable even without a complete formulation incor-
porating all the subtleties. One of the goals of the work I will describe is to try
to understand this in a quantitative way, by starting from a manifestly complete
formalism so that the effects of discarding certain pieces of physics can be studied
without ambiguity.
The rest of this paper is organized as follows. In the next section I discuss infrared
regularization using a finite volume, and the initial-value problem in such a box. I show
that with a careful treatment of the boundary surfaces this formulation is completely
equivalent to equal-time quantization. In particular, all conserved charges (including
the Hamiltonian) are guaranteed to be identical to the operators we would construct
in equal-time quantization, and the vacuum state is complicated. I then describe the
formalism in detail for scalar and Fermi fields. Finally, I sketch how to use this con-
struction to obtain effective Hamiltonians for use in light-cone field theories with a
simple vacuum.
226
Figure 1: Finite volume used for infrared regulation.
II. LIGHT-CONE INITIAL VALUE PROBLEM IN A FINITE VOLUME
As discussed above, one aspect of the vacuum problem on the light cone is that it
is difficult to find an infrared regulator that does not automatically remove the vacuum
structure. One way of achieving this involves using a finite volume as a regulator and
treating the initial-value problem carefully [7,10,l1J. This analysis will show that extra
degrees of freedom, beyond what is present in the conventional light-cone quantization,
are required to construct a theory that is equivalent to the equal-time theory. These
additional degrees of freedom are what will allow complicated vacuum structure to
occur.
To be more concrete, let us consider a free massive scalar field in 1+1 dimensions
confined to the region shown in Fig. 1. (The extension to higher dimensions is straight-
forward and is discussed in [11 J.) The first question is: What classical data are required
in order to determine a completely general solution to the field equation
(3)
These data will correspond to ind'ependent operators in the quantum field theory.
Let us begin by assuming that we have specified the value of <p on the conventional
light-cone initial-value surface x+ = 0, for - L ::; x- ::; L. Eqn. (3) then allows us to
evolve 8_<p infinitesimally in x+:
(4)
where
(5)
In order to go on, however, we need <p itself on the new x+ -slice. This is obtained by
integration,
(6)
which of course requires knowing the value of the field at one point on the surface; here
we have arbitrarily taken this point to be (E, -L). Now that we have <p at x+ = t, we
can continue to the next slice, and so on. Clearly the process can be iterated to fill out
227
(a) (b) (c)
Figure 2: Some possible initial-value surfaces. The dark lines indicate where initial
data are specified.
the entire box, provided that in addition to knowing > on x+ = 0, we are also given >
along a surface of constant x-.
We thus find that in order to obtain a general solution to Eqn. (3) we must
specify > both at x+ = 0 (or more generally on any surface of constant x+) and on a
surface of constant X-. This means that when we build the quantum field theory we
should include independent operators initialized on some surface of constant x- tha.t.
represent this classical freedom. This simple analysis does not tell us which surfaces
of constant x- are most useful, however. In order to answer this question it is helpful
to think about actually setting up the quantum theory with fields initialized on both
surfaces. The basic problem is to write down expressions for the fields and find a set of
commutation relations among them such that, e.g., the Poincare generators satisfy the
correct algebra and generate the proper transformations on the fields. One important
point is that in all but the simplest cases (e.g., free field theories), the commutation
relations between field operators at time-like separations are a priol'i unknown. We
should therefore only consider surfaces that contain no time-like separated points. Two
particularly useful surfaces of this type are shown in Figs. 2( a) and (b). The surface
shown in Fig. 2( c) is not suitable.
The two "good" surfaces each have particular advantages. The symmetric one [Fig.
2( a)] is manifestly parity invariant, for example. (Recall that under parity x+ +-4 x-,
so that the two wings are interchanged.) This is helpful because initializing the fields
is completely symmetric on the two wings. The asymmetric surface [Fig. 2(b) 1 will be
more useful, however, when we attempt to extract an effective Hamiltonian for use in
light-cone theories of the usual type, that is, involving the fields initialized at x+ = O.
In fact, the fields on the asymmetric surface can be obtained from those on the
symmetric one by means of a shift in x+ and a minor reorganization of the degrees
of freedom [11J. I shall therefore concentrate first on the symmetric surface, where
initialization is easier, and shift to obtain the formulation on the asymmetric surface
only when we are ready to discuss effective Hamiltonians.
Conserved Charges
How do these additional degrees of freedom enter the dynamics? That is, how are
they incorporated into the Hamiltonian and other conserved charges of the theory? As
we shall see, understanding this leads to certain clarifications regarding the equivalence
of the light-cone and equal-time formalisms.
228
A D
Figure 3: Integration contour for conserved charges.
The basic point is quite simple [7]. For any conserved current,
(7)
we have
f da I" JI" -- 0 , (8)
where the integral is taken over a closed surface. If this is taken to be the surface shown
in Fig. 3, for example, then Eqn. (8) implies
Q (9)
(10)
The middle term in the second line is the usual expression for the charge in light-cone
quantization. We therefore see that inclusion of contributions from the boundaries is in
general necessary if the various charges are to be identical to those we would construct
in equal-time quantization.
To see why the vacuum is complicated when the boundaries are retained, consider
the particular case of the energy-momentum tensor. We have, for example,
p+ = jB d:r+T-+ + rB dx-T+- + rD. dx+T-+. (11 )

A k k
The boundary contributions involve T-+, which contains interaction terms. Thus P+
will no longer be diagonal in a Fock basis, and the physical vacuum state can be
complicated.
Recognizing the role played by the boundaries is crucial in resolving certain spe-
cific puzzles of light-cone quantization. For example, in the Schwinger model (electro-
dynamics of massless fermions in 1+I dimensions) the vector and axial- vector currents
are related by
j+
5 (12)
-r. (13)
229
In the usual light-cone formalism, where charges are obtained by integrating only over
x+ = 0, this implies
Q = Jdx- J+ = Q5 . (14)
But Q5 is supposed to be anomalous, while Q is conserved. The resolution of this
paradox is that contributions from the boundary surfaces also must be included. These
involve integrals of J(5) over a surface of constant x-, so that a difference between Q
and Qs can arise [9].
It might be thought that this sort of problem arises only for massless fields in 1+1
dimensions, or for the PJ.. = 0 modes (only) of a massless field in higher dimensions.
This is incorrect, however, and is a result of considering only free field theories. In a
free massive theory the limit of large box size can trivially be taken, since there are
no divergent matrix elements of the Hamiltonian. 3 Thus the boundary fields decouple
completely and what remains is the usual light-cone theory. This does not happen
for a massless field (or for the PJ.. = 0 modes of a higher-dimensional theory); in this
case the left- and right-moving modes are not coupled and so they can never decouple.
This is what leads to the apparent paradox regarding the difference between massless
and massive fields on the light cone, and is a point I would like particularly to stress.
The need to include boundary fields in order to obtain a light-cone formulation that is
equivalent to the equal-time theory is quite general, and applies to both massive and
massless fields in any number of dimensions. What leads to confusion is that in free
field theory the boundaries can be decoupled for massive fields. (This is because the
vacuum really is trivial for free field theories!) In an interacting theory, however, the
infinite-volume limit need not be trivial, and singularities in the interactions can lead
to observable consequences-vacuum condensates, for example-that would simply be
absent if the theory were formulated without the boundary degrees of freedom.
In the next two sections I will show how to realize this scheme in some simple
cases. The central problem is to obtain general expressions for the fields on the initial
surfaces such that all requirements of relativistic quantum field theory, in particular
causality, are satisfied, and no boundary conditions are employed that are in conflict
with the dynamics. The discussion will necessarily be rather incomplete; my goal is
mainly to give a flavor of how the formalism is set up. Further examples and many
additional details are given in [11].
III. SCALARS
Let us begin by considering a free, massless scalar field, quantized on the symmetric
surface 1. 4 This is a very simple example, but it serves to illustrate many of the basic
points.
It is convenient to impose periodic boundary conditions at equal time,
<1>( -L, L) = (L, -L) . (15)
3More precisely, there is no divergent coupling between states. The free energies diverge for p+ --> 0,
but we can argue that this is irrelevant, as the states for which this occurs are not in the spectrum of
either the light-cone or equal-time theories-they have infinite energy.
4This theory does not actually exist except with an infrared regulator such as the box in place.
The fact that massless elementary scalars are afflicted with unmanageable infrared divergences is
the essence of Coleman's theorem regarding the impossibility of spontaneous breaking of continuous
symmetries in two spacetime dimensions; the necessary Goldstone boson cannot exist. Here we shall
never try to remove the regulator, so that the theory is always well defined.
230
This insures that the Hamiltonian and other conserved charges are time-independent.
Note that because the two points thus related are separated by a space-like interval,
there is no conflict with causality. The equation of motion is simply
(16)
which has as its general solution
(17)
where </> are arbitrary functions. It is clear that initial data on both x+ = - Land
x- = - L are required to obtain the full solution space of the theory. Our first problem
is to find the most general expression for the fields consistent with, e.g., causality, the
equations of motion, and the condition (15).
Now the equation of motion itself tells us what boundary conditions are allowed
on the light-cone surfaces. To see this, integrate it over one or the other of the wings.
We obtain, for example,
o_</>(L, -L) = o_</>( -L, -L) , (18)
which implies, because of the equal-time periodicity condition,
o_</>(-L,L) = o-</>(-L,-L). (19)
Thus the derivative o_</> may be taken to be periodic on x+ = -L. Similarly, o+</> may
be taken to be periodic on the surface x- = - L. We can therefore write general Fourier
expansions for these,
o_</>(-L,x-) o_<p(x-) + 7 (20)
o -( +) (21)
o+</>(x+, -L) = +0 '
where 0 and o are the zero-momentum
modes. (The factors of L are introduced for later convenience.) Explicitly,
(22)
(23)
We can then integrate Eqns. (20)-(21) to obtain the fields on the initial value surfaces,
in terms of the given derivatives and the value of the field at the corner point (- L, - L).
We obtain
<p(x-) - <p(-L) + 7 (x- + L) + </>(-L, -L) (24)
</>(x+,-L) = rp(x+)-rp(-L)+~(X++L)+</>(-L,-L). (25)
Imposing the condition (15) th~n leads to the identification
</>0 = o. (26)
231
Finally, it is convenient to define
J{ -L, -L) == <pC -L) + <p{ -L) + 'r/J (27)
so that
J{-L,x-) = <p{x-) + <p{L) + ~o (x- + L) + 'r/J (28)
J{x+,-L) = <p{x+)+<p{L)+~(X++L)+'r/J. (29)
These represent the most general expressions for the fields on the initial surfaces that
are consistent with the equations of motion and the periodicity condition (15). The
classical initial data are the Fourier modes a q and aq , and the numbers Jo and 'r/J. In
the quantum theory these become operators whose commutation relations are to be
determined by demanding that the correct Heisenberg equations and Poincare algebra
are obtained. 5
Note that the fields are not themselves periodic on either of the surfaces. Imposing
periodicity on the fields, as one would do in DLCQ, corresponds to a less general
expression for the fields, or, equivalently, to focusing on a subset of the classical solution
space. In the quantum theory this corresponds to omitting degrees of freedom.
The Poincare generators receive contributions from both wings of the initial-value
surface:
plJ.=jL dx+T-IJ.{x+,_L)+jL dx-T+J.t(-L,x-). (30)
-L -L
The only nonvanishing components of the energy-momentum tensor are
T++ (31)
T-- (32)
i:
so that
i:
p+ = 2 dx- [8_J(-L,x-)r (33)
p- = 2 dx+ [8+J(x+, -L)r (34)
These are guaranteed to be the same operators, though expressed in a different repre-
sentation, as we would obtain in an equal-time quantization of this theory. Requiring
that the Heisenberg equations reproduce the field equation (16) leads to the commuta-
tion relations
i i
[<p{x-), <p{y-)j = --f(X- - y-) + -(x- - y-) (35)
4 4L
[<p{x+), <p{y+)j = - -i f (+
4
x - Y+) + -4Li (+
x - Y+) (36)
i
['r/J, Jo] = 4' (37)
5Note that the regulator breaks longitudinal boost invariance, so that the commutator [P, Kj =
=F2ip is not recovered for finite L. In practice the Heisenberg equations are usually sufficient to
determine the field algebra, perhaps supplemented by the requirement that [P+, P-] = 0 and that the
fields commute for space-like separations.
232
where f is the antisymmetric step function. In terms of the Fock operators, the first
two commutators correspond to
[a at] = [a at] =
q, q, Dq,p (38)
Furthermore, with the redefinitions
1/J = lrn(b + bt ) (39)

2v2
tPo - 2~(bt - b) , (40)
Eqn. (37) can be recast as [b, btl = 1.

With these commutation relations the Poincare generators correctly translate the
fields in x, and furthermore satisfy, [P+, P-] = O. We thus have a satisfactory repre-
sentation of the theory, including all the relevant degrees of freedom. The Fock space
is spanned by states of the form
(41)
where 10) is annihilated by a q , aq, and b. The Poincare generators are given by
(42)
(43)
In fact, the entire construction can be shown to be correct by mapping it onto the
equal-time theory of a massless scalar in a box. That is, we can quantize the theory at
equal time with the periodicity condition (15), and solve the resulting theory to obtain
the fields on the boundary surfaces. We then identify the light-cone Fock operators in
terms of the equal-time ones, and so obtain the relation between the two representations.
Not surprisingly, the a q modes correspond to right-moving quanta, the aq to left-moving
quanta, and tPo to the zero-momentum mode in the equal-time Fourier expansion. The
Hamiltonian and momentum operators can also be constructed and are identical to the
appropriate combinations of (42) and (43).
The massive case is more involved, essentially because the left and right movers
are coupled by the mass term (for example, a finite boost can now change a left mover
to a right mover, and vice versa). The commutator we wish to reproduce in this case
is the Pauli-Jordan function
where p, is the mass. This is properly causal; it vanishes whenever (x - y) is space-

like. Furthermore, the Bessel function is unity on the light cone, reproducing the usual
light-cone commutator of the fields. Light-cone commutators involving derivatives of
the field, however, can be p,-dependent. For example,
(45)
This suggests that p, should be built in to the expressions for the fields, in such a way
that those commutators involving derivatives that are needed are correctly reproduced.
233
This can in fact be done, and the details are given in [11]. For now, however, let
us turn to the case of a massive Fermi field. This is similar in many ways due to the
presence of the mass, but allows for a slightly less cluttered treatment.
IV. FERMIONS
Consider a free, massive Dirac fermion in 1+1 dimensions, quantizing again on the
symmetric surface. The standard light-cone decomposition of a Fermi field,
1 0
1/J = fr 7 1/J, (46)
allows the separation of the Dirac equation into the coupled pair
m
= 21/J- (47)
( 48)
A classical analysis of these, analogous to that given in Sect. II for the scalar field,
indicates that in order to determine their general solution we must specify w+ on
x+ = -Land t/J- on x- = - L. The problem is thus to find expressions for the
quantum fields on these surfaces that furnish a representation of the free-field canonical
anticommutation relations, are properly causal, etc. The relevant commuta.tors are
{1/J+( -L, x-), t/Jt( -L, y-)} b(x- - y-) ( 49)
{1/J-(x+, -L), t/J~(x+. -L)} b(x+ - y+) (50)

zm
{t/J+( -L, x-), 1/J~( -L, y-)} = --f(X- - y-) . (51)
4
In addition, all field operators should anticommute for space-like sepa.rations.

For m = 0 the necessary construction has been given in Ref. [7]. After imposing
anti periodicity at equal time,
(52)
an argument identical to that given for the massless scalar shows that '1/.'+ may be chosen
to be antiperiodic on x+ = -L and '1/-'- may be chosen to be antiperiodic on x- = -L.
Thus we may write 1/J+(-L,x-) = 1/J(x-) and 1/J-(x+, -L) = ~(;r+), where
(.53)
(54)
and k; = ml" / L. The correct commutation relations are
{1/J(X-),1/Jt(y-)} = b(x--y-) (5.5 )

{~(x+),~t(y+)} = b(x+ - y+) , (56)
234
with all other anticommutators vanishing, which correspond to
(57)
The Poincare generators receive contributions from both wings of the initial-value sur-
face [see Eqn. (30)), and correctly translate all fields in x. Note that if the degrees
of freedom represented by (in and bn are not included, then the theory contains only
right-moving particles and is not equivalent to the equal-time theory of a massless Dirac
fermion.
There are various ways of arriving at a correct construction for the case m '# O.
Here I shall simply motivate one possible way, and then show that it works.
Let us start out by trying the same procedure as in the massless case. That is,
we choose tP+ to be an antiperiodic function tP( x-) on x+ = -Land tP- to be an
antiperiodic function ~(x+) on x- = -L. We shall also attempt to preserve the simple
commutation relations (57). Now the Dirac equation allows us to obtain tP- on the
surface x+ = -L:
(58)
(59)
(the treatment of tP+(x+,-L) is completely symmetric, of course). It is helpful to

rewrite this in the form
(60)
from which we see that the simple commutation relations (55)-(56) do not correctly
reproduce the anticommutator (51). The second term on the RHS of Eqn. (60) gives us
what we want; the last term is the troublemaker. Let us attempt to fix this by defining
tP-( -L, x-) to be
(61)
which gives the right commutator but changes the values of 4'- at the endpoints. We
now have
_ im
tP-( -L, L) = tP( -L) =f -
jL
dY-1fJ(Y-) . (62)
4 -L
Note that equal-time antiperiodicity then implies
-tP_(-L,L)
-
If,(L) + -imjL dy-tP(y-) . (63)
4 -L
suggesting that we should also modify the value of tP- at the point (L, -L).
We are thus led to seek an expression for tP- (x+ , - L) that is more complicated
than a simple antiperiodic function~. Furthermore, the modified tP- (x+, - L) should
involve tP, that is, should depend on tP+(-L,x-). A clue as to how to achieve this is
obtained by thinking about causality. Recall that tP- (x+ , - L) and tP+ (- L, x-) should
anticommute for x+ '# -L and x- '# -L. It would therefore be dangerous to include
in tP- (x+ , - L) anything involving tP except at the endpoints :r+ = L; anything else
235
would be likely to cause t/J- and t/J+ to fail to anticommute at space-like separation.
This suggests a modification of t/J- (x+, - L) at the endpoints only, of the particular
form
t/J-(x+, -L) = ;j;(x+) + im

4
[2 + f(X+ - L) - f(X+ + L)] JL dy-'l/'(Y+) .
-L
(64)
That is, we add in a discontinuity at each endpoint designed to reproduce Eqns. (62)-
(23). This leaves the fields properly causal; away from the corner points t/J-(x+, -L) =
t/J(x+) and t/J+( -L, x-) = t/J(x-), which anticommute. Of course, the situation is
symmetric with respect to + t-t - so that we should also take
im [2+f(X--L)-f(X-+L) ] JL dy+t/J(y+).
t/J+(-L,x-)=t/J(x-)+- - (65)
4 -L
Eqns. (64) and (65) are the final expressions for the independent fields on their respec-
tive initial-value surfaces.
As mentioned previously, there are other ways of arriving at these, but this one
highlights the physical motivations behind the construction. Note that the fields are not
strictly antiperiodic on the initial surfaces. That this must be the case follows immedi-
ately from integrating the equation of motion and imposing equal-time antiperiodicity;
we obtain, e.g.,
(66)
We have constructed the fields to be as antiperiodic as possible, however, while pre-
serving the simple anticommutation relations (55)-(56). The motivation for this is that
ultimately we would like to use this construction to obtain an effective Hamiltonian for
the usuallight-con~ degrees of freedom (t/J+). We would therefore like the formalism to
resemble as much as possible the usual one, with an antiperiodic '1/'+. We could perhaps
have arrived at simpler expressions for the fields had we not insisted on maintaining
simple Fock space commutation relations.
The energy-momentum tensor has components
(67)
T++ = 2it/Jt8_t/J+ + H.c. (68)

T-- = 2it/J~8+t/J- + H.c. (69)
and the Poincare generators work out to be [Eqn. (30)]
p- = _ im2jL dx-jL dy-t/Jt(X-)f(X- _ y-)t/J(y-) + 2ijL dx+J;t(x+)8+;j;(x+)

4 -L -L -L
(70)
p+
(71)
236
These can be shown to correctly translate all fields in x, including the fields at the
corner points, and furthermore satisfy [P+, P- J = O. We thus have a satisfactory
representation of the theory with the boundary degrees of freedom present. Note the
presence of terms coupling together the fields on the two wings.
A few technical points are worth noting. First, the discontinuities in the fields
at the corner points lead to singularities in their derivatives, and these give rise to
important contributions in the pieces of pI" involving T++ and T--. Second, in verifying
the Heisenberg relations or the Poincare algebra, it should be kept in mind that since
'l/J( x-) is antiperiodic, it follows that, for example,
(72)
Furthermore, when the argument of a delta function vanishes at an endpoint of the
integration region, the proper definition of the integral is, e.g.,
1 L
-L
1
dx- f(x-)b(x- - L) = - f(L) .
2
(73)
The Asymmetric Surface
It is now fairly simple to obtain suitable expressions for the fields for use with the
asymmetric initial-value surface shown in Fig. 2(b). These may be obtained from Eqns.
(64) and (65) by a shift of x+ and a slight reorganization of the resulting expressions
[11 J. For the independent fields we obtain
(74)
and
'l/J_(x+,L) = ~(x+) - im [t(x+) l11L dy-~'(y+). (75)
4 -L
Integrating the equation of motion then gives us the constrained fields:
'l/J+(x+,L) = 'l/J(L) - -im 1L dy+t(x+ - -

y+)~'(y+) (76)
4 -L
-
'l/J-(O,x-) = 1,/>(0) - -imjL dy-t(x- -y-)~'(y-). (77)
4 -L
The Poincare generators are now obtained by integrating
P=~jO dx+T-(X+,L)+~jL dx-T+(O,x-)+~ (L d:r+T-(x+,-L). (7S)

2 -L 2 -L 2 Jo
We find
(SO)
237
Figure 4: Decoupling the boundaries in the large-L limit.
Again, these can be shown to properly translate the fields and to commute with each
other.
V. DISCUSSION
I have argued that in a finite volume, introduced as an infrared regulator, includ-

ing degrees of freedom not present in the usual light-cone formalism is necessary if
the resulting theory is to be equivalent to the corresponding equal-time theory. This
follows from analyzing the equations of motion, as well as from constructing Poincare
generators and other charges that are the same as the operators we would construct
in equal-time quantization. I have also attempted to give some flavor of the formalism
necessary to achieve this equivalence. A major advantage of this approach is that it
gives a regulated theory but does not force the physical vacuum to be trivial. This is
a welcome feature, as the QeD vacuum, for example, is certainly not trivial.
Its main disadvantage is that it is quite complicated, even at the level of free field
theory. In fact, it is more complicated than equal-time field theory, so that if the point
were to use it directly as a calculational tool it would be largely useless. However, recall
that vacuum triviality on the light cone really means that the effects of the vacuum
should be expressible in terms of local effective interactions for a set of fields initialized
at a fixed x+. That is, we should be able to construct an effective theory which has a
trivial vacuum state, but nontrivial vacuum physics. The light-cone framework is the
only one in which such an approach is possible, due to the unusual connection between
long distances (in x-) and high (light-cone) energies reflected in the dispersion relation
(1 ).
This construction can be used to obtain such effective Hamiltonians in a natural
way. The basic idea is to consider the limit in which the box size L becomes large
compared to some fixed scale I of physical interest [Fig. 4]. In this case, the fields that
live on the boundaries appear to decouple from those that live at x+ = 0; by causality,
quanta from the boundaries that can influence the region of interest have their world
lines pushed closer and closer to the light cone as the box gets large. This decoupling
may be illusory, however, because the couplings of these states can simultaneously
diverge. The goal is to extract the finite remainders in this limit, and build them in
to an effective Hamiltonian for use an a "traditional" light-cone calculation. Details of
238
some simple examples are given in [11].
This formulation of the light-cone theory with boundaries is also instructive from
the point of view of the puzzles listed in Sect. I. For example, in the language of
partial differential equations the construction corresponds to the "characteristic" initial-
value problem, as distinct from the Cauchy problem. This is perfectly well defined
from a mathematical point of view. There is no real problem with using surfaces of
characteristics to specify initial conditions, but more than one such surface is necessary
in general.
The question of what boundary conditions may be imposed on the characteristic
surfaces can be addressed naturally in the course of setting up a concrete realization
of the scheme. In general, imposing strict periodicity or anti periodicity is seen to be in
conflict with the dynamics, and results in the removal of degrees of freedom.
The massless limit of the construction is perfectly smooth; the degrees of freedom
that seem to appear only for m = 0 are actually necessary for the massive theory as
well, and so are present from the outset. As discussed above, the confusion over whether
the extra fields are necessary in the massive case arises from considering the artificially
simple case of free theory, where the boundaries do decouple for massive fields. In
general they are required, and when they are present there is nothing discontinuous, or
otherwise unusual, about the massless limit.
Regarding the "No-problem" problem, the construction allows us to study directly
the effects of discarding various pieces of physics. It would be very helpful to have a
quantitative understanding, for example, of what could be computed reliably without
including the boundaries. The present scheme allows this to be studied concretely.
There are also more specific questions along these lines that can be considered. For
example, what is the relation between the effective interactions induced by the con-
strained zero modes in DLCQ and the effective interactions one obtains in the full
theory? Do they in fact capture some sort of "mean field" vacuum properties, and
if so, what can be calculated reliably in that framework without addressing the full
boundary problem?
It is perhaps worth restating that the potential advantage of light.-cone quantiza-
tion over equal-time quantization is not that the vacuum becomes trivial, but rather
that the vacuum can trivially be made trivial, by means of a simple cutoff on longitudi-
nal momenta. This has several immediate advantages. When coupled with the insight
that the states thus removed are mainly high-energy states in ;3+1 dimensions, it leads
to the hope that we might be able to derive a, constituent app1"O.1:imation to QCD. In
fact, it is the only framework that offers any realistic hope of achieving this. In addition,
for problems where the vacuum does not play an important physical role the ability
to ignore it so simply may be quite useful. In QCD, for example, it is presumably
not necessary to completely understand the pion or to have strictly linear long-range
potentials in order to obtain a reasonable description of, e.g., heavy qua,rkonia. For
this sort of problem it may be quite reasonable to ignore the effective interactions that
represent the vacuum, or, equivalently, to use the conventional light-cone formalism
without boundary fields.
For problems where the vacuum does playa role, however, its structure must
be properly accounted for. The formalism described here, or something equivalent,
will be necessary to obtain the effective interactions that mediate vacuum physics.
For QCD this can be expected to be quite difficult, in part because the problem is
intrinsically nonperturbative. Of course, the QCD vacuum is a difficult problem in any
(known) calculational scheme! In any case, the possibility of obtaining a constituent
approximation to QCD, complementary to the lattice, and of making contact with the
natural and very successful phenomenology based on the light-cone Fock representation,
makes this set of problems of considerable interest.
239
ACKNOWLEDGMENTS
It is a pleasure to thank the organizers of ORBIS SCIENTIAE 1996 for providing such a
stimulating and enjoyable atmosphere. This work was done in collaboration with Kent
Hornbostel, and I am grateful to him for many enjoyable and illuminating discussions
on the subject of light-cone field theory. I also thank Gary McCartor and Robert
Perry for many helpful conversations regarding the light-cone vacuum. This work was
supported in part by a grant from the the U.S. Department of Energy.
REFERENCES
1. For an overview of QCD phenomenology from the light-cone point of view, see:
S. J. Brodsky, these proceedings; S. J. Brodsky and D. G. Robertson, "Light-
Cone Quantization and QCD Phenomenology," hep-ph/9511374, to appear in
the proceedings of the ELFE Summer School and Workshop on Confinement
Physics, Cambridge, UK, July 1995.
2. K. Hornbostel, Phys. Rev. D 45, 3781 (1992);
3. R. J. Perry, Ann. Phys. 232, 116 (1994); K. G. Wilson. T. S. Walhout, A.

Harindranath, W.-M. Zhang, R. J. Perry, and St. D. Glazek, Phys. Rev. D 49,
6720 (1994).
4. S. J. Brodsky and H.-C. Pauli, in Recent Aspects of Quantum Fields, H. Mitter

and H. Gausterer, Eds., Lecture Notes in Physics, Vol. 396 (Springer-Verlag,
1991), and references therein.
5. T. Heinzl, S. Krusche, S. Simbiirger, and E. Werner, Z. Phys. C 56, 415 (1992);

D. G. Robertson, Phys. Rev. D 47, 2549 (1993); C. M. Bender, S. S. Pinsky,
and B. van de Sande, Phys. Rev. D 48, 816 (1993); S. S. Pinsky and B. van
de Sande, Phys. Rev. D 49, 2001 (1994); J. Hiller, S. S. Pinsky, and B. van de
Sande, Phys. Rev. D 51, 726 (1995).
6. F. Rohrlich, Acta Phys. Austr. 32,87 (1970); P. J. Steinhardt, Ann. Phys. 128,
425 (1980).
7. G. McCartor, Z. Phys. C 41, 271 (1988).
8. H. Bergknoff, Nucl. Phys. B 122, 215 (1977); T. Eller, H.-C. Pauli, and S. J.
Brodsky, Phys. Rev. D 35, 1493 (1987).
9. G. McCartor, Z. Phys. C 52,611 (1991); ibid. 64,349 (1994).
10. K. Hornbostel, Ph. D. thesis, Stanford University (1988).
11. K. Hornbostel and D. G. Robertson, manuscript in preparation.
240
THE TRANSVERSE LATTICE IN 2+1 DIMENSIONS*
Brett van de Sande a and Simon Dalleyb
aMax Planck Institut fiir Kernphysik

Postfach 10.39.80, D-69029 Heidelberg, Germany
bDepartment of Applied Mathematics and Theoretical Physics

Silver Street, Cambridge CB3 9EW, England
INTRODUCTION
Based 011 an idea due to Bardeen and Pearson, we formulate the light-front Hamil-
tonian problem for SU(N) Yang-Mills theory in (2+ I)-dimensions using two continuous
space-time dimensions with the remaining space dimension discretized on a lattice. We
employ analytic and numerical methods to investigate the string tension and the glue-
ball spectrum in the N - t 00 limit. Our preliminary results show qualitative agreement
with recent Euclidean lattice Monte Carlo simulations. In the following, we attempt to
give a more pedagogical introduction to the idea of the transverse lattice; more detail
may be found in Ref. [1].
MOTIVATION
There exists a large gap between the quantum field theory of QCD with its many
successes in the context of perturbation theory and experimental observables associated
with QeD bound states: the hadron mass spectrum, structure functions, form factors,
et cetera. There has been little progress during the last 20 years in building a bridge
between the field theory and these non-perturbative properties of the hadron spectrum.
Let us contrast two attempts to bridge this gap: the Euclidean lattice Monte Carlo
(ELMC) approach and light-front field theory where one uses Hamiltonian techniqes
on a theory which is quantized on a surface of constant x+ = (Xo + x 3)/V'i. As far as
progress is concerned, ELMC is much further along, benefiting from a large research
effort since the mid 1970's. In contrast, the light-front approach is not so far along;
most research effort has occured since late 1980's. For ELMC, further progress is limited
mainly by the speed of available computers. For light-front, progress is currently limited
by wuceptual issues exempli grati renormalization. Even when successful, ELMC is not
able to measure many interesting observables, such as structure functions, directly. For
'Presenled by B. van de Sande
241
light-front field theory, physically interesting observables are quite easily calculated from
the bound state wavefunctions. The approach that we will discuss here, the transverse
lattice, uses ideas from both the light-front and lattice approaches.
Instead of solving the full theory of QCD, we will focus on a particular model:
SU(N) Yang Mills theory in 2+1 dimensions in the N -7 00 limit. Why 2+1 dimensions
instead of 3+ 1 dimensions? Aside from the obvious fact that it has fewer degrees
of freedom, one should note that the theory is super-renormalizable. In the context
of lattice calculations, this means that there is no critical point and the associated
'critical slowing' is absent. Consequently, excellent lattice spectra are available for 2+ 1
dimensions [2J which one can use as a comparison. Why large N? The N -7 00 limit
allows considerable simplification of our computational problem; in addition, the lattice
data indicates that 1/N corrections to the low energy spectrum are quite small.
THE TRANSVERSE LATTICE
A number of years ago Bardeen and Pearson [3, 4J formulated a light-front Hamil-
tonian version of lattice gauge theory, which makes use of the fact that two components
of the gauge field are unphysical. In this approach two spacetime dimensions are kept
continuous x = (XO x 2 ) / yI2 while the remaining transverse spatial dimension Xl
is discretized on a lattice. Lattice sites are labeled by integer-valued index i and the
lattice constant is a. Following Bardeen and Pearson, we associate longitudinal gauge
fields At (x+, x-) with lattice site i and link variable Ui ( x+, x-) with the link between
sites i and i + 1,
Ui A~l
----------x----------x----------
i +1
(1)
Note that At and Ui are N x N color matrices where At is Hermitian and traceless
and Ui E SU(N). We can define the color Maxwell tensor
(2)
where Q, (3 E {+, -} along with a covariant derivative
(3)
Using this notation, one can write down an action,
which has several important properties:
If we take the naive continuum limit a -7 0 where Ui = exp( -igaAD, we recover

the continuum Yang-Mills action .
By construction, this action is 100% gauge invariant.
242
It has an automatic confinement mechanism. We will say more about this later.
Linearization
In an ideal world, Eqn. 4 is the action one would quantize. However, we are
faced with a problem. The field theory on each link of this transverse lattice is an
SU(N) x SU(N) non-linear a-model. Upon quantizing this theory, we must enfor~e
N2-nonlinear constraints on each link. Although some attempt has been made to do thIs
quantization [5], success has remained elusive. Instead, we perform the lin~arizat~on
suggested by the original authors [3]. In this scheme one replaces the umtary lmk
variables U with N x N complex matrices' Mi, that is, Ui - t J2ag 2Mi.
The obvious next step would be to include an effective potential to enforce the
constraint J2ag 2Mi E SU(N) dynamically. For instance the termt
Ac Tr{ (M/ Mi - 2ai r} (5)
Ac Tr{M/ MiM/Mi} - 4ai Tr{ M/Mi} + (2a i)2 (6)
is minimized precisely when ..;'2{i?Mi E U(N). One could imagine adding such a term
to the Hamiltonian, taking the Ac - t 00 limit, and recovering the transverse lattice
action (4). Unfortunately, a closer inspection reveals several problems:
The second term in (6) is a mass term with negative a (mass/ coefficient. Quan-
tization of the theory breaks down for negative (mass)2 for the same reason that
it breaks down in the 't Hooft model [6] and the spectrum becomes unbounded
from below. Presumably this negative (mass)2 term is a signal for the presence of
spontaneous symmetry breaking. One could imagine solving the broken phase of
the theory using zero mode techniques [7]. The resulting 'shifted' theory would
have some complicated effective potential representing the effects of the spon-
taneous symmetry breaking; this effective potential would undoubtedly be quite
different than the above form (5).
Tadpole contractions associated with four-point interactions, such as the first term
in (6), produce a divergent shift in the mass term Tr{M/M;}. Renormalization
leaves the mass as a free parameter in the Hamiltonian.
A successful renormalization group analysis would introduce couplings between

neighboring links. For instance, one expects terms associated with the constraint
2ag 2 MiMi+! E SU(N).
From a more practical viewpoint, implementing the Ac - t 00 limit forces half of

the dynamical degrees of freedom at each link to decouple from the theory. This
is not good news for numerical calculations where the computational difficulty
depends critically on the number of dynamical degrees of freedom.
As a consequence of these considerations, a first-principles construction of the correct

effective potential Vi is not a simple matter. Instead, we take a more pedestrian ap-
proach and include in Vi all operators up to fourth order in Mi containing one color
'The O(n) a-models are a good example of where this linearization is known to work.
tFor large N. we can ignore the distinction between U(N) and SU(N).
243
trace
(7)
and try to determine the associated coupling constants empirically. Note that the Al
term is local to one link and the A2 term acts on two adjacent links.
Quantization
Next, we quantize the theory, write down the Hamiltonian P-, and construct a
basis of states. Further details may be found in Ref. [1]. At each lattice site i we have
a 1+1 dimensional gauge theory with conserved current
(8)
We set (LAt = 0 by choice of gauge and throwaway the associated dynamical zero
mode f dx- At. The Ai field obeys the equation of motion
(9)
As evidenced by the absence of time derivatives 0+ in Eqn. (9), Ai is not a true

dynamical degree of freedom but is constrained. We solve the constraint equation (9)
for Ai and remove it from the theory. At this point, quantization of the theory is
straightforward. The momentum conjugate to M;(x-) is o_M/(x-) and we impose the
usual equal x+ commutation relations*
(10)
The longitudinal momentum operator
p+ = 2 L
,
Jdx- Tr{ o_M;o_M/} (11 )
and Hamiltonian
(12)
generate translations in the x- and x+ directions, respectively.

Finally, we construct a basis of states. The zero mode of the Ai constraint equation
(9) generates the GauB' law constraint
(13)
.. The factor of 1/2 comes from the Dirac procedure for constrained systems. Also, we drop the constrained zero mode Jdx- Mi.
244
which we must impose on the basis of physical states. That is, physical states must be
color singlets at each lattice site. We construct a basis of closed color loops:
Tr{MiMn 10) :( r x
Tr{ MiM/ MiM/} 10) E } x
Tr{ M iM i+1 M/+ IMn 10) :( x ): (14)
Tr{MiMi+1Mi~IMi+IM/+IMn 10) { ::c }

et cetera. i+ 1 i+2
where the x- co-ordinate of each link field remains arbitrary and the number of links
M J must equal the number of anti-links MJ. Since the loop-loop coupling constant is
non-leading in N, we do not include states with more than one color trace in the Hilbert
spacf'; we deal with a free string theory. * Incidentally, if we do introduce a state that
is not a color singlet at some site, exempli grati
(15)
we find that its energy diverges.
Confinement
This theory has built-in linear confinement. Consider two test charges separated
in the x- direction. The first term in Eqn. (12) acts as a linear confining potentiatt
Now consider two charges at lattice sites i and j. Due to the GauB' law constraint (13)
we must construct a color string of at least Ii - j I link fields between the two charges. If
/1 2 > 0, there is some minimum energy associated with each link field and the transverse
string tension is nonzero. As we shall see in the next section the resulting potential is,
in fact, linear. Similar arguments hold for the original action (4).
N uITlerical Techniques
We solve the spectrum on the computer using DLCQ techniques [9]. For the x-
coordinate, we impose anti-periodic boundary conditions Mi(X-) = -Mi(X- + L) and
use a momentum space representation. For integer valued cut-off ]{ = LP+ /(27r),
momenta are labeled by odd half integers "m E {I /2,3/2, ... } where Lm "m = ]{. This
yields a finite basis of states. Also, we employ a truncation in particle number.
The first term of the Hamiltonian (12) dominates the behavior of the theory. Thus
it is natural to choose units such that the associated coupling constant g2 N/ a is set
equal to 1. This choice of units is assumed in the following.
STRING TENSION
Next, we introduce a method for measuring the string tension in the xl direction.
Consider a lattice of n transverse links and periodic boundary conditions. We construct
'In fact it is possible to choose V, such that one has a theory of free bosonic strings [8].
tNotethat (lL )-l f\x-) = f dy-Ix- - y-I f(y- )/2 inEqn. (12).
245
6
4
2
M 3
0
0 1 2 3 4 5
n
Figure 1. Lowest M2 eigenvalue vs n for states that wind once around

the lattice. Here, j{ = 10.5 or 11 ((n + 4)-particle truncation), J.l2 = 0.1 ,
= =
Ada 1, and A2/a -1. Also shown is a fit to a quadratic.
0.25
11 2
Figure 2. Parameters such that the lowest M2 eigenvalues are equal for
n = 4 and 5 (see Fig. 1), where j{ = 10.5 or 11 ((n + 4)-particle truncation) .
This is an estimate of vanishing string tension. Also shown is a line such that
the M2 eigenvalues are approximately degenerate for n = 3, 4, and 5.
a basis of states that wind once around this lattice and calculate the lowest eigenvalue
of the invariant (mass)2 operator M211J1) = 2P+ P-IIJI). The continuum limit string
tt'nsion is
. 1 D.M
hm - - - na fixed. (16)
a-+O a D.n '
This definition is be equivalent to the standard definition of string tension using two
test charges if the charges are placed sufficiently far apart. If we plot M2 vs. n in
Figure 1, we obtain a quadratic corresponding to a linear confining potential.
In fact, we find that the string tension decreases with decreasing f-l2, leading us to
the conclusion that f-l2 is a measure of the physical lattice spacing. The continuum limit
is partially fixed by requiring vanishing D.(M)/ D.(n); see Fig. 2. Above the surface, the
string tension is positive corresponding to nonzero lattice spacing.
In dddition we have an analytic ansatz for these states that wind once around the
lattice that is valid in the >'d a < 0 region of parameter space. It agrees well with the
numerical results.
246
Figure 3. Parameters such that the lowest M2 eigenvalue is zero, /{ = 10 to
14 with extrapolation using.a fit to L~=o cm /{-m/2 (6-particle truncation).
Below this surface, the spectrum is tachyonic; above the surface, it is well
behaved.
SPECTRUM
Before we look at the details of the spectrum, we can make some statements about
the allowed region in parameter space. A numerical estimate of the 'edge' of this region
is plotted in Fig. 3. Comparing Figs. 2 and 3, we see that a continuum limit with a
nOll-tachyonir spectrum occurs only for the "wedge shaped region" -AI::; A2 ::; Ad2.
In this region, the continuum limit occurs for /1 2 ::::i 0 to within numerical errors. Finite
lattice spacing corresponds to /1 2 slighly above the surface in Fig. 2.
Symmetries
The theory possesses several discrete symmetries. Charge conjugation induces the
symmetry C : (Mi)I,m +-t (M/)m,1 where 1, m E {l, ... , N}. Parity is the product of
two reflections PI : Xl ----t _xl and P 2 : x 2 ----t _x 2. In light-front quantization, PI is
an exact symmetry P, : Mi +--> M!i while P 2 : x+ +--> x-, is complicated. Its explicit
operation is known only for free particles [10], which we call "Hornbostel parity." The
latter is nevertheless useful since it is often an approximate quantum number and can
be used to estimate P 2 [11]. Given P 2 and PI we can determine whether spin.:1 is even
or odd using the relation (-1).:7 = PI P 2 If rotational symmetry has been restored in
the theory, states of spin .:1 ::f 0 should form degenerate PI doublets 1+.:1) 1-.:1) [2].
As with the lattice results, we use "spectroscopic notation" 1.:1I P 'c to classify states.
One expects the lowest two eigenstates to be approximately two-particle states
(17)
with the lowest state having a symmetric wavefunction corresponding to 0++ , and the
first excited state having an antisymmetric wavefunction, corresponding to 0--. Of
course, these states also have 6-particle et cetem contributions.
Results
Ideally, we would like to predict the effective potential based on some connection
247
120
100
80
I
j
t It It Ii It
ft
j
2
M 60 +t I~ +t +t It
+t +
40 tt +
t + lattice
20
+ +ours
0
++ +- -+ ++ -+ +- -+ +-
0 0 0 0 2 2 2 2 1 1 1
Figure 4. A comparis()n of our spectrum with SU(3) ELMC data in units of

the physical string tension [2] for various 1.1ll'c. The parameters g2 N/a =
=
3.90, p2 0.134g 2 N/a, Al =
0.487g 2 N, and A2 =
1.108g 2 N were chosen by
=
a best fit to the lattice data, X2 40, where K = 10 (8-particle truncation).
Our error estimates are solely for the purpose of performing the X2 fit.
to the continuum theory, restoration of rotational invariance, et cetera. However, as a

first step, we use instead a best X2 fit to the ELMC results of Teper [2].
An example spectrum is shown in Fig. 4. Similar spectra are found in other regions
of coupling constant space above the 'wedge-shaped region' (Fig. 3). We label the lowest
2-- and second 0-- states based on the expectation value of the number operator and
0
-
detennined (_1).1" based on Hornbostel parity [11]; the exception is the 1..11+- sector
where Hornbostel parity gave exactly the opposite of the desired results. Beyond this,
J is determined by a best fit to the lattice data.
Since this spectrum is the result of a best fit, it is not very predictive. However,
we can use the result to tell us about our model. At first glance, we seem to have a
pretty good match. However, we note several problems:
The energy of the lowest 0-- state is too low.
The lowest 2++ and 2-+ states form a degenerate doublet if rotational symmetry
is restored (as indeed happens for the lattice data). In our case, the splitting is
large. This discrepancy dominates the error in our X2 fitting procedure.
Let us review the possible sources of error in our calculation.
Large N. We compare N - t 00 spectra to SU(3) lattice results. However, based

on lattice calculations for SU(2), SU(3), and SU(4), liN corrections to the low
energy spectrum are small [2].
Finite K. Our discretization of the longitudinal momentum introduces some

error. However, we have generated spectra for K = 10,11, 1?, 13, 14, extrapolated
248
to large I< (6-particle truncation), and compared to large N extrapolated ELMC
spectra. We saw no real improvement in our results.
Particle Number. We also impose a truncation in the number of particles. We

have examined spectra for 4-, 6- , and 8-particle truncations (I< = 10), extrapo-
lated to large number of particles, and compared to large N extrapolated ELMC
spectra. We saw no real improvement in our results.
Hamiltonian. The effective potential Vi that we chose (7) did not contain any
6-point or higher interactions. In addition, we did not include any operators
containing multiple traces, for instance (Tr{ MiMl})2. *
CONCLUSIONS
We have investigated the transverse lattice model of Bardeen and Pearson [3] for
(2 + I)-dimensions in the large-N limit using linearized link variables and an empirical
eJfective potential Vi. We identified a choice for Vi corresponding to vanishing string
tension. The glueball spectrum in the vicinity agreed qualitatively with that coming
from the presumably reliable Euclidean lattice Monte Carlo results [2]. Most impor-
tantly however, we did not see significant signs of rotational invariance which could lead
one to conclude that the transverse gauge dynamics were correctly accounted for by Vi,
We believe that our choice (7) is probably too simple and that higher order terms are
necessary to see improvement in Qur spectrum.
Future work includes the addition of more operators in the effective potential. Also,
we can use our method of measuring string tension to measure the physical lattice
~pacil1g. This issue needs further investigation. Most importantly, we need a more
COllcrete connection between our model and the continuum theory. This would allow
us to better predict the correct effective potential.
ACKNOWLEDGMENTS
This work was supported, in part, by the Alexander Von Humboldt Stiftung. B.
van de Sande would like to thank the organizers of ORBIS SCIENTIAE 1996 for providing
a stimulating and enjoyable atmosphere.
REFERENCES
[1] S. Dalley and B. van de Sande, preprint No. DAMTP-96-21 and hep-ph/9602291.
[2] M. Teper, Phys. Lett. 289B, 115 (1992); Phys. Lett. 311B, 223 (1993); Lecture
at Rutherford-Appleton Laboratory, U. K. (December 1995).
[3] W. A. Bardeen and R. B. Pearson, Phys. Rev. D 14,547 (1976).
[4] W. A. Bardeen, R. B. Pearson, and E. Rabinovici, Phys. Rev. D 21, 1037 (1980).
[5] P. Griffin, Nucl. Phys. B372, 270 (1992) .
Although this tenn is generally not leading order in N, it does act on the two panicle subspace of the theory.
249
[6] G. 't Hooft, Nucl. Phys. B72, 461 (1974); M. Einhorn, Phys. Rev. D 14, 3451
(1976).
[7] D. Robertson, Phys. Rev. D 47, 2549 (1993); S. S. Pinsky, and B. van de Sande,
Phys. Rev. D 49, 2001 (1994); S. S. Pinsky, B. van de Sande, J. Hiller, Phys. Rev.
D 51, 726 (1995).
[8] I. R. Klebanov and 1. Susskind, Nucl. Phys. B309, 175 (1988).
[9] H.-C. Pauli and S. Brodsky, Phys. Rev. D 32, 1993 and 2001 (1985).
[10] K. Hornbostel, Ph. D Thesis, SLAC-PUB No. 333 (1988).
[11] B. van de Sande and M. Burkardt, preprint No. MPI H-V 1995 and
hep-th/9510104, to appear in Phys. Rev. D.
[12] S. Dalley and l'. R. Morris, Int. Journal Mod. Phys. A5, 3929 (1990).
250
SECTION VI - THE MATTER OF DARK MATTER
SUSY DARK MATTER WITH UNIVERSAL AND NON-UNIVERSAL
SOFT BREAKING MASSES
R. Arnowitt 1 and Pran Nath 2

1Department of Physics
Texas A&M University
College Station, TX 77843-4242
2Department of Physics
Northeastern University
Boston, MA 02115
ABSTRACT
Predictions for rates for detecting SUSY dark matter are given within the frame-
work of supergravity grand unification models with R-parity. Here the dark matter
candidate is the lightest neutralino for almost the entire parameter space and event
rates range from (10- 5 to 10) events/kg da. Effects produced by non-universal soft
breaking masses are discussed.
INTRODUCTION
Supergravity grand unified models [1] with R-parity offer a natural explanation
of the origin of the dark matter (DM) observed in the universe. In these models, su-
persymmetry is spontaneously broken at a scale > MCUT ~ 1016 GeV in the "hidden"
sector. This spontaneous breaking then generate; a spontaneou~ breaking of SU(2) x
U(l) at the electroweak scale by radiative effects using the renormalization group equa-
tions (RGE) in [2]. One is lead then to a relatively predictive theory. The R-parity
invariance implies the existance of a stable lightest supersymmetric particle, the LSP.
Over almost the entire parameter space the LSP is the lightest neutralino [3,4], X~, and
the relics of these from the Big Bang is the dark matter candidate.
In the Minimal Supergravity Model (MSGM), all the masses and interactions
of the 32 SUSY particles are determined by four parameters and one sign [5]. At the
GUT scale there are the four soft breaking parameters: the universal gaugino mass
m1/2, the universal scalar mass mo, and the universal cubic and quadratic soft breaking
parameters Ao and Bo. In addition there is a fifth SUSY preserving parameter, the
Higgs mixing parameter /Lo . The condition of radiative breaking of SU(2)x U(l) at
the electroweak scale determines /L5 in terms of the other parameters leaving the sign
of /Lo arbitrary. Using the RGE, one may replace the above parameters by the gluino
mass m g, the scalar mass mo, the t-quark A parameter at the electroweak scale At, and
253
tanJ1 = (H2 ) / (HI) (where H 1 ,2 are the two Higgs doublets of the low energy theory).
Experiment currently allows for two options for the gluino mass. Either the
gluino is heavy i.e. mg:' 170 GeV [6] or the gluino is light mg~ 10 GeV [7]. We consider
here only the heavy gluino possibility. In our discussion we will restrict the parameter
space as follows: 0 < mo ::;1 TeV, 170 GeV < mg ::; 1 TeV, 2::; tanJ1 ::; 25, -6::; Atmo ::;
6 and both signs of J.L. In principle, once SUSY is discovered, four measurements wOlild
determine the four parameters of the MSGM. Already two measurements, the top mass
mt ~ 175 GeV [8] and the branching ratio BR (b -+ s + 1'), eliminate about 65% of this
parameter space.
For a simple GUT group, it is difficult not to have universality of the gaugino
masses ml/2 at M CUT . (One might have at most a splitting of ~ 10% from Planck slop
terms.) However, a complicated Kahler potential, or the assumption that universal soft
breaking occurs at a scale above MCUT can lead to non-universal mo and Ao parameters
at M CUT . We will also discuss these possibilities and some of their effects on DM event
rates.
RELIC DENSITY AND EVENT RATE CALCULATIONS
While the X~ produced by the Big Bang are absolutely stable, they can an-
nihilate in the early universe into ordinary matter: X~ + X~ -+ f + 1... , where f =
quarks, leptons, etc. If we let n be the number density of X~ (no the density at thermal
equilibrium) then the Boltzman equation in the early universe is [9]:
~~ = -3H(t)n - (av) (n 2 - n~) (1)
where H is the Hubble constant a is the annihilation cross section (v is the relative X~
velocity), and < > means thermal average:
(av) = Jdvv 2 EXp[_V2/4X] av/ Jdjvv 2 Exp [-v2/4X] (2)
drops below the expansion rate. This occurs at = kTt/m given by x,

Here x == kT /m where m is the X~ mass. Freezeout occurs when the annihilation rate
x,-1 =rv n
~n [ ex,1/2()
av x I m N ')-1/2]
(G N, (3)
out and c = 0 (1). Eq. (3) gives

time [10]:
x,
where G N is Newton's constant, N, the effective number of degrees of freedom at freeze-
~ 1/25. One finds then for relic density at present
0-0 h2 ~ 2.48 X 10- 11 (

T
x.~ )3 (T"() 3_,_
N 1/ 2
(4)
Xl T"( 2.73 J(x,)
where 0x.~ = PiPe (p is the X~ mass density, Pe = 3H 2 /(87fG N ) is the critical mass
density), (Tx.?/T"()3 is the reheating factor, h = H/100kms- 1 Mpe and
(5)
We note the factor 1/ J (x,) shows that the larger the annihilation cross section, the
less relic neutralinos exist at present time. We will require in the following:
254
(6)
Since experimentally 0.4~ h ~0.8, the range of Eq. (6) covers most of the DM models
that are consistent with current evaluations of the power spectrum e.g. CHDM (cold-
hot DM model with nx? ~ 0.7, nHDM ~ 0.3, nB = 0.05, h = 0.5) gives n Z1 h 2 = 0.18;
tCDM (tilted model with nx? = 1.0, h = 0.5 - 0.6 n = 0.8) gives nx? = 0.25 - 0.36 and
ACMD (cosmological constant model with nx? = 0.2 - 0.4, n A = 0.6 - 0.8, h = 0.8)
gives nZ1 h 2 = 0.13 - 0.26.
The formula for nx? h2 depends on a wide array of disparate physics: gravita-
tional (G N ), cosmological (H), statistical (k) and electroweak (0:2)' It is a remarkable
success of supergravity models that they can accomodate such a narrow window of
nx?h2 as Eq. (6) when they might have given values of nx?h2 many orders of magni-
tude different.
We turn next to the detection of dark matter particles. Rotation curves for the
Milky Way imply a density p ~ 0.3 GeV /cm 3 of DM particles in the Galaxy which are
impinging on the Solar System with velocity v~ 300 km/s. The most promising mode
of seeing these is from terrestial detection of the scattering of the X~ by quarks in a
nucleus [11]. The X~ - quark scattering (Fig. 1) can be characterized by an effective
Lagrangian:
where q(x) and mq are the quark fields and their mass,Xl (x) is the X~ field and
PL,R = (1 =t= "/)/2. In Eq. (7), Aq, Bq arise from the Z t-channel and ij s-channel
pole, while Cq arises from the light and heavy Higgs (h,HO) t-channel and ij s-channel
poles. The first term of Eq. (7) gives rise to spin dependent (incoherrent) scattering,
while the second term gives rise to coherrent scattering from each quark in the nucleus
and hence scales by the nuclear mass M N . The expected scattering event rates that
can be seen by a detector takes the form
(8)
where p and v are the local density and velocity of impinging Milky Way X~. Rcoh and
R spin have the form
16mMN 1 12 (9)
R spin = [ ]2 A spin
MN+m
We note that for large MN that R spin rv l/MN while Rcoh rv MN and hence Rcoh
R spin for large M N .
The calculation proceeds then as follows [12,4]: (1) Calculate nx?h2 and restrict
the SUSY parameter space so that Eq. (6) is obeyed; (2) Calculate the X~ - quark
scattering for the allowed parameter space of (1); (3) Assemble quarks into nucleons
and nucleons into nuclei (by making use of nuclear form factors) to obtain the event rate
R expected for a detector made of a specific nucleus. The experimentally observable
domain of interest for the near future is for R~ 0.01 events/kg da and so we will restrict
the discussion to this domain.
Figs. (2) and (3) give the maximum and minimum event rates for a Pb detector
(the largest possible choice for M N ) for Jl < 0 and Jl > 0 respectively [4].
255
Zo
Fig. 1 X~ - quark scattering diagrams contributing to Leff of Eq. (7).
0
V
::I.. -1
f
W -2
E-

~
E-
Z
~
:>
~
8..J
At/roo
Fig. 2 Maximum and minimum event rates for /1 < 0 (solid curve) for a Pb DM detec-
tor as a function of At/mo. The dashed curve includes the restrictions from b -+ s + 'Y
data.
One sees that event rates run from (10- 5 to ~ 10) events/kg da as one ranges over
the entire parameter space. With an experimental sensitivity of R;:::: 0.01 events/kg da
one will be sampling a sizable portion of the parameter space. However, a significant
increase in sensitivity will be needed to examine the full parameter space.
RADIATIVE BREAKING AND COMPOSITION OF X~
The condition of radiative breaking of SU(2) x U(l) determines /1 2 , and the size
of /1 turns out to govern much of what one might expect for DM detection rates. One
has
(10)
256
,,
,,
0
,,
A
,,
~
-1 ,,
f ,,
,,
t.d' ,,
E-
< -2 ,,
~ ,,
E-
Z
,
LtJ
>
LtJ -3

-4
Fig. 3 Same as Fig. 2 for f-L > o.

where f-L7 are the running Higgs masses for HI and H2 (plus I-loop correction). Since
tan 2 (3 > 1, radiative breaking occurs when f-L~ turns negative. In general, the f-L7 are
scaled by m6 and my 2 and for most of the parameter space f-L6, ms > > M~. Thus for
most of the parameter space f-L2 > > M1 and is an increasing function of m6, my 2.
The LSP which enters in the detector scattering cross section is the lightest
eigenvector of the neutralino mass matrix. It will therefore be a mixture of gaugino
( W3, B) and Higgsino (ih, ih) states: X~ = nl W3 + n2B + n3ih + n4H2. The
coefficient ni can be calculated numerically. However, in the domain f-L2 > > M1, one
can get an approximate analytic formula which exhibits more clearly the physics of the
phenomena. One finds [4].
1 Mz 1 Mz . [ . _]
(11)
rv
nl = ---A - - szn28w szn2(3 - mI/f-L
2 f-L m2 - ml
n2 ~ 1- ~ ~z ~sin8w[1 + (mI/f-L )sin2(3 + mUf-L2] (12)
n3 ~ ~z ~sin8wsin(3[1 + (mI/f-L )ctn(3] (13)
n4 ~- :z ~sin8wsin(3[ctn(3+mdf-L] (14)
where A = 1 - mU f-L2 and ml = (ad a3 )my. Thus in the f-L2 > > M~ domain, the X~ is
mostly Bino (B) i.e. n2 ~ 0.90-0.95, but with a significant Higgsino contribution, n3,4 ~
0.20. Now the ZI - q scattering diagrams of Fig. 1 show that the Higgs poles contribute
only to Rcoh and give Rcoh rv (n2n3)2, (n2, n4)2 = O(MV f-L2) while R spin gets its main
contribution from the Z pole of R spin rv (n~ - n~) 2 = 0 ( Mi / f-L4). Combined with the
257
- 800
-
~
CJ
10
10
II
::I
-
0
~ 400
CJ
::.:=
200
o 5 10 15 20
Tanbeta
Fig.4 Maximum value of mg as a function of tan (3 for At/mo = 0,2,3,4 (ascending
order on the right) for the part of the parameter space where R::::: 0.01 event/kgda.
nuclear mass behavior discussed above, one has then for most of the parameter space
that Rooh R spin ' Further, since J.1 increases with mg, the currently experimentally
accessible region of R~ 0.01 implies a maximum value for mg. This can be seen in Fig.
4 [4]. Thus if accelerator experiments were to show that m/:~ 650 GeV, it would imply
that current DM detectors would not be capable of detecting the local DM X~.
b -+ s + l' AND CONSTRAINTS FROM THE VALUE OF mt
While there is no direct experimental evidence for the existance of SUSY parti-
cles, other experiments have already begun to restrict the parameter space. Thus the
observed b -+ s + l' branching ratio [13] of BR ~ (2.32 30%) x 10- 4 (compared to
the SM expectation of BR ~ (2.9 30%) x 10- 4 ) and the measured top quark mass
[4] mt ~ 175 GeV together eliminate about 2/3 of the SUSY parameter space.
When J.1 and At have the same sign, the supersymmetric contributions increases
the BR(b -+ s + 1'), while with opposite sign, they tend to cancel. Since already the
central point of the experimental data lies below the SM, the current data tends to
exclude regions where J.1/ At is positive. This can be seen from the dashed lines of Figs.
(2,3) [4] which shows that almost all of the region with At > 0 is eliminated.
The largeness of mt indicates that the top quark is close to its quasi infra-red
fixed point. Thus the top Yukawa coupling constant ht is close to its Landau pole
at the GUT scale. One finds [14], h;(O) = hW)/(E(t)Do) where Do = 1 - mUm},
t = 2n (MCUT/Q) (with Q the running scale), E(t) is a form factor, and mf ~ 199sin(3
GeV is the fixed point mass. Thus ht(O) -+ 00 as mt -+ mf' The RGE couple this pole
to other parameters of the theory. In particular, the light stop squark obtains a mass
of the form [15]
258
mgluino(GeV)
mu (GeV)
4 2 o
At/mO
Fig. 5 Three dimensional plot of /1, mg and At/mo for rna = 100 GeV, tan {J = 20, mt
= 175 GeV, 51 = 0 = 52.
ml 1 = -~A~/Do + (Non - Pole) (15)
where AR ~ At - 0.613mg. The light stop is thus eventually driven tachyonic as

mt ---+ m f' eliminating such regions of parameter space. One sees that this effect is
more significant for At < 0 since then Ah is larger. This is seen in Figs. (2,3), where
most of the negative At domain is eliminated, even for /1 > 0 where the b ---+ s + I
constraint does not operate. The combined effects of the b ---+ s + I data and the
heavy top then leads to the elimination of about 60% of the parameter space i.e. At is
restricted to [4]: -0.25::; At/mo ::; 0.5 (/1 > 0) and -0.6::; At/mo ::; 5.5 (/1 < 0) 4.
Fig. 5 shows the situation for universal masses and illustrates that /1 2 is generally
large except along the Landau pole "valley" AR = 0 (more precisely along the mini-
mum of the second plus third term in Eq. (20)] which runs diagonally across the figure
(from front right to far left). In Fig. 6, the effects of the non-minimal masses force /1 2
negative for the deepest part of the Landau pole valley, eliminating that part of the
parameter space. Otherwise, /1 is not significantly changed. Fig. 7 shows that the
effects of the non-minimal terms are significantly enhanced when rna is increased.
259
mgluino(GeV)
mu(GeV)
o
At/mO
Fig. 6 Same as Fig. 5 except 01 = -1,02 = +1.
NON-MINIMAL SOFT BREAKING MASSES
As we have seen in the previous section, the parameters that control much of the
predictions of DM detection rates are J.L2 and mp..
1
We consider in this section the effects
on J.L2 if the soft breaking masses at MCUT are non-universal. For these parameters, the
most general assumption one can make is to have the following soft breaking masses at
M CUT :
(16)
where 151,2 represent deviations from the universal (mass)2 rh6, and (mQ; mu) are addi-
tional non-universal (iL , bL ; i R , bR ) masses. From Eq. (10) (neglecting loop corrections
for simplicity) J.L2 is given by
J.L 2 + 2t
= t 2 1_ 1 [11 ma + t 2 1_ 1 [151 - 2t
2 ( 1 - 3Da)]-2 12( 1 + Da ).]-2
U2 ma
(17)
260
mgluino(GeV)
mu(GeV)
4 2 o
At/mO
Fig. 7 Same as Fig. 6 except mo = 300 GeV.
where one finds that m6 = ! (M~ +m~ ), t = tanf3 and the ai are 0 (1). Since rno and mg
range up to 1 TeV, and the Landau pole term A~/ Do is large, one generally has p,2 > >
M~. The 51 ,2 corrections will then not qualitatively effect results if 151 ,2 12!. However,
there are particular cases where the non-universal corrections can become important:
(1) For tan 2 f3 > > 1, the universal (first) term of Eq. (20) "accidentally" vanishes
at Do ~ ~. This occurs at mt ':::' 168 GeV which is quite close to the experimental
value of mtl (2) The Landau pole term is generally large (making the 51,2 corrections
a relatively small effect) except where the residue AR "accidentally" vanishes i.e. from
Eq. (7) along the line At ~ 0.613m g. (3) 51 ,2 will produce negative contributions,
reducing J.l2 when 52 > 0, and 51 < 0, and this effect becomes large for large m6.
The above effects can be seen in Figs. (5-7).
When J.l2 is significantly reduced, the X~ will no longer be mainly Bino and so
this will effect the event rate analysis given above , though mostly in specific regions of
parameter space. Thus as seen in Fig. 2, event rates for universal masses tend to peak
when 1 ~ At/mo ~ 3, and if this region lies in the Landau valley, then there can be an
overall reduction of the expected event rates.
261
CONCLUDING REMARKS
Supergravity grand unified models with R-parity predict over almost their entire
parameter space that the lightest supersymmetric particle is the lightest neutralino, the
X~. One of the successes of this theory is the fact that the expected relic density for
such particles fall within the narrow bounds required by astronomical estimates of the
amount of dark matter for a significant amount of the parameter space. Calculated
expected detector event rates for the local dark matter of the Milky Way range from
(10- 5 to 10) events/kg da. Thus while current detectors (sensitivity of 10- 2 events/kg
da) should be able to explore a portion of the parameter space, major improvements
would be needed to examine the full parameter space.
For a large majority of the parameter space, the coherrent scattering in terrestial
detectors dominate over the spin dependent scattering. This implies that event rates
would scale with the nuclear mass of the detector target. If such an effect were seen, it
would be a strong indication that the incoming dark matter particle was the X~. This
shows the importance of having a number of detectors based on different nuclei. Fur-
ther, accelerator experiments can be expected to delimit the SUSY parameter space,
making event rate predictions more precise.
REFERENCES
1. A.H. Chamseddine, R. Arnowitt and P. Nath, Phys. Rev. Lett. 29, 970 (1982).
For reviews see P. Nath, R. Arnowitt and A.H. Chamseddine "Applied N=1
Supergavity" (World Scientific, Singapore, 1984); H.P Nilles, Phys. Rep. 100,
1 (1984); R. Arnowitt and P. Nath, Proc. of VII Swiece Summer School, ed. E.
Eboli (World Scientific, Singapore, 1994).
2. K. Inoue et al. Prog. Theor. Phys. 68, 927 (1982); L. Ibanez and G.G. Ross,
Phys. Lett. BllO, 227 (1982); L. Alvarez-Gaume, J. Hagelin, D.V. Nanopoulos
and K. Tamvakis, Phys. Lett. B125, 2275 (1983); L. Ibanez and C. Lopez, Nucl.
Phys. B233, 545 (1984); L. Ibanez, C. Lopez and C. Munos, Nucl. Phys. B256,
218 (1985).
3. G. Kane, C. Kolda, L. Roszkowki and J. Wells, Phys. Rev. D49, 6173 (1994).
4. R. Arnowitt and P. Nath, NSF-ITP-95-67; CTP-TAMU-14/95; NUB-TH-3/21/95,

to be pub. Phys. Rev. D.
5. L. Hall, J. Lykken and S. Weinberg, Phys. Rev. D27, 2359 (1983); P. Nath, R.
Arnowitt and A.H. Chamseddine, Nucl. Phys. B227, 121 (1983).
6. E. Gallas (DO Collaboration), R. Kaom (CDF Collaboration) talks at Rencontres

de Moriona (1996).
7. P. Nath, A.H. Chamseddine and R. Arnowitt, Proc. of 1983 Coral Gables Con-
ference on High Energy Physics, ed. Mintz, Perlmutter (Plenum Press, 1985), L.
Alvarez-Gaume et al. Ref. [2]. For recent discussion of the light gluino see E.W.
Kolb, Theol. Proceedings.
8. L. Galtieri (CDF Collaboration), R. Partiridge (DO Collaboration) talks at Ren-

contres de Moriond (1996); P. Langacker, talk at Unification: From Weak Scale
to Planck Scale, Santa Barbara (1995).
262
9. For discussion of calculations of relic density, see E.W. Kolb and M.S. Turner,
"The Early Universe" (Addison-Wesley, Redwood City, 1989); G. Jungman, M.
Kamionkowski and K. Greist, "Supersymmetric Dark Matter", to be pub. Phys.
Rep.
10. B.W. Lee and S. Weinberg, Phys. Rev. Lett. 39, 165 (1977); D.A. Dicus, E.
Kolb and V. Teplitz, Phys. Rev. Lett. 39 168 (1977); H. Goldberg, Phys. Rev.
Lett. 50, 1419 (1983); J. Ellis, J.S. Hagelin, D.V. Nanopoulos and N. Srednicki,
Nucl. Phys. B238, 453 (1984).
11. M.W. Goodman and E. Witten, Phys. Rev. D31, 3059 (1985); K. Greist, Phys.
Rev. D38, 2357 (1988); D39, 3802 (1989); (E); R. Barbieri, M. Frigeni and G.F.
Giudice, Nucl. Phys. B313, 725 (1989); M. Srednicki and R. Watkins, Phys.
Lett. B225, 140 (1989).
12. J. Ellis and R. Flores, Phys. Lett. D263, 259 (1991); B300, 175 (1993); Nucl.
Phys. B400, 25 (1993); K. Greist, Phys. Rev. Lett. 61 666 (1988); A. Bottino
et al. Astro. Part. Phys. 1 (1992); 2, 77 (1994); M. Drees and M. Nojiri, Phys.
Rev. D48, 3843 (1993); R. Arnowitt and P. Nath, Mod. Phys. Lett. A10, 1257
(1995); P. Nath and R. Arnowitt, Phys. Lett. B336, 395 (1994), Phys. Rev.
Lett. 74,4592 (1995); G. Kane et al. Ref. [3].
13. M.S. Alam et al. (CLEO Collaboration), Phys. Rev. Lett. 74,2885 (1995).
14. C. Hill, Phys. Rev. D24, 691 (1981); Ibanez et al. Ref. [2].
15. P. Nath, J. Wu and R. Arnowitt, Phys. Rev. D52, 4169 (1995).
263
Search for SUSY in the D0 Experiment
Sharon Hagopian *
Introduction
Supersymmetry (SUSY) is a spacetime symmetry which relates bosons to fermions

and introduces supersymmetric partners for all the Standard Model (SM) particles!.
The discovery of such particles would verify that supersymmetry is the correct extension
of the standard model and open up vast new areas of work for both experimentalists
and theorists. This report will discuss three searches for supersymmetric particles using
data taken with the D0 detector in the 1992-1993 Fermilab pp collider run at Vs = 1.8
TeV: search for squarks and gluinos, search for charginos and neutralinos, and search
for top squarks.
The D0 Detector
The D0 detector is described in detail elsewhere 2 It has uranium-liquid argon

calorimeters which provide very nearly hermetic coverage for good missing transverse
energy ($t) measurement and good hadronic and electromagnetic resolution for good
electron and jet energy measurement. It also has a central tracking system and a muon
spectrometer with coverage at large and small angles with respect to the colliding
beams.
The Minimal Supersymmetric Model
One of the simplest supersymmetric extensions of the Standard Model (SM) is

the minimal super symmetric standard model (MSSM)3. Besides introducing sparticles
corresponding to all the SM particles, the MSSM adds two more Higgs doublets and
assumes that R-parity, the SUSY multiplicative quantum number is conserved. This
means the sparticles are produced in pairs and decay to the stable Lightest Supersym-
metric Particle (LSP), which is usually assumed to be the lightest Neutralino (Zd. In
the MSSM with Super Gravity (SUGRA) inspired mass relations at the GUT scale, the
large number of SUSY parameters can be reduced to five 4 6 . These can be chosen at
the low energy scale to be tan~ (ratio of the Higgs vacuum expectation values), M H +
* For the De> Experiment
265
(mass of the charged Higgs), JL (Higgino mass mixing parameter), and the if (squark)
and 9 (gluino) masses.
Search for Squarks and Gluinos
The experimental signature of squark and gluino cascading decays to the LSP is
jets and/or leptons and missing transverse energy CIA), due to the energy carried away
by the LSP which does not interact in the detector. In the D0 search, events with
leptons were rejected to reduce the background from Wand Z leptonic decays. Two
independent searches were done: one search required three jets and a high Itt threshold
while the other search required four jets and a lower Itt threshold. Results from these
two searches have been combined to obtain the final squark and gluino mass limits.
Three jet Analysis
The trigger sample was 9,625 events from an integrated luminosity of 13.50. 7 pb -1.
Omine requirements were:
1) a single interaction
2) Itt > 75 GeV

3) three or more jets with E t > 25 GeV, passing quality cuts
4) reject electrons and muons
5) no jet-Itt correlations
Of the 17 events surviving these cuts, one event was rejected because it contained a
muon consistent with a cosmic ray, and two other events were rejected because their
large .$t was caused by vertex reconstruction errors. The final candidate data sample
contained 14 events, consistent with the 14.2 4.4 background events expected from
W + 2,3 jets and QeD. Results from this analysis have been published 5.
Four jet Analysis
The same trigger sample and integrated luminosity was used. The different omine
requirements were:
1) four or more jets with E t > 20 GeV, passing quality cuts
2) Itt > 65 GeV

The estimated background was 5.2 2.2 events. The final candidate data sample con-
tained five events, again consistent with the background from the SM.
MSSM Signal Simulation
The MSSM model was used for the signal calculation, assuming SUGRA-inspired
degeneracy of squark masses 6. Only squark and gluino production were considered, no
slepton or stop production. The mass of the top quark was set to 140 Ge V / c2 , and for
the MSSM parameters, the following values were used:
266
1) Ratio of the Higgs vacuum expectation values tanf3 = 2.0
2) Mass of the charged Higgs M H + = 500 GeV jc 2
3) Higgsino mass mixing parameter J. = -250 GeV

Using the Monte Carlo program ISASUSY, 7 squark and gluino events were gen-
erated for pairs of m(q)-m(g) points in the search region and processed through Del
detector simulation, trigger simulation and event reconstruction programs. The signal
efficiencies determined from this simulation ranged from 10% -20%.
Calculation of Mass Limits from Cross Sections
U sing these signal detection efficiencies, the luminosity and the number of visible
events above SM background, a 95% confidence level (C.L.) upper limit cross section
was determined. This was compared with a leading order theoretical cross section 7 to
determine the lower mass limit for each squark and gluino mass combination. Figure 1
shows the region in the m(g)-m(q) plane excluded by Del, along with previous results
of other experiments 8. For heavy squarks, a lower gluino mass limit of 173 Ge V j c2
was obtained, and for equal squark and gluino masses, a mass limit of 229 GeV jc 2 was
obtained at the 95% C.L.
500
:1 D0 Update 95%
.'
:1 / (Preliminary)
N
U :j
---~ :j
CJ
400
:;
:i
:\
C/)
~ 300
~ \
.;
: \
ro
5- 200 . \., ...... ./
"
(f)
100 : UA1/UA2
: 95% C.L.
100 200 300 400 500 600

Gluino Mass (GeV Ic 2 )
Figure 1: Del, CDF, LEP and UA1jUA2 squark and gluino mass limits as a function
of squark and gluino mass.
CharginofNeutralinos and The Minimal Supersymmetric Model
In the MSSM and mini~al SUGRA model there are two chargino states (Wi ,i=1,2),
and four neutralino states (Zi,i=1,4) corresponding to mixtures of the SUSY partners of
267
t.!te Higgs bosons, Wand Z bosons and the photon. As before, the lightest neutralino,
ZI is assumed to be the LSP. The five SUGRA parameters can be taken as mo (the
.common scalar mass at the GUT scale), ml/2 (the common gaugino mass), tanj3 (ratio
of the Higgs vacuum expectation values), Ao (the common trilinear interaction), and
the sign of p, (Higgsino mass mixing parameter). These parameters determine the values
of m Wl and m Z2 '
Chargino/Neutralino Search
Charginos and neutralinos are produced in pairs at pp colliders with the WI Z2 pair
having the largest cross section over much of the SUSY parameter space9 . The WI can
decay into qql or ill plus a LSP, while the Z2 can decay into qq, lIV or II plus a LSP.
The final state of three leptons + Itt has the least SM background and is the subject
of the D0 search.
Data Analysis
Combinations of single lepton and dilepton triggers were used for the four final
states (eee, eep" ep,p" and p,p,p,). Offline, events were required to have Itt> 10GeV,
at least three leptons, but not more than one electron in the forward region (where
extra material causes photon conversion into e+e-), and mass (p,p,) > 5 GeV to reduce
background from J /1/1 events and combinatoric background. No candidates are seen
consistent with WI Z2 pair production and subsequent decay into trilepton final states.
The background consists primarily of single lepton and dilepton events with one or
more spurious leptons, except for the p,p,p, channel, where it is mostly from heavy flavor
(bb and cc) events. Table 1 gives the integrated luminosity per channel, the number of
events passing the omine requirements and the background per channel.
Channel
eee eep, ep,p, p,p,p,
J Ldt(pb I) 12.5 12.5 12.5 10.8
Require Number of events passed
Ne + Nm ~ 3 5 2 5 7
Ne!wd<2 4 0 - -
Itt> 10 GeV 0 - - -
M"" > 5 GeV /c 2 - - 0 0
Backgrounds 0.80.5 0.80.4 0.60.3 0.10.1
DATA 0 0 0 0
Table 1. Analysis cuts for each of the search channels, showing the number of events
left after a cut has been applied, and the predicted background per channel.
Cross Section Limits
Detection efficiencies were determined using a combination of data and Monte Carlo
simulation. Signal events were generated using ISAJET 7.06 10 and processed through
D0 simulation, programs to determine kinematic and reconstruction efficiencies. Elec-
tron identification efficiencies were calculated from monte carlo +data and verified
268
using Z _ e+e- events. Muon identification efficiencies were based on Z - pp and
J /1/J _ pp event samples. From these efficiencies, the luminosity and zero candidate
events a 95% confidence level (C.L.) upper limit on the cross section for producing W1
Z2 pairs times the branching ratio into anyone of the trilepton final states was deter-
mined for W1 masses from 45 - 100 Ge V / c2. The results from the four channels were
combined with the assumption that B( eee) = B( eep) = B( epp) = B(ppp). Figure 2
shows the resulting limit in the region above the LEP limit l l as a function of the mass
of the W1
:0- 10' DO 95% CL Excluded Rog'on
-3:
LL 1()O
OJ
\:) (01
10 '
40 50 60 70 80 90 100
W 1 mass (GeV/c 2 )
Figure 2: The 95% C.L. limit on cross section times branching ratio into anyone
trilepton final state, as a function of M w" along with the region of M w, excluded by
LEP. Also shown are bands of theoretical predictions, as described in the text.
For comparison three bands of theoretical curves are shown. Band (a) shows the
ISAJET production cross section obtained with a wide range of input parameters,
multiplied by a branching ratio of 1/9, which is the maximum branching ratio obtained
when the W1 and Z2 decay purely leptonically. Bands (b) and (c) show the values
obtained from ISAJET with mo=200-900 Ge V / c2, Ao=O, m1/2=50-120 Ge V / c2 and
p < O. Band (b) is for tan,B = 2 and band(c) for tan,B = 4. Upper limits on U(W1 Z2 )
B(W1 -lvZt} B(Z2-> liZ1) range from 3.14 pb for m w, = 45 GeV/c 2 to 0.63 pb
for m w, =100 GeV /c 2.
The Top Squark and The Minimal Supersymmetric Model
Early MSSM calculations assumed that all squark masses were degenerate. But the
high mass of the top quark 12 implies its Yukawa interactions are large, which can drive
the top squark mass lower than the other squark masses. Mass splitting by left/right
mixing may make one stop state, t1 , the lightest squark of all 13.
Search for the Top Squark
Top squarks are produced in pairs in pp collisions, with a cross section about 1/10
that of top quark pairs at the same mass, since top quarks are fermions and have extra
helicity factors relative to the scalar top squarks. The decay modes of the top squark
depend on its mass relative to that of its possible decay products. If mi, > mb + m w,'
then the favored decay will be i1 - b + W1, which is a top-like signature. If the top
squark has a mass heavier than the lightest slepton and sneutrino, then it will have
three body decays into b + slept on + neutrino or b + lepton + sneutrino. But if the
top squark is lighter than the lightest chargino, slept on and sneutrino, then the only
decay channel open is i1 - C + Zl 14. This mode has a signature of two acollinear jets
and Itt and was the subject of the D0 search.
269
80
........
C\J
:; 60
(!)
~ 40
IN
E /
/
20
25 50 75 100 125
my (GeV/c 2 )
1
Figure 3: The D0 95% Confindence Level exclusion contour. Also shown is the result
from the OPAL experiment at LEP.
Top Squark Analysis
The trigger sample, with a requirement of Itt > 35 GeV, was 83,474 events from a
integrated luminosity of 13.5 0.7 pb -1. Offline requirements were:
1) a single interaction
2) Itt > 40 GeV

3) two jets with E t > 20 GeV, passing quality cuts
4) 165 0 > 4>(jet1,jet2) > 90 0 (acollinear in 4
5) no jet-$t correlations
6) reject electrons and muons
Three events survived these cuts. The QCD background from back-to-back jet pairs
was eliminated by requiring the jets to be acollinear in 4>, the opening angle in the
plane perpendicular to the beam. The background from lepton Wand Z decays was
calculated to be 3.49 1.1. 7 events. The data is therefore consistent with the SM
background.
Mass Limit Calculation
Signal events were generated for various combinations of m;, and m Zl using ISAJET
7.13 10 , and processed through D0 simulation programs to determine efficiencies. From
these efficiencies, which ranged from 1% - 5%, and the theoretical production cross
section, exclusion limits were obtained at the 95% C.L. Figure 3 shows the region in
the m tl vs m Zl plane excluded by D0 along with previous results from the OPAL
experiment 15 at LEP.
The allowed region for this decay is bounded by mb +mw +m Zl > m tl > me +m Zl .
The gap between the LEP excluded region and the D0 excluded region for 60 GeV Ic 2 >
m(tl) > 45 GeV/c 2 is due to the D0 trigger requirement of Itt> 35 GeV. For the 1994-
1995 run, this trigger threshold was reduced to Itt > 25 GeV. When these data are
270
analyzed, the gap should be filled. The highest mt;. value excluded is 93 GeV for m ZI
= 8 GeV /c 2 For m ZI = 44 GeV, masses are excluded for mt;. > 85 GeV /c 2
Future Searches
New results from analyzing the 1994-1995 data with an integrated luminosity of
over 90 pb- 1 should be available in the next year. Several studies on selecting the
correct vertex in multi vertex events have already been done allowing the new analyses
to remove the single interaction requirement. Work in progress includes search for
squarks and gluinos into leptons, search for charginos/neutralinos into dileptons, and
search for R-parity violating SUSY decays. The next year should bring many interesting
results and perhaps with ten times as much data some surprises may be in store. The
region beyond the Standard Model may be within in our grasp at the Tevatron.
References
1. X. Tata, The Standard Model and Beyond,p.304, ed. J. Kim, (World Scientific
1991)
2. S. Abachi et al., Nuclear Instruments and Methods A33S, 185 (1994).
3. H. Nilles, Phys. Reports 110, 1 (1984); P. Nath et al., Applied N-1 Supergravity,
(World Scientific 1984); H. Haber and G. Kane, Phys. Reports 117, 75 (1985).
4. R. Arnowitt and P. Nath, Phys. Rev. Lett. 69, 725 (1992).
5. S. Abachi et al., Phys. Rev. Lett. 75, 618 (1995).
6. G. Ross and R. G. Roberts, Nucl. Phys. B377, 571 (1992).
7. H. Baer, et al., Proc. of the Workshop on Phys. at Current Accel. and Super-
colliders 1993, p.703; Eds. J. Hewett, A. White and D. Zeppenfeld (Argonne
Nat. Lab, 1993).
8. F. Abe, et al., (CDF collaboration) Phys. Rev. Lett. 69, 343A (1992), P. Abreu,
et al., (DELPHI collaboration) Phys.Lett. B247, 148 (1990), T. Barkow, et al.}
(MARK II collaboration) Phys. Rev. Lett. 64, 2984 (1990), C. Albajar, et
al.} (UA1 collaboration) Phys. Lett. BI9S, 261 (1987), J. Alitti, et al.} (UA2
collaboration) Phys. Lett. B235, 363 (1990).
9. H. Baer, et al., Phys. Rev. 052, 1565 (1995).
10. F. Paige and S. Protopopescu, BNL Report 38304 (1986).
11. T. Medcalf, "The Search for Supersymmetry with the Aleph Detector at LEP";
P. Lutz, "SUSY with DELPHI"; R. Brown, "Searches for New Paricles in OPAL";
all in International Workshop on Supersymmetry and Unification of Fundamen-
tal Interactions, P. Nath ed., World Scientific, Singapore (1993).
12. S. Abachi et al., Phys. Rev. Lett. 74, 2632 (1995), F. Abe, et al., Phys. Lett.
74, 2626 (1995).
13. J. Ellis and S. Rudaz, Phys. Lett. 12SB, 248 (1983); A. Bouquet, J. Kaplan
and C. Savoy, Nucl. Phys. B262, 299 (1985).
14. H. Baer, et al., Phys. Rev. 044,725 (1991); H. Baer et al., Phys. Rev. 050,4517
(1994).
15. R. Akers, et al., (OPAL collaboration) Phys. Lett. B337, 207 (1994).
271
Formation of a Photosphere
Around Microscopic Black Holes
Andrew F. Heckler
NASA/Fermilab Astrophysics Center
Fermi National Accelerator Laboratory
Batavia, IL 60510
INTRODUCTION
As first shown by Hawking in 1975 [1], quantum theory predicts that a black hole
emits thermal radiation. The possibility of observing this thermal or "Hawking" ra-
diation from, say, a solar mass black hole is impractically small: the entire black hole
would emit only a few hundred quanta per second, and this is much too small of a flux
to possibly be observed at astronomical distances.
However, since the temperature TBH of the black hole, hence flux of radiation, is
inversely proportional to the black hole mass, the possibility of detecting Hawking ra-
diation from much smaller mass black holes becomes observationally feasible. Page and
Hawking, and several other authors [2, 3, 4] have placed upper limits on the density
of very small mass (therefore very hot) black holes by constraining the total radiation
produced to be less than the observed gamma ray background radiation. This method
constrains the density of black holes with temperatures of order lOOMeV. This particu-
lar number arises form the fact that a 100MeV black hole has a lifetime on the order of
the age of the universe, and although it is true that higher temperature black holes have
higher fluxes, they also have much shorter lifetimes, and the important quantity (for
background measurements) is the time integrated flux. Therefore 100MeV black holes
contribute the most to the gamma ray background. We should note here that since
there are no known astrophysical processes that can produce these small mass black
holes, these constraints all assume they were produced in the early universe, hence they
are called "primordial black holes" .
An important issue in finding the constraints on the density of black holes is to
determine the emission spectrum of the black hole. At first glance one might expect
the observed spectrum of radiation from a black hole to be thermal, since the black
hole emits thermal radiation from its surface (taking into account, of course, finite
size effects). MacGibbon and Webber [51 have shown that the observed spectrum is
not thermal simply because emitted particles such as quarks fragment into hadrons,
photons, neutrinos etc., and this fragmentation plays a major role in determining the
spectrum.
There is another possibility, however, for affecting the observed spectrum. If the
273
particles emitted from the surface interact with each other as they propagate away,
then the spectrum observed far away from the black hole will not be the same as the
emitted spectrum. In fact, if the particles interact strongly enough, then a photosphere
will develop around the black hole, and the average energy of the particles at the outer
surface of the photosphere will be much less than the average energy of particles emitted
directly from the black hole. A similar effect also occurs in the sun, where the surface
is much cooler than the central core, which produces the energy.
Previous authors have considered the possibility that the emitted particles do inter-
act [3, 6], and they use (perhaps too) simplistic arguments that the radiation emitted
from the black hole interacts too weakly to form a photosphere. In paticular, they con-
sider 2 - 2 body interactions, such as compton scattering and correctly show that these
are negligible. However, using standard QED, we will show that bremsstrahlung (and
photon-electron pair production) processes are important (2 - 3 body interactions),
and, for high enough black hole temperatures, the particles scatter and dramatically
lose energy as they propagate away from the black hole. The principle idea is that at rel-
ativistic energies, the bremsstrahlung cross section is roughly constant (independent of
energy). Since the density of emitted particles around the black hole increases with the
black hole temperature, there will be a temperature at which bremsstrahlung (and pair
production) scattering will become dominant. Although this scattering is not enough to
completely thermalize the emitted particles, a sort of near-thermal photosphere forms,
and the average energy of the particles decreases dramatically.
In this paper we will neglect all general relativistic effects (except for Hawking
radiation itself), because most of the interactions take place at a radius r rBH,
where rBH is the radius of the black hole.
BREMSSTRAHLUNG CROSS SECTION
For simplicity, let us first consider QED only. In a later section we will consider
the effects of QeD. If we take an energy averaged cross section (J = J w( d(J / dw )dw / E,
we can obtain an approximate expression for the cross section that is insensitive to the
well known infrared divergence of the bremsstrahlung cross section [7]
2 2E
(Jbrem ~ 8ar0 In - . (1)
me
This is in the center of mass frame and in the relativistic limit. Examination of the
function w( d(J / dw) reveals that the average energy lost in each collision is :'S E [8].
The interesting and well known behavior of the the relativistic bremsstrahlung (and
pair production) cross section is that it does not decrease with energy. However, one
must keep in mind that the cross section is large compared to the energy scale of the
particles, and there is a minimum interaction volume required in order for the process to
take place. We will see in the next section that the finite interaction volume is important
for two reasons: bremsstrahlung needs a finite amount of volume to occur, and any
particles within a separation m;l can interact via the bremsstrahlung process.
<"V
Note that along with the bremsstrahlung process (e + e - ewy), there is the similar
process of photon-electron pair production (e +l' - ee+e-). When we speak of brems-
strahlung, we will also tacitly include pair production because they have cross sections
with the same functional dependence at relativistic energies [7] shown in eq. (1).
ONSET OF PHOTOSPHERE
In a pure QED theory, electrons, positrons and photons are emitted from a black
hole that has a temperature TBH > mOe, where mOe is the vacuum electron mass. We
274
will assume that the energy spectrum of the particles emitted directly from the black
hole is perfectly thermal, though in reality there are spin and finite source size effects
[5]. We approximate the density of particles at radius r from the center of the black
hole:
J d3p ()r~H 1 TBH
no(r) :::::: 2 (27r)3f P -;:2:::::: 7r2 (47r)2r2
( )
2
where the radius of the black hole rBH = 1/(47rTBH ), f(p) = (expE(p)/TBH 1)-1 (for
fermions and bosons), and recall that the () angle should only be integrated to 7r/2,
since the particles are being emitted from a surface, the subscript 0 is to remind us
that we have not taken particle production from scattering into account. We will use
this approximation for density for both photons and fermions.
In calculating the density we must also take into account that bremsstrahlung and
electron-photon pair creation are particle creating processes. Therefore with each scat-
tering, more particles are being produced, and the particle density increases. Since the
combinations e + e -+ ee')' and e + ')' -+ ee+ e- are all possible, let us simplify the
picture and treat the electrons positrons and photons all as "particles" that undergo
2 -+ 3 body interactions: every two particles that interact create a third particle. Then,
for.N particle scatterings (see eq. (5)), we can account for the effect of particle creation
by replacing the density the term no by
3)N(r)
n(r) ="2
( no(r), (3)
which indicates that the density grows exponentially with each scattering.
The relativistic bremsstrahlung mean free path of electrons and positrons at radius
r is
1
'\(r) = n(r)ubremlvl
(4)
where we have approximated the vector velocity difference between colliding particles
to be Ivl : : : 1, and the average energy of the electrons E:::::: TBH .
Before proceeding, let us digress for a moment and discuss why one can approximate
Ivl : : : 1. For the sake of clarity, let us consider localized particles escaping from the black
hole with an energy E and a size roughly 1/E. The bremsstrahlung interaction involves
an exchange of a virtual photon between the particles and is an "action at a distance" .
The question is, what is the typical range of the interaction, or rather, how far apart
can the particles be and still interact? The answer is not 1/ E, rather the answer is that
the range of the interaction is l/k, where k is the the momentum exchanged between
the particles i.e. the momentum of the virtual photon. Using the same arguments used
in the last section, one can show that since the average angle bewtween the incoming
and out going particle is roughly mel E, the average momentum exchanged is '" me.
Therefore particles as far apart as l/me will interact via bremsstrahlung.
Now if a particle is emitted from the black hole of radius 1/(47rTBH ) and it can
interact with particles within a distance lime, then it is easy to see that for TBH ~ me,
the particle can interact with other particles all around the black hole. Therefore, if one
averages the vector velocity difference between the "colliding" particles, one can see that
this is similar to averaging over an isotropic distribution, and for relativistic particles
we can then approximate Ivl = 1. To summarize, localized particles a distance d apart
do not have to be "moving towards" each other to collide, the important quantity to
consider is the possible momentum exchanged k if they interact: if d < k- 1 then the
particles will interact, if d ~ k- 1 then the interaction will be negligible.
275
Let us define N to be the number of scatterings that an average particle has un-
dergone as it travels from r min to r max from the black hole,
(5)
Certainly if N > 1, the particles will begin to significantly interact, and possibly form
a photosphere.
We must be careful in using eq.(5) and (1), since we must take into account the
plasma mass of the electron [9]. Let us define the total electron mass in a plasma to be
(6)
where we have explicitly noted that the plasma mass is a function of density, temper-
ature and momentum. We will use the estimate of Ref.[9] that the plasma mass of a
particle near a black hole is
(7)
where we have used eq. (3) and we have assumed that since the average energy of
a particle emitted from a black hole is '" abTBH, where ab ~ 5 [5], and since the
average energy of the a particle decreases with each scattering, we estimate through
conservation of energy that E ~ aTBH/(3/2)N.
Combining eqs. (3), (4), and (5), we obtain a formula for number of bremsstrahlung
scatterings a particle undergoes travelling to a distance R from the black hole:
(8)
where we have explicitly shown the electron mass as a function of distance away from
the black hole, since the plasma mass varies with density.
Eq. (8) can be solved numerically, but we can also determine solutions analytically
by approximating (3/2)N ~ 1, and approximating the argument of the logarithm to
be TBH/mOe (because most of the integrand comes from the region where mpm '" mOe).
Integrating from rBH to infinity, we obtain
.t
JVbrem"""
,...., (a ba35 ) 1/2 (In TBH) TBH , (9)
47r mOe mOe
keeping in mind that this expression is valid only for N ~ 1. Using the definition
N(Tcrit) = 1, and approximating ab ~ 5, we obtain a critical temperature Tcrit for the
black hole
Tcrit '"
( (47r3/ab)1/2) mOe
In (a- 5/ 2 ) a 5/ 2 '" 45GeV. (10)
Numerically solving (8), we find that N brem = 1 when

Tcrit ~ 45.2GeV. (11)
At this temperature, outgoing particles will scatter on average at least once via brems-
strahlung and photon-electron pair production processes.]
276
THE THERMAL PHOTOSPHERE
We have shown that when a black hole rises above a critical temperature Tcrit ,
the Hawking radiation particles bremsstrahlung scatter many times as they propagate
away from the black hole. As N gets large, the radiation will have to be described in a
qualitatively different way. If the mean free path is short enough (.\ < r), then a kind
of photosphere will form, and a fluid description is more appropriate.
The unique environment of microscopic black hole photosphere, however, compli-
cates the issue: the average energy and density of the particles is changing so fast as
they propagate away from the black hole, that the particles never have enough time
to fully thermalize. That is to say, the fluid in the photosphere is an imperfect fluid.
Heckler [9] has shown that one can determine the approximate behavior of the imper-
fect fluid, and by applying the conservation of energy-momentum 8JlTJlV = 0, one can
show that the radius, temperature and gamma factor of the fluid element at the inner
surface of the photosphere is
411" 9
rO =
loa .IBH
2 4'7' ' and /0 = 8' (12)
where the "0" subscript denotes that these are values at the inner surface of the pho-
tosphere. Notice that the size and temperature of the photosphere are functions of the
black hole temperature T BH .
We can then determine T and / at larger radii by employing the conservation of
energy. Assuming u ~ 1, the conservation of energy requirement in the photosphere
becomes roughly
(13)
and with eq. (12) we find
Figure 1: A schematic of black hole and QED photosphere. For black holes with temperature
TBH > Tcrit ~ 45Ge V, the electrons, positrons, and photons emitted from the black hole travel
a distance'" (a 2TBH) -1 before they begin to bremsstrahlung scatter. Eventually, the particles
scatter enough that a near-perfect fluid photosphere forms at ro with some temperature To.
The temperature of the fluid decreases as it flows outward, and at T p , T = mOe and most of
the e annihilate. The photons then free stream to infinity-now with an average energy of
Eobs '" mOe(TBH/Tcrit)1/2.
277
T( r) = (:;~:;2) 1/3, 'Y( r) = ('YOO:~BHr r/3 (14)
The temperature of the photosphere decreases as the radius increases, and eventually
the temperature will decrease to mOe. This is the outer boundary of the photosphere
since most of the electrons and positrons will annihilate at this point, and the mean free
path will become so large that the particles will simply free stream to infinity. Using
eqs. (14) and (12) we find the parameters of the outer surface of the photosphere to be
'YOTBH)I/2 1
rp = (- - --,
47rmoe omOe
Tp = mOe,
and =0 ('YoTBH )1/2 (15)
'Yp 47rmOe
The average energy observed far away from the black hole (beyond the photosphere) is
roughly
_ (TBH)I/2 ( )
Eobs ~ ath'YpTp '" mOe -r.. ' 16
Crlt
where ath ~ 3 is obtained from a pure thermal plasma, as opposed to ab ~ 5, which is

obtained from the black hole spectrum.
Since the density of photons at rp in the rest frame of the black hole is '" 'Ypm~e, we
can estimate the total number of photons emitted by the black hole per second
. TBH (TBH) 1/2
Nphotons'" - 2
o
-r..
Crlt
' (17)
which is roughly (TBH /Tcrit ) 1/2 / 0 2 times greater than the particle emission rate directly
from the black hole.
THE BIG PICTURE
Now we can describe the whole picture of the photosphere which forms around a
hot black hole (see Figure 1). At black hole temperatures TBH < Tcrit '" 30GeV, there is
no scattering and no photosphere- the particles free stream away from the black hole,
and the far away observer sees the thermal spectrum directly emitted by the black hole
(we are neglecting the fragmentation of quarks etc [5]).
As the black hole temperature rises to TBH ;;:: Tcrit. bremsstrahlung and photon-
electron pair production begins to occur. The particles will begin to scatter (N", 1)
via bremsstrahlung and pair production at a radius of r = (S02TBH)-I, and as they
continue to scatter, they will eventually form a thermalized photosphere at a radius
ro '" 47r/(04TBH ). The characteristics of the photosphere are described in the previous
section, and represented in Figure l.
As the black hole continues to increase in temperature, the photosphere grows. For
low enough black hole temperatures, we can treat the photosphere as steady state-that
is the effect of the changing black hole temperature is negligible. Once the black hole
temperature increases to the point where its lifetime 'TBH is equal to the light crossing
time of the photosphere, that is when
(lS)
then the photosphere is no longer in a steady state. Beyond this point the black hole
will evaporate quicker than the photosphere dissociates, and there will be a period when
the there is a remnant photosphere, with no black hole in the center. Strictly speaking,
one must also include a factor of the relevant degrees of freedom such as quarks, gluons
278
TBH Eobs Tph TBH MBH
Photosphere 45GeV (02mOe )-1 108 sec 1011 gr
mOe
initially forms
TBH =10 TeV 10 TeV 10 MeV 10-5 cm 10 sec 109 gr
Tph = CTBH 109 GeV 1 GeV 10-3 cm 10- 13 sec 104 gr
Photosphere - 105 GeV 102cm - -

finally dissipates
Table 1: Characteristics of black hole and QED photosphere at various times. Note
that the average particle energy Eobs observed far from the black hole (beyond the
photosphere) is much smaller than the black hole temperature. Note also that when
the black becomes so hot that T ph = CTBH, the black hole quickly evaporates and leaves
behind a remnant photosphere, which eventually dissipates. All values are approximate.
etc.[5], but since we are only including the presence of photons, electrons and positrons,
we will neglect this factor. The lifetime Tbh of the black hole is then
M;I
TBH ~ 67',3 ' (19)
BH
where Mpl is the Planck mass.

The important features of the black hole and its photosphere are summarized in
Table 1 for various times during its ~volution. For the final moments of the black hole,
we are assuming a naive picture that the black hole simply evaporates into radiation,
with a maximum temperature of M pl .
INCLUDING QCD
The bremsstrahlung process also occurs in QeD. In this case, two quarks collide
and emit a gluon, and the coupling Os is much larger than the QED coupling. Adding
quarks, gluons and QeD interactions to the picture will therefore change the critical
temperature, size, and structure of the photosphere.
In order to show this, let us use a simple model of QeD. First of all, because of the
asymptotic freedom of QeD interactions, the average interparticle spacing must be less
than AQCD ,
(20)
in order for the coupling constant to be small enough that perturbation theory is valid.
When the interparticle spacing is larger than this, the quarks and gluons simply form
hadrons etc, which is described by MacGibbon and Webber [5].
For simplicity, let us assume that the QeD bremsstrahlung cross section has the
exact same form as the QED cross section, only one must now use the strong coupling
Os, and the mass of the quark involved m q ,
QCD 80~ I 2E
ubrem '" -2- n-. (21)
mq mq
Once again, we must take into account the plasma mass of the quark, and we do so
279
by following the same prescription as for the plasma mass of the electron (7). We also
will simplify the problem by estimating that the smallest mass of the quark is AQCD.
Therefore we estimate
(22)
This estimate will always insure that the important scale is the interparticle spacing,
and that the smallest energy scale is AQCD. Following the same procedure as in the
QED case, we estimate that the critical temperature at which quarks begin to QCD
bremsstrahlung scatter (i.e. NQCD = 1)
QCD A QCD
Tcrit '" ---sj2 ~ A QCD (23)
as
The true value of Tc~i~D is difficult to estimate, since as is such a sensitive function of
energy at these scales. Clearly the scale A QCD will be important, but a detailed model
of QCD is needed to obtain a more accurate number. Since Tc~i~D > Tc~i~D '" 100MeV,
it is also clear that QCD will play an important in the formation of a photosphere
around a microscopic black hole.
CONSEQUENCES FOR OBSERVATION
As stated in the introduction, the Page-Hawking limit, which constrains the total
flux of Hawking radiation from all of the evaporating black holes to be less than the ob-
served gamma ray background [2, 3, 4], most stringently constrains black holes emitting
100MeV photons. Since Tc~i~D '" 45GeV, the QED photosphere will not have a large ef-
fect on this constraint. Even as the temperature of the black hole increases to the point
where the photosphere begins to emit 100MeV photons again (when TBH '" (104)Tcrit),
the black hole will have such a small mass compared to its mass when it was a100MeV
black hole, that its contribution to the background (when it is converted to photons)
at this point will be negligible. Therefore the QED photosphere will have a negligible
affect on Page-Hawking limit.
However, since Tc~i~D '" AQCD '" 100MeV, the QCD photosphere may playa domi-
nant role in determining the number 100MeV photons emitted by a black hole. Clearly,
a full calculation including QCD interactions is needed in order to determine the energy
spectrum of particles coming from the photosphere. In general, one would expect an
added flux of particles starting at energies'" AQCD.
Another way of constraining the density black holes is based on the fact that we
have not, as far as we know, observed an individual black hole evaporating in its final
stages [3, 4, 13]. For example, Halzen et. al. [4] show, using the thermal radiation
plus quark fragmentation model, that if a a black hole with TBH = 100GeV, which has
a lifetime of about 107 seconds and radiates 100GeV photons is closer than about 1
parsec, then its emission will be above the 100 GeV background, and can be observed.
If we apply the effect of the photosphere to the standard constraints on the distance
individual black holes, we find that, since the QED photosphere decreases the energy
of the particles emitted from the black hole, observation becomes much more difficult
because the background is much higher at lower energies. To illustrate this, let us use
the example of the TBH = 100GeV black hole. The observer at infinity will only see
photons that have been processed through the photosphere, and as a result, will only
see photons with an average energy of Eobs '" mOe (TBH/Tcrit)l/2 '" 1MeV. In order to
conserve energy, the photon flux will consequently increase by a factor ofTBH / Eobs over
the standard no-photosphere assumption. But the the observed gamma ray background
flux is proportional to E-2.5 [3]. In order to see the 100GeV black hole, one must look
280
in the 1MeV energy range, where the background is much higher. Even though the
black hole flux is increased by a factor of TBH/ Eobs by the photosphere, this is not
nearly enough to compensate for the increase in background. Therefore, from this
point of view, the photosphere makes the observation of individual black holes much
more difficult, and the present limits must be reconsidered. That is to say, individual
primordial black holes may be a lot closer than the present constraints prescribe.
The presence of the photosphere will change the constraints on individual black
holes in several other important ways. If we consider observing the sky at some energy
Eobs, then for some range of energies mOe ~ Eobs ~ Tcrit there will be black holes at
two different temperatures which will both produce photons of average energy Eobs
and contribute to a signal. For example, if Eobs =10MeV, then a black hole with
TBH ..... 10MeV, which has no photosphere, and a black hole with TBH ..... 10TeV, which
has a (QED) photosphere will both radiate photons Eobs ..... 10MeV . The 10TeV black
hole, however, will emit photons at a much higher rate than the 10MeV black hole. In
particular, in the no-photosphere case, Eobs ..... TBH and the total photon flux is of order
Eobs; however with the photosphere, we find from eqs. (16,17) that the photon flux is
roughly
IV ..... (EObS)3 Tcrit (24)
Tcrit Q19f2
which is much greater than the flux from the lower temperature black hole.
However, one should note that the 10TeV black hole has a lifetime of only about
10 seconds, which severely limits the integration time of the observation. This exem-
plifies an important observational consequence of the photosphere. The photosphere
decreases the average energy of the particles to such an extent that very high energy
photons (Eobs) can only be produced by extremely high temperature black holes, which
have such extremely short lifetimes, that they are, practically speaking, unobservable
(see Table 1). This will dramatically weaken the constraints made by high energy
observations such as in Ref. [13].
Because the photosphere decreases the energy of the emitted particles, the possibil-
ity that black holes are the source of ultra-high energy background photons (or other
cosmic rays) seems remote. Even by including QCD and electro-weak theory, it would
be difficult to produce ultra high energy e or photons that would not be processed in
the photosphere. Of course, the black hole could emit other high energy particles such
as neutrinos, but even in this case a neutrino photosphere will eventually form, for hot
enough black hole temperatures.
CONCLUSIONS
The main purpose of this paper is to show that by using simple QED theory, one
can show that a photosphere does indeed form around a black hole, and that this can
have important observational effects. In order to understand the full consequences of
the photosphere, however, one must include QCD interactions, which will cause the
photosphere to form at lower black hole temperatures.
Nonetheless, even with the inclusion of QCD one might expect from our analysis that
the Page-Hawking limit on the density of primordial black holes will at most be affected
by only an order of magnitude or two. The observation of individual black holes, on the
other hand, will be dramatically affected by the photosphere. The photosphere makes
observing individual black holes much more difficult, and this opens up the possibility
that an individual black hole can be much closer than previous constraints prescribe.
I would like to thank Eric Braaten for discussions about the plasma mass, and I
would also like to thank Craig Hogan, Scott Dodelson, Rocky Kolb and Chris Hill for
281
many helpful discussions. This work was supported in part by the DOE and by NASA
(NAG5-2788) at Fermilab, and NAG5-2793 at the University of Washington.
References
[I] S.W. Hawking, Commun. Math. Phys. 43, 199 (1975).
[2] D.N. Page and S.W. Hawking, Astrophys. J. 206, 1 (1976).
[3] J.H. MacGibbon and B.J. Carr, Astrophys. J. 371, 447 (1991).
[4] F. Halzen, E. Zas, J.H. MacGibbon & T.C. Weekes, Nat. 353, 807 (1991).
[5] J. H. MacGibbon and B.R. Webber, Phys. Rev. D4l, 3052 (1990).
[6] J. Oliensis and C.T. Hill, Phys. Lett. B238, 492 (1984).
[7] J.M. Jauch and F. Rohrlich, The Theory of Electrons and Photons, (Springer-
Verlag, New York, 1975).
[8] E. Haug, Z. Naturforsch, 30a, 1099 (1975).
[9] A.F. Heckler, Fermilab-Pub-95/059-A, astro-ph/9601029, submitted to Phys.

Rev. D
[10] L.D. Landau and 1. Pomeranchuk, Dokl. Akad. Nauk SSSR 92, 553 (1953); A.B
Migdal, Phys Rev., 103, 1811 (1956); E.L. Feinberg and 1. Pomeranchuk, Nuovo
Cimento, Supplement to Vol. 3, 652 (1956).
[11] W.A. Weldon, Phys. Rev. D26, 2789 (1982); J.F. Donoghue and B.R. Holstein,
Phys. Rev. D28, 340 (1983); J. 1. Kapusta, Finite-temperature field theory, (Cam-
bridge University Press, Cambridge, 1989). See also the appendix of E. Braaten
and D. Segel, Phys. Rev. D48, 1478, (1993).
[12] J. Goodman, Ap. J. 308, L47 (1986).

[13] D.E. Alexandreas et al., Phys. Rev. Lett. 71, 2524 (1993).
[14] A.D. Sakharov, JEPT Lett. 5, 24 (1967); A.G. Cohen, D.B. Kaplan, and A.E.
Nelson, Ann. Rev. Nuc. Par. Sci. 43, 27 (1994).
282
A supersymmetric Model
for Mixed Dark Matter
Antonio Riotto
Fermi National Accelerator Laboratory
Batavia, IL 60510
The idea that the large scale structures seen today evolved from very small primeval
density inhomogeneities has been strenghtened by the recent detection of large scale
anisotropies in the cosmic microwave background [1]. Nevertheless, one of the necessary
ingredients for the structure formation, namely the nature of dark matter, remains
unknown.
The most satisfactory model for structure formation is perhaps the cold dark matter
(CDM) theory [2] where the Universe is assumed to be spatially flat (0 = 1) and
with tv 0.9 of the mass density formed by CDM particles, cold in the sense that they
decouple from the expanding thermal bath at temperatures much smaller than their
mass. CDM can successfully explains galaxy-galaxy and cluster-cluster correlation
functions on scales of order of 1-5 Mpc. However, it now appears to be inconsistent
with large scale structure data as the automatic plate machine (APM) galaxy survey
[3], which suggest more power on large scales than the standard CDM predictions. On
small scales, the observed pairwise velocity dispersion for galaxies appears to be smaller
than those predicted by CDM [4].
One alternative is the hot dark matter model (HDM). HDM is taken to be a light
neutrino, which decouples from the thermal bath when still relativistic, with m" =
(920" h2 ) eV where H = 100 h Km/sec Mpc is the Hubble parameter. In HDM the
processed fluctuation is characterized by the distance a neutrino travels over the history
of the Universe, A" ~ 40 (30 eV /m,,) Mpc. The problem with HDM is that A" is too
large with respect to the scale which is just becoming nolinear today, tv 5 h- 1 Mpc. If
galaxy formation occurs early enough to be consistent with high-redshift galaxies and
quasars, then structures on 5 h- 1 Mpc will overdevelop.
The hope is that cold + hot dark matter (C + HDM) will combine the success of
both models. Indeed (C + HDM) models with OCDM ~ 0.6, 0" ~ 0.3, Obaryon ~ 0.1
and a Hubble constant h ~ 0.5 has the best fit for microwave anisotropy data, large
scale structure surveys, and measures of the bulk flow with a few hundred megaparsecs [5l.
Even if C + HDM is appealing for the large scale structure phenomenology, it might
seem rather unpalatable from the point of view of particle physics for the following
283
reasons. It is well-known that low-energy supersymmetric theories provide an elegant
solution to the hierarchy problem [6]. The lightest supersymmetric particle (LSP) is
the most attractive candidate for CDM and is made stable by imposing a discrete
symmetry, called R-parity. Nevertheless, in the minimal supersymmetric standard
model (MSSM), neutrinos are not massive and therefore there are no candidates for
HDM. If one considers, for instance, minimal extension of MSSM by including a right-
handed neutrino superfield V C and a singlet field S with lepton number L(S) = +2
[7], the U(l)B-L and R-parity symmetries get spontaneously broken, neutrinos acquire
a mass, but the LSP is no longer stable and cannot play the role of CDM. So, when
trying to build up a supersymmetric model with both CDM and HDM components,
one has to face the problem of making neutrinos massive and, at the same time, the
LSP stable. Mo~eover, even if one is successful on this way, the relative abundances of
light massive neutrinos and LSP's are expected to be set by two uncorrelated scales,
namely the U(l)B-L breaking scale, VBL , and the supersymmetric breaking scale, M s ,
and a sort of fine-tuning on VBL and Ms must be done to satisfy the requirements of
C + HDM model.
In this talk we give an example of extension of MSSM where it is possible to im-
plement C + HDM model for large scale structure formation and where no fine-tuning
on the scales VBL and Ms is necessary. Indeed, the relative abundances of the HDM
and CDM components are set by the same scale VBL ~ 103 GeV, which is quite likely
to manifest itself in rare decay processes.
The model [8] is a sort of extension of the original proposal given in ref. [9J and has
been recently studied in ref. [10]. The superpotential can be written as the sum of two
terms
W=WO +W1 , (1)
where Wo is the usual MSSM piece and [8]
WI = vLH2v c + jVCVCSI +.x (SIS2 - M2) Z. (2)

Here M is an explicit mass scale of order of VBL ; V is a right-handed neutrino superfield
C
which is a singlet under SU(2h @ U(1)y and carries lepton number L = -1; S1, S2
and Z are SU(2h @ U(1)y singlet superfields as well and carry lepton numbers L = 2,
-2 and 0, respectively.
It easy to work out from eqs. (1) and (2) the full scalar potential [6]. As the tem-
perature falls down below the value T ~ VBL , the U(l)B-L symmetry is spontaneously
broken by the following vacuum expectation values (VEV's)
(3)
whereas it is easy to show that Z acquires a VEV
(4)
The crucial point here is that, whenever VBL > M s , the right-handed sneutrino i/C does
not acquire a VEV so that the discrete R-parity symmetry, which is proportional to
( _l)L, is preserved. As a consequence, the LSP in our model remains stable and can
provide a suitable candidate for CDM.
After the spontaneous breaking of the U(l)B-L symmetry, a Nambu-Goldstone
boson, the majoron, will appear
284
Vi (1m Sl) - V2 (1m S2)
(5)
J= JVl+Vi '
whereas its fermionic superpartner, the majorino, and its real superpartener, the sma-
joron, will acquire a mass proportional to Ms. In particular, the majorino 'ljJJ receives
two different mass contributions: one, at the tree level, of order of A(Z}, and the second
from one-loop diagrams involving i7 and lIC of order of UMs /167r 2). Without any fine-
tuning of the parameters, we can have m.pJ ~ (10 - 50) GeV and consider the majorino
lighter than any other neutralino. As a consequence, the majorino can be taken the
LSP and a suitable candidate for CDM. The HDM component will be provided by the
light tau neutrino which, after the breaking of SU(2h U(I)y, will acquire through
the see-saw mechanism, a mass of order of m v , ~ (h~viw/fVBd, where VEW stands
for the scale of SU(2h U(I)y breaking. We now proceed to the estimation of the
relic abundance of majorinos.
For T < VBL all the heavy fields, such as lIc , decay into lighter particles so that
the majorino, as well as the smajoron and the majoron, can only interact among each
other and with the "standard model" particles, such as leptons and quarks, through
the coupling hvLHll1c. Therefore, the coupling of 'IjJ~s and J's to the "standard model"
particles is suppressed by powers of E ~ U2h~/167r2) and they decouple from the
"standard model" thermal bath at a temperature [10]
~
T,D ~ 104 VBL ( 2.!:..
)1/3 (1O-6f2)2/3
_.- - GeV. (6)
Mp1 E
To get a feeling of the numbers, T,D ~ 102 GeV for VBL ~ 103 GeV. Therefore, the
majorino is expected to decouple from the "standard model" thermal bath when still
relativistic. Nevertheless, its number density for T < T,D does not decrease only due
to the expansion of the Universe. Indeed, the key point here is that, even after T,D,
the number density of majorinos continue to follow its equilibrium value due to the
fact that they keep into equilibrium with the thermal bath formed by majorons via
interactions mediated by heavy particles with mass rv VBL such as ReZ or the two
fermionic combinations of 'ljJl, 'ljJ2 and 'ljJz orthogonal to 'ljJJ.If we indicate with TJ the
temperature of the thermal bath formed by majorons and majorinos and T, the one
relative to the "standard model" particles, for T, < T,D we do have TJ = a(T,)T,
where
(7)
takes into account the various annihilation thresholds for massive "standard model"
particles and g.s(T,) counts the effective relativistic degrees of freedom contributing to
the entropy density.
It is easy to show that majorinos can annihilate into a pair of majorons only through
a p-wave and that the today contribution of majorinos to the 0 parameter is then given
by [8]
(8)
where T"tod ~ 2.75 K is the today temperature of the relic photons.

Taking h ~ 0.5 and O.pJ ~ 0.6 as suggested by the latest C + HDM simulations, we
get ao ~ 10- 14 GeV-2. Since a detailed calculation of ao with standard techniques [11]
285
gives, for oX ~ 0.5, 0"0 ~ 0.1(M1m~)VA2), we obtain that majorinoscan form the CDM
component of the Universe for a relativley small value of VBL , VBL ~ 103 GeV. Such a
small value of VBL is quite likely to manifest itself in rare decay processes, whereas it
would be rather difficult to detect the CDM component either through direct searches
or through indirect detection of annihilation products of majorinos that annihilate in
the Sun, in the Earth or in the galactic halo since the majorino is very weakly coupled
to matter. In conclusion, we have given an example of extension of MSSM where it is
possible to implement the C + HDM scenario for large scale structure formation. In
particular, we have shown that the relative abundances of CDM and HDM components
in the Universe can be set by the same scale, the U(l)B-L breaking scale.
References
[1] G.F. Smoot et al., Astrophys. J. Lett. 396 (1992) Ll.
[2] For a recent review, see M. Davis, G. Efstathiou, C.S. Frenk and S.M. White,
Nature (London) 356 (1992) 489.
[3] G. Efstathiou, W.J. Sutherland and LJ. Madrox, Nature (London) 348 (1990)
705.
[4] M. Davis, P.J. Peebles, Astrophys. J. 267 (1983) 465; M. Davis, G. Efstathiou,
C.S. Frenk and S.M. White, ibidem 292 (1985) 37l.
[5] Q. Shaft and F.W. Stecker, Phys. Rev. Lett. 53 (1984) 1292; R.K. Shaefer, Q.
Shaft and F.W. Stecker, Astrophys. J. 347 (1989) 575; J.A. Holtzman, Astrophys.
J. Suppl. 71 (1989) 1; E. Wright et al. Astrophys. J. Lett. 396 (1992) L13.
[6] See, for instance, H. Nilles, Phys. Rep. 110 (1984) 1 and references therein; H.E.
Haber and G.L. Kane, Phys. Rep. 117 (1985) 75 and references therein.
[7] G.F. Giudice, A. Masiero, M. Pietroni and A. Riotto; Nucl. Phys. 396 (1993) 243;
M. Shirashi, I. Umemura and K. Yamamoto, Phys. Lett. 313 (1993) 89.
[8] R.N. Mohapatra and A. Riotto, Phys. Rev. Lett. 73 (1994) 1324.
[9] C. Aulakh and R. Mohapatra, Phys. Lett. B119 (1982) 136.
[10] R.N. Mohapatra and X. Zhang, UMDHEP preprint 94-04 (1993).
[11] K. Griest, M. Kamionkoski and M.S. Turner, Phys. Rev. D41 (1990) 3565.
286
Light photinos and supersymmetric dark matter
Edward W. Kolb
Fermi National Accelerator Laboratory, Batavia, IL 60510, and
Department of Astronomy and Astrophysics, Enrico Fermi Institute
The University of Chicago, Chicago, IL 60637
1 Supersymmetric relics
1.1 Supersymmetry and Supersymmetry breaking

There are two fundamental reasons for believing that nature is supersymmetric. The
first reason is that supersymmetry can rescue the standard electroweak model from the
embarrassment of finely tuned coupling constants. The standard electroweak model em-
ploys fundamental scalars, usually refereed to as "Higgs" scalars, to break the gauge
symmetry spontaneously. But scalar particles have very bad ultraviolet behavior, which
has the effect of dragging the electroweak Higgs mass up to the mass scale of any encom-
passing theory, such as a grand-unified theory. Thus, unless coupling constants are very
finely tuned or some other dynamics enters the picture, light scalar masses (of order the
electroweak scale) would not be possible. Supersymmetry (SUSY) is an example of "some
other dynamics." Because of the relative factor of -1 between fermionic and bosonic
loops, the addition of fermionic loops can mitigate the bad ultraviolet behavior of scalar
loops. The way to realize this possibility is if for every boson there is a corresponding
fermion appearing in the calculation of the quantum corrections to the Higgs mass.
This correspondence between fermions and bosons implies that both fermions and
bosons appear in multiplets, and they are transformed into each other by supersymmetry
transformations. Thus, SUSY is intimately related to Poincare symmetry. In fact, the
commutator of SUSY transformations generates the momentum operator. SUSY is the only
know way to unify spacetime and the internal symmetries of the S-matrix. Thus, SUSY
seems to be a fundamental part of any attempt to unify gravity with the fundamental
forces. This aesthetic reason is the second motivation for SUSY.
The particular realization of SUSY I will consider is the supersymmetric extension
of the standard model. Although this model, the minimal supersymmetric standard
model (MSSM), has many parameters in addition to the plethora of parameters of the
non-symmetric version of the standard model, it is sufficiently restrictive to have some
predictive power.
287
Of course in nature SUSY is broken-there is no massless fermionic photon for in-
stance. I will return to the question of SUSY breaking in a moment. But the first
relevant issue for SUSY dark matter is the existence of something known as R-parity, a
discrete multiplicative symmetry. The R-parity of a particle is given in terms of its spin
S, baryon number B and lepton number L, by R = (_1)3(B-L)+2s. Known particles
all have even R-parity, while their SUSY partners are all R-odd particles. If R-parity
is conserved, then the lightest R-odd color singlet (LROCS) must be stable-and hence
a candidate for dark matter. There are inconveniences with any theory if R-parity is
broken. Rapid proton decay, for instance. If one works hard enough these difficulties
can be overcome, but the assumption of exact R-parity is very attractive and naturally
leads to dark matter candidates. So I will assume exact R parity.
Now let's return to the issue of SUSY breaking. The details of SUSY breaking will
determine the identity of the LSP, as well as its mass and interaction strength. Un-
fortunately, essentially nothing is known about SUSY breaking. The only reasonable
constraint one might imagine is that any susY-breaking Lagrangian terms must have
mass dimension less than four. 1 Since we have no other choice, let's consider all possi-
ble dimension-two and dimension-three susY-breaking terms consistent with gauge and
Lorentz symmetry:
il(SUSY BREAKING) = -mi1H112 - m~IH212 - mi2(HI H 2 + H;H;)

-t 2 - -t 2 - :Jt 2 - -t 2 - -t 2-
-QL;(Mi)ijQLj - URi(M-vJijURj - dRJMJ)ijdRj - LdMz)ijLLj - eRi(Me)ijeRj
-Hi:JLi(huAu)i/URj - HlJLi(hdAd)ijdRj - HILi(heAe)ijeRj
1 =- 1 =- 1 -=--
--MI B B - -M2W aW a - -M3GaGa. (1)
2 2 2
The tilde superscript denotes the SUSY partner of familiar particles: e is the selectron, d
is the down squark, Ga are the gluino fields, Wa and B are the supersymmetric partners
of the familiar SU(2) and U(l) gauge fields, and Q and are SU(2) doublets containing
the SUSY partners of left-handed quarks and leptons. The fields HI and H2 are the two
Higgs necessary in SUSY. The parameters M.2 and A.. are 3 x 3 symmetric matrices.
The matrix A.. has mass dimension one. Note that the operators appearing in the first
two lines of (1) are operators of mass dimension two, while the last two lines contain
operators of mass dimension three.
The usual procedure is to choose a set of parameters including the constants appearing
in (1), requiring that the resulting low-energy theory leads to the usual standard model.
The choice of these mass parameters, along with the Higgsino mass parameter jL, results
in a mass matrix for the neutralinos: the bino Ii, the zino W3, and the Higgsinos and Hr
Hg. In terms of the mass of the Z, the weak mixing angle Bw , and tan (3 (the ratio of the
vacuum expectation values of the two Higgs fields responsible for electroweak symmetry
1 Dimension-four susY-breaking terms suffer from the bad ultraviolet behavior we are trying to fix.
288
breaking), the neutralino mass matrix in the basis (B, W 3 , HP, Hg) is given as
Ml o -mz cos (3 sin ew mz sin (3 sin ew )

( M2 mz cos (3 cos ew -mz sin (3 cos ew .
-mz co~ (3 sin ew mz cos (3 cos ew o -fJ,
(2)
mz sin (3 sin ew -mz sin (3 cos ew -fJ, o
The susY-breaking masses Ml and M2 are commonly assumed to be of order mz or larger,

and if the SUSY model is embedded in a grand unified theory, then 3Md M2 = 501d 012.
If we assume the relation between Ml and M 2, then there are three parameters in the
neutralino mass matrix: fJ" tan (3, and Ml/2 (the zino-bino mass parameter).
Now the game is set: for a given set of parameters, diagonalize the mass matrix,
find the mass of the lightest supersymmetric particle and its field content (of course in
general it is a linear combination of the four neutralino fields), determine its annihilation
cross section, and put the above information into the freeze-out machinery to determine
the relic abundance.
Different groups who study the problem come up with slightly different composite
sketches for the dark matter suspect. Some believe the particle content is mostly Hig-
gsino, while some find mostly bino. However just about all groups find a relatively large
mass for the suspect, between 30 and several hundred GeV.
In this part of the talk 2 I would like to propose a different picture for the wanted
poster: a particle of low mass (500 MeV to 1.6 GeV) and "photino-like," with an SU(2) x
U(l) content almost identical to the photon.
The motivation for this light photino comes from our lack of knowledge about SUSY-
breaking. Referring to the susY-breaking terms in (1) we see that there are dimension-
two and dimension-three terms. There are theoretical reasons to believe that dimension-
three terms might be much smaller than the dimension-two terms. It appears difficult
to break SUSY dynamically in a way that produces dimension-three terms while avoiding
problems associated with the addition of gauge-singlet superfields. In models where
SUSY is broken dynamically or spontaneously in the hidden sector and there are no
gauge singlets, all dimension-three susY-breaking operators in the effective low-energy
theory are suppressed compared to susY-breaking scalar masses by a factor of (Cf!)/mpl,
where (Cf!) is the vacuum expectation value of some hidden-sector field and mpl is the
Planck mass. Thus, dimension-three terms may not contribute to the low energy effective
Lagrangian. This would imply that at the tree level the gluino is massless, and the
neutralino mass matrix is given by (1) with vanishing (00) and (11) entries. However,
non-zero contributions to the gluino mass and the neutralino mass matrix come from two
sources: radiative corrections such as the top-stop loops for the gluino and neutralinos,
and "electroweak" loops involving higgsinos and/or winos and bin os for theneutralinos
(but not for the gluino).
2Reference to all the material presented here can be found in Farrar and Kolb (1996) and Chung,
Farrar, and Kolb (1996)
289
The generation of radiative gaugino masses in the absence of dimension-three SUSY
breaking was studied by Farrar and Masiero. They found (taking M ~ 40 GeV) that as
the typical susY-breaking scalar mass, Mo, varies between 100 and 400 GeV, the gluino
mass ranges from about 700 to about 100 MeV, while the photino mass ranges from
around 400 to 900 MeV. This estimate for the photino mass should be considered as
merely indicative of its possible value, since an approximation for the electroweak loop is
strictly valid only when M and M o are much larger than mw. The other neutralinos are
much heavier, and the production rates of the photino and the next-lightest neutralino
in ZO decay are consistent with all known bounds.
The conclusion is that light gluinos and photinos are quite consistent with present
experiments, and result in a number of striking predictions. One prediction is that the
photino i should be the relic R-odd particle, even though it may be more massive than
the gluino. This is because below the confinement transition the gluino is bound with a
gluon into a color-singlet hadron, the RO, whose mass (which is in the 1 to 2 GeV range
when the gluino is very light) is greater than that of the photino. In this scenario, LSP
is an ambiguous term: the gluino is lighter than the photino, although the photino is
lighter than the RO. As discussed above, a more relevant term would be LRocs-lightest
R-odd color singlet.
However, models with light gauginos were widely thought to be disallowed because it
had been believed that in such models the relic density of the lightest neutralino would
exceed cosmological bounds unless R-parity would be violated allowing the relic to decay.
In the next subsection I will rehash that argument, and then point out how the restriction
can be evaded if the RO mass is close to the i mass (here "close" means within a factor
of two).
1.2 Self-annihilation and freeze out

The reaction rates that determine freeze out will depend upon the i and RO masses, the
cross sections involving the i and RO, and possibly the decay width of the RO as well. In
turn the cross sections and decay width also depend on the masses of the i, 9 and RO,
as well as the masses of the squarks and sleptons. We will denote the squark/slepton
masses by a common mass scale (expected to be of order 100 GeV). Even if the masses
were known and the short-distance physics specified, calculation of the width and some
of the cross sections would be no easy task, because one is dealing with light hadrons.
Fortunately, our conclusions are reasonably insensitive to individual masses, lifetimes,
and cross sections, but depend crucially upon the RO-to-i mass ratio, denoted by r.
When we do need an explicit value of the photino mass m, the RO mass M, or the
squarkjslepton mass M s' we will parameterize them by the dimensionless parameters
Ms, r, and Ms:
m = 800Ms MeV; M=rm; Ms = lOOMs GeV. (3)
The standard procedure for the calculation of the present number density of a thermal
relic of the early universe is to assume that the particle species was once in thermal
equilibrium until at some point the rates for self-annihilation and pair-creation processes
290
became much smaller than the expansion rate, and the particle species effectively froze
out of equilibrium. After freeze out, its number density decreased only because of the
dilution due to the expansion of the universe.
Since after freeze out the number of particles in a comoving volume is constant, it is
convenient to express the number density of the particle species in terms of the entropy
density, which in a comoving volume element is also constant for most of the history
of the universe. This number-density-to-entropy ratio is usually denoted by Y. If a
species of mass m is in equilibrium and nonrelativistic, Y is given simply in terms of
the mass-to-temperature ratio x == m/T as YEQ(x) = 0.145(g/g.)x 3 / 2 exp(-x), where
9 is the number of spin degrees of freedom, and g. is the total number of relativistic
degrees of freedom in the universe at temperature T = m/x. Well after freeze out Y(x)
is constant-we will denote this asymptotic value of Y as Yoo.
If self annihilation determines the final abundance of a species, Y00 can be found by
integrating the Boltzmann equation (dot denotes d/dt) n+3Hn = -(lvlaA) (n2 - n~Q)'
where n is the number density, nEQ is the equilibrium number density, H is the expansion
rate of the universe, and (IvlaA) is the thermal average of the annihilation rate.
There are no general closed-form solutions to the Boltzmann equation, but there
are reliable, well tested approximations for the late-time solution, i.e., Yoo. Then with
knowledge of Yoo, the contribution to 0.h 2 from the species can easily be found. Let us
specialize to the survival of photinosassuming self-annihilation determines freeze out.
Calculation of the relic abundance involves first calculating the freeze-out value of
x, known as x I, where the abundance starts to depart from the equilibrium abun-
dance. Using standard approximate solutions to the Boltzmann equation gives XI =
In(0.0481mplmaO) - 1.5ln[ln(0.0481mplmao)], where we have used 9 = 2, g. = 10, and
parameterized the nonrelativistic annihilation cross section as (IvlaA) = aox- 1 . Using the
diagram shown in Figure 1, ao = 2x10-11J.t~J.ts4mb, which leads to xI:::::: 12.3+ln(J.tVJ.t~).
The value of XI determines Yoo:
(4)
Once Yoo is known, the present photino energy density can be easily calculated:
= mn::y = 0.8J.tsGeV Y00 2970cm- 3 . When this result is divided by the critical
fJ'::y
density, Pc = 1.054h2 X 10-5 GeV cm- 3 , the fraction of the critical density contributed
by the photino is found to be 0.::yh 2 = 2.25 X lOsJ.t sYoo For Yoo in (4), 0.::yh 2 = 167J.t82J.t~.
The age of the universe restricts 0.::yh2 to be less than one, so for J.ts = 1, the photino
must be more massive than about 10 GeV or so if its relic abundance is determined by
self-annihilation.
1.3 RO-catalyzed freeze out

Farrar and I pointed out that for models in which both the photino and the gluino
are light, freeze out is not determined by photino self annihilation, but by ::y ~ R O
interconversion. The basic point is that since the R O has strong interactions, it will
stay in equilibrium longer than the photino, even though it is more massive. As long as
291
;;y +----+ RO interconversion occurs at a rate larger than H, then through its interactions
with the RO, the photino will be able to maintain its equilibrium abundance even after
self annihilation has frozen out.
Griest and Seckel discussed the possibility that the relic abundance of the lightest
species is determined by its interactions with another species. They concluded that the
mass splitting between the relic and the heavier particle must be less than 10% for the
effect to be appreciable. We find that ;;y +----+ RO interconversion determines the ;;Y relic
abundance even though the RO may be twice as massive as the ;;Yo The difference arises
because Griest and Seckel assumed that all cross sections were roughly the same order
of magnitude. But in our case the RO annihilation is about 10 12 times larger than other
relevant cross sections.
I will now consider in turn the reactions we found to be important in our scenario.
The diagrams for the individual constituent processes can be found in Figure 1. However,
as we shall see, it is not a simple task to go from the constituent diagrams to the cross
sections and decay width.
;;Y;;Y ---+ X: For photino self-annihilation at low energies the final state X is a lepton-
antilepton pair, or a quark-antiquark pair which appears as light mesons. The process
involves the t-channel exchange of a virtual squark or slept on between the photinos,
producing the final-state fermion-antifermion pair. (See the upper third of Figure 1.)
In the low-energy limit where the mass Ms of the squark/slepton is much greater than
vis, the photino-photino-fermion-antifermion operator appears in the low-energy theory
with a coefficient proportional to ell Ml, with ei the charge of the final-state fermion.
Also, as there are two identical fermions in the initial state, the annihilation proceeds as
a p-wave, which introduces a factor of v 2 in the low-energy cross section. The resultant
low-energy photino self-annihilation cross section is:
(5)
where we have used for the relative velocity v 2 = 6/x with x == miT, and qi is the
magnitude of the charge of a final-state fermion in units of the electron charge. For the
light photinos we consider, summing over e, fJ" and three colors of u, d, and s quarks
leads to L qt = 8/3.
RO RO ---+ X: In RO self-annihilation, at the constituent level the relevant reactions are
'9+'9 ---+ g+g and '9+'9 ---+ q+ij (see the middle third of Figure 1), which are unsuppressed
s'
by any powers of M so the cross section should be typical of strong interactions. In the
nonrelativistic limit we expect the RO RO cross section to be comparable to the pp cross
section, but with an extra factor of v 2 , accounting for the fact that there are identical
fermions in the initial state so annihilation proceeds through a p-wave. There is some en-
ergy dependence to the pp cross section, but it should be sufficient to consider (Ivl(JRORO)
to be a constant, approximately given by (Ivl(JRO RO) c::: 100v 2 mb = 600 x- 1 r- 1 mb, where
we have used for the relative velocity v 2 = 6TIM = 6/(rx), with x == miT.
We should note that the thermal average of the cross section might be even larger
if there are resonances near threshold. In any case, this cross section should be much
larger than any cross section involving the photino, and for the relatively small values of
r we employ it will ensure that the RO remains in equilibrium longer than the ;;y, greatly
292
yxq,l
~ y q, I
y 2
e M_
-2
S
q, T
y~q
q~!i
s s
Figure 1: Feynmann diagrams for the constituent processes determining the relic photino
abundance. The top left diagram is for ;Y self annihilation, and on the top right is the
effective low-energy operator for that process. The two diagrams in the middle are the
diagrams for RO self annihilation. Finally in the lower left-hand corner is the diagram for
the ;Y f--+ RO interconversion processes (all interconversion processes can be obtained
from crossings of this diagram), and on the lower right-hand side is the effective low-
energy operator.
simplifying our considerations.

;YRo -+ X: This is an example of a phenomenon known as co-annihilation whereby
the particle of interest (in our case the photino) disappears by annihilating with another
particle (here, the RO). Of course co-annihilation also leads to a net decrease in R-
odd particles. We can estimate the cross section for ;YRo -+ X in terms of the ;Y self
annihilation cross section by comparing the lower third of Figure 1 to the upper third:
(6)
where the ratio of a's arises because the short-distance operator for co-annihilation is
proportional to erg} rather than e;, the second factor is the color factor coming from
the gluino coupling, and the third factor comes from the ratio of Li q; / q;
Li for the
participating fermions. We have replaced m 2 appearing in (5) by mM, although the
actual dependence on m and M may be more complicated. Finally, the annihilation is
s-wave so there is no v 2 /3 suppression as in photino self-annihilation.
Although the short-distance physics is perturbative, the initial gluino appears in a
light hadron, and there are complications in the momentum fraction of the RO carried
by the gluino and other non-perturbative effects. For our purposes it will be sufficient
to account for the uncertainty by including in the cross section an unknown coefficient
A, leading to a final expression
293
(7)
RO -+ 1'1f+1f-: In what we call interconversion processes, there is an R-odd particle

in the initial as well as in the final state. Although the reactions do not of themselves
change the number of R-odd particles, they keep the photinos in equilibrium with the
ROs, which in turn are kept in equilibrium through their self annihilations. An example
is RO decay. It can occur via, e.g., the gluino inside the initial RO turning into an
antiquark and a virtual squark, followed by squark decay into a photino and a quark. In
the low-energy limit the quark-antiquark-gluino-photino vertex can be described by the
same type of four-Fermi interaction as in co-annihilation (see Figure 1). One expects on

dimensional grounds a decay width f oc (XEM(XsM 5 / M/. The lifetime of a free gluino
to decay to a photino and massless quark-(anti)quark pair was computed by Haber and
Kane. However this does not provide a very useful estimate when the gluino mass is less
than the photino mass.
In an attempt to account for the effects of gluino-gluon interactions in the RO, which
is necessary for even a crude estimate of the RO lifetime, Farrar developed a picture
based on the approach of Altarelli et al.: The RO is viewed as a state with a massless
gluon carrying momentum fraction x, and a gluino carrying momentum fraction (1- X),3
Table I: Cross sections and the decay width used in the calculation of the relic photino abundance.
The dimensionless parameters J.Ls and J.Ls were defined in (3), and F(r) was defined below (9). The
coefficients A, B, and C reflect uncertainties involving the calculation of hadronic matrix elements.
Process Cross section or width

RO self annihilation: (IvluRO RO) 600 x- 1 r- 1 mb
::; self annihilation: (Iv Iu-y) 2.0 x 10- 11 x- 1 [J.L~J.Ls4] mb
co-annihilation: (Ivlu::YRO) 1.5 x 10- 10 r [J.L~J.Ls4 A] mb
RO decay: r R--+,1T7r
- 2.0 x 10- 14 F(r) [J.L~J.Ls4 B] GeV
::; - RO conversion: (IvluRU or ) 1.5 x 10- 10 r [J.L~J1s4C] mb
having therefore an effective mass Mvr=;r. The gluon structure function F(x) gives
the probability in an interval x to x + dx of finding a gluon, and the corresponding
effective mass for the gluino. One then obtains the RO decay width (neglecting the mass
of final state hadrons):
f(M, r) = fo(M, 0) t- r
-
2
dx (1 - X)5/2 F(x) f(1/rvT=X) , (8)
where in this expression the factor fo(M, 0) is the rate for a gluino of mass M to decay
to a massless photino, and f(y) = [(1 - y2)(1 + 2y - 7y2 + 20y3 - 7y4 + 2y 5 + y6) +
24y3(1 - Y + y2) log y] contains the phase space suppression which is important when
30f course there should be no confusion with the fact that in the discussion of the RO lifetime we use
x to denote the gluon momentum fraction whereas throughout the rest of the paper x denotes miT.
294
the photino becomes massive in comparison to the gluino. Modeling K decay in a
similar manner underestimates the lifetime by a factor of 2 to 4. This is in surprisingly
good agreement; however cautio)1 should be exercised when extending the model to RO
decay, because kaon decay is much less sensitive to the phase-space suppression from the
final state masses than the present case, since the range of interest will turn out to be
r rv 1.2 - 2.2. For r in this range, taking F(x) rv 6x(1 - x) leads to an approximate
behavior
(9)
where F(r) = r 5 (1- r- 1 )6, and the factor B reflects the overall uncertainty. We believe
a reasonable range for B is 1/30 :s :s
B 3. Lattice QCD calculation of the relevant
hadronic matrix elements would allow a more reliable determination of B.
RJr -+ ;YJr: The short-distance subprocess in this reaction is q + g -+ q +;y, again
described by the same low-energy effective operator as in co-annihilation and RO decay.
At the hadronic level the matrix element for RJr -+ ;YX is the same as for R;y -+ Jr X
for any X, evaluated in different physical regions. Thus the difference between the
various cross sections is just due to the difference in fluxes and final state phase space
integrations, and variations of the matrix element with kinematic variables. Given the
crude nature of the analysis here, and the great uncertainty in the overall magnitude of
the cross sections, incorporating the constraints of crossing symmetry are not justified
at present. We will therefore use the same form as for (7), letting C parameterize the
hadronic uncertainty in this case: (iviCTRO,,) '::::' 1.5 x 10- 10 r [f.L~f.LS4C] mb.
This completes the discussion of the lifetimes, cross sections, and their uncertainties.
The results are summarized in Table I.
Once the cross sections and decay rate is known, one can develop the Boltzmann
equations for the system and numerically solve them to find the relic abundance. This is
being studied by Chung, Farrar, and Kolb. But for the purpose of illustrating the main
issues, it will suffice to compare equilibrium reaction rates to the expansion rate near
freeze out.
To obtain an estimate of when the rates will drop below the expansion rate, we will
assume all particles are in LTE (local thermodynamic equilibrium). In LTE a particle of
mass m in the nonrelativistic limit has a number density
(10)
Here g counts the number of spin degrees of freedom, and will be 2 for the RO and the
;Yo Of course all rates are to be compared with the expansion rate. In the radiation-
dominated universe with g. rv 10 degrees of freedom
(11)
There are two striking features apparent when comparing the magnitudes of the
equilibrium reaction rates in Table II. The first feature is that the numerical factor in
RO RO -+ X is enormous in comparison to the other numerical factors. This simply
reflects the fact that RO annihilation proceeds through a strong process, while the other
processes are all suppressed by a factor of MS-4.
295
Table II: The ratio of the equilibrium reaction rates to the expansion rate for the indicated reactions.
Shown in [...J is the scaling of the rates with unknown parameters characterizing the cross sections and
decay width.
Process r EQ / H Scaling
77 -+ X 1.2 X 107 X- 1 / 2 exp( -x) [J.!~J.!S4J
f?!J RO -+ X 3.5 X 1020 x- 1/ 2 r 1/ 2 exp(-rx) [J.!S]
7RO -+ X 8.9 X 107 X1 / 2 r 5 / 2 exp( -rx) [J.!~J.!s4A]
77r+7r- -+ RO 7.1 X 104 x2 r3/ 2 F(r) exp[-(r - l)xJ lJ.t~J.!s4B]
77r -+ R7r 9.6 X 106 X1 / 2 r 5 / 2 exp[-(r - l)xJ exp( -0.175J.!SlX) lJ.t~/2J.!S4C]
-2
-4
8~T.l~~.T~r.~~~~~~~~
._._._._._._ .. r('RO .... X)/H -.-.-.-.-.. r(" .... X)/H

- - - - - - _. r('n .... RD)/H - - - - r('jI7r .... ROn)/H
- - - r/H = 1 ....................... r(RoRD .... X)/H
Figure 2: Equilibrium reaction rates divided by H for r = 1.25 and r = 2.0, assuming
f.,Ls = f.,Ls = 1, and that the factors A = B = C = 1. The rates can be easily scaled for
other choices of the parameters.
The other important feature is the exponential factors of the rates. They will largely
determine when the photino will decouple, so it is worthwhile to examine them in detail.
The exponential factor in ;:y;:y -+ X is simply exp( -miT), which arises from the
equilibrium abundance of the;:Y. It is simple to understand: the probability of one ;:y
to find another ;:y with which to annihilate is proportional to the photino density, which
contains a factor of exp( -miT) = exp( -x) in the nonrelativistic limit.
The similar exponential factor in RO RO -+ X is also easy to understand. An RO must
find another RO to annihilate, and that probability is proportional to exp( - MIT) =
exp(-rx).
The process ;:YRo -+ X is also exothermic, so the only exponential suppression is the
probability of a ;:y locating the RO for co-annihilation, proportional to the equilibrium
number density of RO, which is proportional to exp( - MIT) = exp( -rx)
In ;:Y7r+7r- -+ RO the exponential factor is exp[-(M - m)IT] = exp[-(r -l)x], which
is just the "Q" value of the decay process.
296
Figure 3: Assuming ;Y freeze out is determined by ;Y +-+ RO interconversion, this figures
shows as a function of'r the values of [fJ,~/2 fJ,s4C] required to give the indicated values of
0::yh2. The uncertainty band is generated by allowing fJ,8 to vary independently over the
range 0.5 :::; fJ,8 :::; 2.
For the process ;Yn -+ ROn, it is necessary for the collision to have sufficient center-
of-mass energy to account for the ;Y~Ro mass difference, which accounts for a factor of
exp[-(M - m)/T]. The number density of target pions is exp(-m1f /T), so this factor
is also present. Combining the two factors leads to the overall factor appearing in Table
II: exp[-(M - m + m 1f )/T] = exp[-(r -l)x] exp( -0.175fJ,8 1X ).
Graphs of the reaction rates as a function of r is shown in Figure 2. There are several
things to notice from the graphs: 1) The RO self-annihilation rate is always larger than
the other rates. This means that the assumption that the RO is in equilibrium during ;Y
freeze out is a good approximation for the values of r considered here. 2) Even for r as
large as r = 2, the interconversion rates seem to be (slightly) more important than ;Y self
annihilation in keeping photinos in equilibrium. 3) The process ;Yn +---+ ROn seems to be
the most important process for r < 2. Detailed numerical integration of the Boltzmann
equation by Chung, Farrar, and Kolb confirms this. 4) The freeze-out temperature is
very sensitive to the value of r. This traces to the exponential sensitivity of the reaction
rates upon r. We will make use of this last feature to find a cosmologically acceptable
range of r.
Assuming that ;Yn +---+ ROn does determine the ;Y relic abundance, Figure 3 shows the
sensitivity of 0::yh 2 to r. From this figure we can draw some very interesting conclusions.
If we assume that [fJ,~/2fJ,s4C] < 10 2 , then r must be less than 1.9. If we demand
that the relic photinos are dynamically important (say 0::yh 2 ;::: 10- 2 ) then r ;::: 1.2 if
[fJ,~/2 fJ,s4C] > 0.1. Finally, if we choose our best guess [fJ,~/2 fJ,s4C] '" 1 and 0::yh 2 '" 0.25,
then the allowed range of r is 1.4 :s r :s 1.6.
1.4 Testing the scenarIO

Direct detection of light photinos is not easy. the interaction cross section decreases
with photinos mass, and more importantly, the kick they would give to a massive target
nucleus also decreases with decreasing photino mass.
297
The case for light photinos hinges upon laboratory experiments. The scenario de-
pends upon the existence of the RO, the gluino-gluon bound state,4 with a mass roughly
1.5 times the photino mass. If the RO can be discovered (and after all, the discovery of
a 1.5 GeV hadron does not sound impossible), from its decay one can learn the ::y mass,
and hence r, as well as the parameters of the short-distance matrix element. While there
is no shortage of candidates for relic dark matter particle species, this proposal extends
the idea that photinos may be the dark matter to a previously excluded mass range by
incorporating new reactions that determine the photino relic abundance. If this scenario
is correct, direct and indirect detection of dark matter might be even more difficult than
anticipated. However the scenario requires the existence of low-mass hadrons, which can
be produced and detected at accelerators of moderate energy. Thus particle physics ex-
periments will either disprove this scenario, or make light photinos the leading candidate
for dark matter.
So the job is straightforward: find the RO:
I would like to acknowledge collaboration with Glennys Farrar and Daniel Chung on
the light-photino work reported here. This work was supported by the Department of
Energy and by NASA under Grant NAG5-2788.
References
[1] FARRAR, G. R. & KOLB, E. W. 1996 Phys. Rev. D 53,2990-3001.
[2] CHUNG, D. J. H., FARRAR, G. R. & KOLB, E. W. 1996 In preparation.
4In fact, one suggested title for this talk is "The RO: sglueball or screwball?"
298
SECTION VII - PROGRESS ON NEW AND OLD
IDEAS- B
Non-Universality and Post-GUT Physics in Supergravity Unification
Pran Nath
Department of Physics, Northeastern University
Boston, MA 02115
R. Arnowitt
Center for Theoretical Physics, Department of Physics
Texas A&M University, College Station, TX 77843-4242
Abstract
The non-universal boundary conditions at the GUT scale are discussed and
their effects at low energy analysed. It is shown that the non-universality effects
that couple with the top quark sector are strongly influenced by the Landau pole
effects in the top quark Yukawa coupling. The possibility of analysing post-GUT
physics in the scenario where non-universalities at the GUT scale of the soft
susy breaking parameters arise only from the renormalization group running of
the parameters is pointed out. Some illustrative examples are considered. The
possibility of analysing post-GUT scenarios at the LHC and at the NLC are
discussed.
1. Introduction: Supergravity unification[l] is an effective theory of particle interac-

tions valid at scales MCUT < MPlanck which neglects loop corrections in the gravitational
sector. However, because of the proximity of the GUT scale to the Planck scale, one
can generate corrections to supergravity GUT which are of size
O(McUTIMPlanck) rv O(few%) (1)
If the measurements at low energy are precise, then such measurements can be used
to determine the size of the Planck scale effects. Specifically, there can be observable
effects[2] to as) biT unification [3] etc. Thus Planck scale effects may already be present
in low energy physics. These results may be relevant in view of the current controversy
on the value of as. The analysis of the LEP data indicates a high value for as of
as = 0.123 0.005 (2)

while the low energy data from DIS, Y, JI1/J, etc indicates a low value of
as = 0.112 0.005 (3)
In supergravity unification on has a value of as > 0.12 for values of Tsusy < 1 TeV
over most of the parameter space of the theory. There are various options discussed
in the literature for lowering the value of as. One possibility is that of Planck scale
corrections[2]. Supergravity unification provides a natural framework for the appear-
ance of such corrections, for example, via corrections to gauge kinetic energy function,
which in general is given by
301
_ ~; Fa F(3/1-V
(4)
4 Ja(3 /1-V
Planck scale corections make contributions to fa(3' For example, in SU(5), fa(3 can take
the form
fa(3 = 001(3 + 2~p da (3-yL,-Y (5)
where L, is the 24-plet of SU(5). After spontaneous breaking of SU(5) the Planck scale
corrections generate a shift in as which is roughly given by
as ~ -O.007c (6)
Thus a c'" 1 is sufficient to lower the value of as to bring it in conformity with eq(3).
2. Non-Universalities: Most of the analyses in supergravity have been conducted
in the framework of the minimal supergravity unification[l] with four parameters, and
one has 28 predictions on the susy mass spectra[4]. Some of the mass predictions can
be converted into sum rules[5]. However, these results are based on certain assumptions
such as that the Kahler potential is generation blind, and that the breaking of the GUT
symmetry and supersymmetry occur at the same scale. Thus, it is useful to investigate
departures from these assumptions, and various possibilities present themselves. We
may classify these non-universalities as follows:
1. Non-universality in the Higgs sector.
2. Non-universality at the GUT scale from renormalization group effects assuming uni-
versalityat M sojt > M euT , where M sojt is the scale where the soft susy breaking occurs
and Me is the scale of the gauge coupling constant unification.
3. Non-universality from a general Kahler potential.
Case 1 is the one most frequently investigated in the literature[6] as it is found neces-
sary to include such non-universalities in the large tan,8 scenarios to affect electroweak
symmetry breaking. Regarding case2, it is generally argued that the scale of super-
symmetry breaking is at the Planck scale and thus Msojt = MPlancd7]. Thus if one
assumes a universal breaking of supersymmetry at the Planck scale and evolves the
soft susy breaking parameters from MPlanck to MeuT one generates a splitting of the
soft SUSY breaking parameters at the GUT scale. However, it should be kept in mind
that the effective supergravity theory exits only below the compactification scale or the
string scale. Thus one should think of M sojt as Mstring rather than MPlanck. Further,
since M sojt is now lowered from MPlanck to Mstring it has the effect of reducing the
non-universal effects on soft susy breaking masses in case 2. In case 3 one assumes that
non-universalities arise from a general Kahler potential[8]. Here the soft susy breaking
parameters would in general be arbitrary unless additional assumptions regarding the
generations, such as, the existence of horizontal symmetries is made. In this talk we
focus on cases 1 and 2. We begin with a discussion of case 1. Here one introduces non-
universalities in the Higgs sector while the squark-slepton sector is assumed universal
at the GUT scale. We parametrize the non-universalities in the Higgs sector at the
GUT scale as follows:
(7)
where one often takes the range for the variation of 101,021 :S 1. Since H2 couples with
the top, non-universality in the H2 channel is affected by renormalization group effects.
At the electro-weak scale one finds
2
m~2(t) = ~o (1 + DO)02 + m6h - m6A6k + mOm1/2Aof + mi/2e (8)
302
and
F(t)
Do = (1- 6Yt E(t)) (9)
where the functions e,f,h,k are as defined in ref[9]and Do = 0 gives the location of the
Landau pole in the top yukawa coupling. Further, /1 2 , m~, m" etc are also affected
via radiative breaking of the electro-weak symmetry. To discuss the effect of non-
universalities on /1 2 we display /1 2 at the electro-weak scale in the form
/1 2 = m6C1 + A6C2 + mlAoC3 + ml C4 - ~2 M~

2 2
(10)
We focus here on the effect of the Landau pole on the universalities at low energy.
Typically the dominant terms in IL2 arise from the parts which are proportional to m6
and A6. However, if the non-universalities conspire to significantly reduce the value of
these terms then the effect of non-universalities will be enhanced. For example, consider
C1 :
1
C 1 = (1- tan 2 (3(3Do - 1)/2 + (51 - 2(1 + Do)52 tan 2 (3))/ ( tan 2 (3 - 1) (ll)
In this case non-universalities get enhanced when the values of 0'1 and 0'2 are such that C1
approximately cancels. A somewhat different phenomenon enhances the effect of non-
universalities in the term which contains C2 . Here it is the Landau pole effect in Ao
which controls the enhancement of non-universality. From the one-loop renormalization
group analysis we can write Ao in the form[lO]
AR
Ao = Do + Ao(nonpole), AR ~ At - 0.6mg (12)
Thus for values of parameters where AR vanishes, one will have an enhancement of non-
universality. Non-universalities of this type can directly affect low energy phenomena
such as susy mass spectra, b =} s + "dark matter, etc and precision measurements
can detect the presence of such effects if there are a sufficient number of independent
experiments.
3. Post-GUT Physics and Tests at the Large Hadron Collider(LHC) and at
the Next Linear Collider(NLC): One of the interesting possibilities for the origin of
non-universality is the case when one has universality at Mso/t and non-universalities in
the soft SUSY breaking parameters arise as one evolves the scalar masses from Mso/t to
M euT . Here the evolution and thus the non-universalities depend on physics between
Me and M so/ t ' This means that the nature and size of non-universalities at Me will
depend on the nature of post-GUT physics, i.e., the nature of the group structure, such
as SU(5), SU(3)3, SO(10), E(6), etc. Thus the sparticle spectrum at the electro-weak
scale will become sensitive to the nature of post-GUT physics and one can deduce sum
rules among sparticle masses which are sensitive to the nature of post-GUT physics. If
our measurements at low energy are sensitive enough, then these mass measurements
can be used to distinguish among various post-GUT scenarios. We discuss below few
such scenarios.
(i) SU(3) x SU(2) x U(l): This is a string like scenario where one directly descends
to the gauge group SU(3) x SU(2) x U(l) at the scale M so/ t ' In this case the non-
universalities at Me will be characterised by seven independent parameters
(13)
(ii) SU(5): Here we assume that the gauge group evolution between Mso/t and Me
303
is from 8U(5). In this case the non-universalities at the GUT scale are conveniently
parametrized by
(14)
where 85 determines the non-universalities in i(ih, eL), dR and 810 determines the non-
universalities in ih(ih, dL ), UR, eR which implies the equality of the iL and dR masses
and the equality of the ih,UR and eR masses at the GUT scale. Thus in this type of
scenario non-universal effects cancel out in the following mass differences:
(15)
(iii) SO(10):In this case the evolution between Msoft and MG is determined by the
group 80(10) and we assume that 80(10) breaks to the standard model gauge group
SU(3) x SU(2) x U(1) at the scale MG. We note that here one has a reduction in
the rank of the group which leads to D term contributions to the soft susy breaking
sector[ll].
From a study of the low energy mass spectrum one can distinguish among the three
cases discussed above. We present below such a strategy, as we consider specific sum
rules which can be used to distinguish among the three scenarios (i), (ii) and (iii)
discussed above. Let us consider, for example, the quantity
6.-ULUR
- = m-2 2
- m-UR - 3 - -11
- QG(-f2 1 )ml - (1- - -4 sin2 Bw ) M z2 cos 2{3 (16)
UL 2 2 "2 2 3
where the functions fI,h,h are as defined in [9]. The condition
(17)
would point to scenario (i) among the three possibilities given above. If, however,
suppose 6. uL UR = 0 and if one also has
(18)
(19)
(20)
that would imply that either scenario (ii) or scenario (iii) is valid. To distinguish further
between scenarios (ii) and (iii) we need to look further at the details of the splittings
of the spectrum. For the 80(10) case if we assume that the higgs doublets ( which
give masses to the up quarks and the down quarks and leptons) belong to the same 10
representation of 80(10) then, one can derive the following sum rule
6. 80 (10) = 0 (21)
where
(22)
The above equation can be converted into an equation involving sfermion masses at
the electro-weak scale using renormalization group evolution. Using the notation x =
(3 - Do)/(1 + Do), y = 2/(1 + Do), z = (3 + Do)/(2 + 2Do) and w = (1- Do)/(1 + Do),
we have
(23)
where R = RA + RB and RA is independent of the specific nature of the particle species
304
being considered and RB depends on the nature of the particles. Now RA consists of
several pieces:RA = RAl + RA2 + RA3 where RAl = zM'i cos 2{3, RA2 = WJ-t2, and
R A3 = (ye - g)mi + ym~( -kA~ + Aofml/mo) (24)

2 2
Similarly RB consists of several parts:RB = RBl + RB2 + R B3 + RB4 where RBl =

-ae(~13 + i5ft)mI/2' RB2 = ~sinB~M'icos2{3, R B3 = x(-~ + ~sinB~) M'icos2{3,
and
_ (8 3 1 2
RB4 = (leX "313 + 212 + 30ft)m~ (25)
mo, ml, Ao and tan{3 are determined from other experiments as discussed below. Va-
2
lidity of the above sum rule eq(21) would attest to the existence of an SO(lO) symmetry
in the post-GUT region and thus distinguish between the scenarios (ii) and (iii). In
addition to the above one can write several other sum rules that can distinguish among
the different post-GUT scenarios.
Next we discuss briefly how well these sum rules can be checked at future colliders.
Our analysis is based on the work of ref [12] which discussed the reach of several future
colliders, including the LHC and the NLC. It is found that the reach of LHC is quite
high for squarks and gluinos. Thus one can detect squarks up to ::::i 2 TeV for mij ::::i my
and up to mij ::::i 1.0 -1.5 TeV for mij my. However, the actual mass measurements
are much harder. Thus the errors for the measurements of the gluino mass are as
my ::::i mij : Errar ::::i 15%; my < < mij : Error ::::i 25% (26)
for my :S 800 GeV. The determination of the squark mass is even more difficult. One can
get a quantitative estimate from jet multiplicities[12]. Thus no. of jets ::::i 3 - 4 =* mij ::::i
my and no. of jets::::i 5 - 6 =* mij my. For charginos and neutralinos the only signal
that seems viable is the xfx~ -4 3l signal[13] which allows a precise mass measurement
of the mass difference mxg - mx~ (from the upper cutoff of, e.g., the J-tee event rate
distribution) up to a precision of perhaps 5 GeV. However, the mass reach is limited to
mx.o :S 150 GeV. After this mass range spoiler modes such as xg -4 X~ho,X~Z begin
to reduce the signal. For sleptons the mass reach is rather limited. Thus one finds[12]
min'" 200 GeV, for ml/2 > mo: (miL> min)' Similarly, min'" 250 - 275 GeV, for
ml/2 < mo: (miL'" min)' However, it appears difficult to obtain any quantitative
mass measurements.
Much more accurate mass measurements are possible at the NLC than at the LHC.
However, the mass reach of NLC is limited because of its center of mass energy of
VB = 500 GeV. In the following we give estimates of the reach of the susy particle
masses at the NLC. For the Higgs, one should be able to explore the full mass range
mh :S 150 GeV, and for mA one should be able to explore the mass range up to 200
GeV. For sleptons, from the decay i -4 l + xf a conservative analysis shows a mass
reach of mi :S 225 GeV Analysis of polarized beams shows that one could determine
miL, mx~ to within 1% accuracy. Study of T -4 T+X~ can give gauginoj higgsino content
of X~. Thus for example the chirality of the final T relative to the chirality of T, i.e.,
same( different) will determine if X~ is dominantly gaugino (higgsino). The mass reach
for xf(at 50") is mx~ :S 248 GeV from xtxl pair production. Here one can determine
mx~' mx~ to within 1% (27)
( with the exception when xf is gaugino-like and VI is light. Here, the mass measure-
ments are good to '" 5%). With the above mass measurements in place,one can then
determine m;;e to within 3%. Regarding the squarks, the first and the second generation
305
mass measurements can be made to within a few GeV. For the third generation, assum-
ing that 11, and tan,6 are determined from chargino/neutralino sector one can determine
At from the i mass matrix. A rough estimate of errors is 8~: :::
3. A similar analysis
for the b-system gives 8~_b ::::: 150, tan,6 = 30, so that Ab is"only roughly determined.
b,
Regarding the first and second generation squarks, the energy reach of NLC may not
be sufficient to make them experimentally accessible. Thus, for example, consider the
up squark ih. One has
m~L ::::: m6 + 0.85m~ + 0.34M~( cos 2,6) (28)
A non-observation ofthe light chargino at LEP1.8 would imply that mw > 90 GeV and
by scaling a mg > 270 GeV. Then even with ma = 0 one has m~L ;::: 250GeV, my ;::: 270
GeV, which would put it out of the reach of NLC. In this circmstance only the t1 ( or
bL ) could be low enough to be seen at the NLC. The solution to this problem is to build
an NLC with Vs = 1 TeV. Returning to the test of the universality vs non-universality
we can determine the soft susy breaking parameter ma given the determination of the
squark masses and other parameters. From eq(28) we also have
m6 : : : mt - 0.85m~ - 0.34M~( cos 2,6) (29)
Assuming some typically realistic errors in the determination of quantities on the right
hand side, i.e., '" 1%error in the determination of m UL ' '" 3% error in the determination
of m1/2 and correspondingly a '" 6% error in the (indirect) determination of my and a
'" 3% error in the determination of tan,6 one has
8ma = 4 - 5% (30)
ma
which implies that the test of universality and of the post-GUT sum rules could be
carried out to a level of accuracy of about 5% or less at the NLC.
4. Conclusions: In the above we have shown that the sparticle mass spectrum at low
energy contains signatures of non-universalities. Further, if one makes the assumption
that universality occurs at a scale MsoJt which is the string scale rather than at the
GUT scale, then the spectrum at low energy will become sensitive to the nature of post-
GUT physics. Thus accurate measurements at low energy will be able to distinguish
among various post-GUT scenarios. LHC will be able to test some of the sugra and
post-GUT sum rules. However, while LHC is an excellent machine for the discovery of
supersymmetry, the mass measurements at LHC have a limited degree of sensitivity.
Much more accurate tests of sugra GUT and post-GUT sum rules are possible at the
NLC to an accuracy of up to 5% or less. However, the reach of NLC is limited and is
not likely to cover the full range of susy particles. Thus an ideal machine for the study
of sugra GUT sum rules, and of post GUT sum rules is a 1 TeVon 1 TeV e+e- or a
11,+ p,-machine. Such a machine will be able to cover a bulk of the spectrum of sugra
GUT.
Acknowledgements: This research was supported in part by NSF grant numbers
PHY-19306906 and PHY-9411543.
References
1. A. H. Chamseddine, R. Arnowitt and P. Nath, Phys. Rev. Lett 29. 970 (1982);
P. Nath, Arnowitt and A. H. Chamseddine, "Applied N=l Supergravity" (World
Scientific, Singapore, 1984); H. P. Nilles, Phys. Rep. 110, 1 (1984); R. Arnowitt
and P. Nath, Proc of VII J. A. Swieca Summer School (World Scientific, Singapore
1994).
306
2. C. T. Hill, Phys. Lett. B135, 47 (1984); Q. Shafi, and C. Wetterich, Phys. Rev.
Lett. 52, 875 (1984); D. Ring, S. Urano, and R. Arnowitt, Phys. Rev. D52,
6623 (1995); T. Dasgupta, P. Mamales, and P. Nath, Phys. Rev. D52, 5366
(1995).
3. V. Barger, M. S. Berger, and P. Ohman, Phys. Lett. B314, 351 (1993).
4. R. Arnowitt and P. Nath, Phys. Rev. Lett. 69, 725 (1992); P. Nath and R.
Arnowitt, Phys. Lett. B289, 368 (1992).
5. S. P. Martin and P. Ramond, Phys. Rev. D48, 5365 (1993).
6. D. Matalliotakis and H. P. Nilles, Nucl. Phys. B435, 115 (1995); M. Olechowski

and S. Pokorski, Phys. Lett. B344, 201 (1995); V. Berezinsky et ai, CERN-TH
95-206.
7. N. Polonsky and A. Pomerol, Phys. Rev. D51, 6532 (1995).
8. S. K. Soni and H. A. Weldon, Phys. Lett. B126, 215 (1983); V. S. Kaplunovsky

and J. Louis, Phys. Lett. B306, 268 (1993).
9. L. Ibanez, C. Lopez, and C. Munos, Nucl. Phys. B256, 218 (1985).
10. P. Nath, J. Wu and R. Arnowitt, Phys. Rev. D52, 4169 (1995).
11. Y. Kawamura, H. Murayama and M. Yamaguchi, Phys. Lett. B324, 54 (1994);

Phys. Rev. D51, 1337 (1995).
12. H. Baer et ai, FSU-HEP-95041, LBL-37016, UH-511-822-95; C-H. Chen et al.,

FSU-HEP-950720.
13. P. Nath and R. Arnowitt, Mod. Phys. Lett. A2, 331 (1987); R. Arnowitt, M.
Barnett, P. Nath, and F. Paige, Int. Jon. Mod. Phys. A2, 1113 (1987); R.
Barbieri, et. al.
Nucl. Phys. B367, 28 (1993); H. Baer and X. Tata, Phys. Rev. D48, 2062
(1993);
J. Lopez et. al., Phys. Rev. D48, 2062 (1993).
307
PADE APPROXIMANTS, BOREL TRANSFORM, AND RENORMALONS:
THE BJORKEN SUM RULE AS A CASE STUDY
John Ellisl, Einan Gardi 2, Marek Karliner 2, and Mark A. Samuel 3
1 Theory Division, CERN, CH-1211, Geneva 23, Switzerland

E-mail: johne@cernvm.cem.ch
2 School of Physics and Astronomy
Raymond and Beverly Sackler Faculty of Exact Sciences
Tel-Aviv University, 69978 Tel-Aviv, Israel
E-mail: gardo@post.tau.ac.il.market@vm.tau.ac.il
3 Department of Physics, Oklahoma State University
Stillwater, Oklahoma 74078,USA
E-mail: physmas@mvs.uss.okstate.edu
ABSTRACT
We prove that Pade approximants yield increasingly accurate predictions of higher-

order coefficients in QCD perturbation series whose high-order behavior is governed by a
renormalon. We also prove that this convergence is accelerated if the perturbative series is
Borel transformed. We apply Pade approximants and Borel transforms to the known pertur-
bative coefficients for the Bjorken sum rule. The pade approximants reduce considerably the
renormalization-scale dependence of the perturbative correction to the Bjorken sum rule. We
argue that the known perturbative series is already dominated by an infra-red renormalon,
whose residue we extract and compare with QCD sum-rule estimates by an higher-twist eff-
ects. We use the experimental data on the Bjorken sum rule to extract a s (M /) = O.l16~:~~:
including theoretical errors due to the finite order of available perturbative QCD calcula-
tions, renormalization-scale dependence and higher-twist effects.
INTRODUCTION
Everybody interested in more precise quantitative tests of QCD, or in its place in some
Grand Unified Theory, would welcome a more precise determination of the strong coupling
strength a s in some well-defined renormalization prescription, say MS, at some reference
energy scale, say M z. Such determination are normally made using perturbative QCD to
interpret data, though lattice QCD may also become competitive once systematic effects are
309
better controlled. Obtaining the desired level of precision using perturbative QCD requires
calculations beyond the next-to-Ieading order. Several processes are calculated to high order
in perturbative QCD \,2 . However, progress in the high-precision determination of as (Mz)
is hampered by the fact that the QCD perturbation series is expected to be asymptotic:
Us
= Lcllx ll ,
C1J
Sex) X= -; CII ~ n!Klln Y (1)

1t
for some coefficients K, y 3 . Under these circumstances, how can one extract the most
information from the QCD perturbation series, and obtain the best value of a s = 1tX? The
usual answer is to calculate up to order nopt : 1111 =icllxlli is minimized, and use the magni-
tude 1111 of this minimum term as an estimator for the residual uncertainty.
"P'
In this paper we study whether it is possible to estimate Sopt = ~ c"x" reliably without
11=0
the labour of calculating all the perturbative coefficients ell : n :-: ; nopt , and comment on the
possibility of summing over the higher-order terms S,,,yt = LC"x" sufficiently reliably to
11=0
reduce the magnitude of the residual uncertainty below 1111 Our approach to these issues is
"P'
based on Pade approximants, which are widely appreciated in other areas of physics, and
which we have recently shown 4 can be used to predict higher-order perturbative QCD coeff-
icients in agreement with exact calculations(where available) and with the effective charge
method 5,6.
In this paper, we present new results on the rate of convergence of Pade approximants
for series of the form (1) expected in QCD. We also demonstrate that they reduce signifi-
cantly the renormalization-scale dependence of the perturbative series for the Bjorken sum
rule, and summarize a comparison with another technique for treating higher-order effects in
perturbative QCD 7. To go further, we transform to the Borel plane, where beha-viors of the
type (1) correspond to discrete renormalon singularities 3. The Pade technique is a priori well
adapted to locating such singularities, and we indeed prove that the conver-gence of the Pade
approximants is accelerated for the Borel transform of a series such as (1). We apply this
combines Pade-Borel technique to the calculated QCD perturbation series for the Bjorken
sum rule, and show that it yields a leading infra-red renormalon pole close to the expected
location in the Borel plane. Assuming this location, we extract its pole residue and use it to
evaluate the possible magnitude of the infra-red renormalon ambiguity in the pertur-bative
contribution to the Bjorken sum rule, which we argue is canceled by a corresponding ambi-
guity in the non-perturbative contribution. We use Pade summation, extracting a,(M/)
= O.l16+~~~: from the available polarized structure function measurements, including theoreti-
cal errors associated with renormalization-scale dependence and higher-twist effects. The
accuracy of this result testifies to the utility of both Pade approximants and the polarized
structure function data.
2 "CONVERGENCE" OF PADE APPROXIMANTS
We denote Pade approximants (PA's) to a general perturbative QCD series

x
Sex) = LC"x" by
1/",,0
310
(2)
i.e. the PA's are constructed so that their Taylor expansion up to and including order N+M
is identical to the original series. We have previously pointed out that the next term in the
Taylor expansion of a [NIM] PA typically provides increasingly accurate estimate C;:~M+I of
the next higher-order perturbative coefficient CN+M+l of the original series. In the following
we refer to such estimate as Pade Approcimants PREDICTIONS (PAP's).
Let us briefly restate the condition8 for the convergence of PAP's. With f (n) == In CII
and g(n) == d 2 f(n) / dn 2 (where the derivative with respect to n is to be understood in a
discrete sense), a sufficient condition for PAP convergence, C':~M+I ~ C N + M + 1 , is lim g(n) = o.
,,-+00
To quantify therate of convergence, we introduce the quantity
(3)
It is easy to check that Ell: 11 n for an asymptotic series of the form (1). When the
asymptotic behaviour of E II is known, it is possible to write down an asymptotic formula for
the relative error 8 [NIM] in the PAP estimate C;:~M+I of the perturbative coefficient CN+M+l.
est
s: CN+M+I - CN+M+I
U[NIM] == (4)
C N + M +1
When Ell: 1/ n , as it is the case for an asymptotic series of the form (1), we have been able
to demonstrate, for all values of N and several values of M, that
M!
8[NIM] : - LM ' where L = N + M +a'M +h. (5)
for some choice of numbers a' , h. This implies that asymptotically
Inl8 [MIM]I: - M[1 + In(2 + a')] (6)
We have verified that this formula is numerically accurate for simple series of the form (1)
in which CII = n!Klln1 Moreover, we have checked the prediction (6) for several different
asymptotic series, including that for the QeD vacuum polarization D function in the large-Nj
approximation 9 , where it agrees numerically very well with the relative error reported in
panel (a) of the figure in Ref. [4]. We note that the large-Nj D function contains an infinite
number of renormalon poles, which could in general provide important corrections to the
leading-order formulae (5,6). The fact this is not the case supports the empirical utility of the
PAP's even beyond the idealized analytic case (1).
In the previous paragraph, we have discussed the use of Pade approximants to estimate
the next term in a given perturbative series, and, as we have seen, sufficient conditions for
the convergence of such PAP's are known, so that the main open issue is their actual rate of
convergence in practical applications. Next we discuss how to use PA's to estimate the
"sum" of the perturbative series, which we term Pade Summation (PS).
311
It is important to note that convergence of the PAP's is largely independent of the
"summability" of a given series. Thus, for example, the PAP method gives equally precise
x x
predictions for the next terms in the series Ln!x" and Ln!(-x)" , even though the latter is
o 0
Borel summable, while the former is not.

The formal issues related to Pade Summation are less clear, especially because most
perturbation series of practical interest are not Borel summable (since their Borel transforms
have poles on the positive real axis). One well-defined prescription for defining the "sum" of
such a series is the Cauchy principal value of the inverse Borel transform integral 10 , so we
ask whether PS "converges" to this prescription for the "sum" of the series. Such series are
in general obtainable from functions with cuts on the positive real axis, with a toy example
'"
being provided by the simple asymptotic series In! x" , which is a formal expansion of
o
e~1 1 '" e~Yx
f - dt =-x f-dY
'"
(7)
01- xt 1- Y 0
We see from the representation on the right-hand side of (7) that this series corresponds in
QCD language to a single simple infrared renormalon pole, whilst the left-hand side of (7)
exhibits a cut on the positive real axis.
3 APPLICATION TO THE BJORKEN SUM RULE
We now discuss a concrete application ofPA's to a perturbative QCD series, namely

that for the Bjorken sum rule, which takes the following form in the MS renormalization
. . 13 I
prescnptIOn . :
frg( (X,Q2) - g;' (X,Q2)] = !lgA If(x) : f(x) = 1- x - 3.58x 2 - 20.22x 3+...+(HT) (8)
o 6
for Nj = 3, as relevant to the f!
range of current experiments, where x = a ,(Q2 )/n , the dots
represent uncalculated higher orders of perturbation theory, and (HT) denotes higher-twist
terms. The perturbative series in (8) is expected to be dominated by renormalons in large
ordersl4, leading to growth in the perturbative coefficients c,~j of the form shown in (1). The
PA's to the series (8) yield the following predictions for the next term
8j
C4[PA] '" -111 ([112] PA)
8j
(9)
c 4 [PA]",-114 ([2/l]PA)
in this series.
In order for these PAP's to be useful, it is important to estimate the errors involved.
The asymptotic error estimate given in Eq. (5) requires as input values of a' and b, which are
not known a priori. Experience with many series shows that typically -1 ~ a' ~ 0 and b '" 0 .
With a' = 0 and b = 0 we obtain a ballpark estimate
(10)
In a previous paper4, we used a different method to estimate the errors of the Pade prediction
(9), obtaining C:(PA] = -112 33' as the error-weighted average of the [112] and [211]
312
approximants. We also pointed out that this prediction is close to an estimate made using the
Effective Charge Method (ECH):
BJ
C 4 [ECH) ~ -l30 (11)
We now note further that even though the error estimate (10) are in principle expected to
hold only asymptotically, in practice (9) and (11) are consistent with each other within these
c:
estimates. We take this as an indication that the true value of J is likely to be in the range
predicted by the PAP and ECH methods. Moreover, as we shall show in section 4, it seems
that the perturbation series in (8) is already dominated by a single infrared renormalon, in
which case the quantities (10) are no longer "statistical" errors, but fractional corrections to
be subtracted from (9), improving the concordance with the ECH estimate (11).
The next step is to apply the PS procedure to estimate the complete correction function
f(x) in (8). The [2/2] PS estimate of f(x) (obtained using the ECH value (11) of the fourth-
order perturbative coefficient),
1[2/2) (x) = 1- 8.805x + 1l.974x 2 (12)

1- 7.805x + 7.753x 2
with the [1/2). [2/1] PS's and with the partial sums of the perturbative series up to order x 3
and x4 (the latter also taken from (11)). We see that the different PS's are numerically quite
stable in the range x::::; 0.1 of relevance to present experiments, which is related to the fact
that the nearest poles are some distance away (x = 0.18 for the [2/1] PA, x = -3.41 and x =
0.18 for the [1/2] PA, and x = 0.15 and x = 0.86 for the [2/2] PA). This means that the
"combined methods" for smoothing ofPA's described at the end of the previous section is
not necessary.
We have also compared the PS' s with results based on the BLM treatment of the
perturbative series in the MS scheme, in which the growing higher-order coefficients are
absorbed into the scales r;! at which x = a ,(Q2) I 1! is evaluated in each of the lower-order
terms. As described elsewhere 12 , we find that our PS procedure agrees very well with the
BLM procedure when applied to the perturbative series for the Bjorken sum rule, adding
further support to our contention that the PA's may indeed accelerate usefully the
convergence of perturbative QCD series, as suggested by the general arguments of Section
2.
4 PADE APPROXIMANTS IN THE BOREL PLANE
If one knows the asymptotic behaviour of the series under study, one can go further. In
particular, if the perturbative coefficients diverge as in (1), which is believed to be the case
in perturbative QCD, corresponding to a discrete set of renormalons 3, it is useful to consider
the Borel transform of the series Sex) in Eq. (1):
IHI
ro
S(y) == Ic;,y" : C = cn +l ( ~) (13)

n=O
n n.'A1-'0 '
If the Borel transform S(y) indeed has a discrete set of renormalon singularities
r k / (y - Y k) P , where the rk's are the residues, PA's in the Borel plane are a priori well suited
to find them. Indeed, if there is a finite set of renormalon singularities, as occurs for the
313
Bjorken sum rule series in the large-Nj approximation, higher-order PA's will be exact. In
general, the removal of the n! factors in the coefficients (13) means that the corresponding
quantity measuring the rate of convergence of the Pade prediction for the next term is
(14)
which is much smaller than the previous E" "" l/n (3). This means that the relative error
O[N,M] in the PA of the Borel-transformed series is also much smaller than (5) for the
original series,
(15)
corresponding asymptotically to
InI8[M'M] 1= -2M[1+ln(2+a')] (16)
We have also checked this prediction for several different asymptotic series, including that
for the QCD vacuum polarization D function in the large-Nj approximation9, which has a
discrete infinity ofrenormalon poles. The prediction (16) again agrees numerically very well
with the relative error reported in panel (b) of the figure in Ref. [4], which is much smaller
than that for the naive PA in panel (a). Again, the success of the prediction (15) gains
significance from the fact that the large-Nj calculation exhibits an infinity of renormalon
poles, indicating that the Borel PA's are useful in the real world, and not only in idealized
simplified situations.
We now apply this combined BorellPade technique to the QCD perturbation series (8)
for the Bjorken sum rule. With the normalizaton of the Borel variable y implicitly defined
through Eq. (13), the [211] PA to the Borel transformation of (8) has a pole at y = 1.05 with
residual r = 0.98 . The appearance of a pole near y = 1 is encouragingly consistent with the
exact large-Nj calculations 14 , which yield poles at y = l, 2. In the MS prescription that
we are using, the residuals of these poles contain factors exp[5y/3].
Therefore, it is not surprising that an infrared renormalon pole at y = 1 emerges more clearly
than an ultraviolet renormalon pole at y = -1. Prima jacie, the message of this analysis is
that the calculated Bjorken series is already dominated by the expected leading infra-red
renormalon.
Encouraged by this success, we have made fits to the Borel transform of the Bjork-en
series with varying numbers of poles whose locations are fixed in accordance with
theoretical expectations. We see the following points:
(i) the residual of the y = 1 pole is consistently found to be positive and around
unity,
(ii) the residual of the y = -1 pole is much smaller, and consistent with zero,
(iii) there is room for a second pole at y = 2, but it is not possible to disentangle
this from a higher-lying pole.
We now discuss the possible implications of this BorellPade exercise for pheno-
menology. First, we note that the dominance by a single infrared renormalon pole indi-cates
that, as already remarked, the fractional errors (10) should be subtracted from the naive
estimates (9), bringing them into better agreement with the ECH estimate (11):
314
(17)
from the [112] and [211] PA's respectively. Secondly, it is well known that the magnitude of
the residual r" of the y = 1 pole corresponds to a possible renormalon ambiguity 7tr,
relative to the Cauchy principal value discussed previously. This is also shown for our toy
series in Fig.l. Taking r,for the Bjorken series from Fig 4, we find an ambiguity
d(rt - rn = ~0.9S1t A: (IS)

6 Q
in the perturbative contribution to the Bjorken sum rule. Numerically, with A = 25050
MeV, this corresponds to 0.0400.016 GeV2/(j, which is to be compared with previous
QCD sum rule higher-twist estimates that yield'6
d (r p _ r") = _ 0.02 0.01 (19)

HT' I Q2
(see also the discussion pertaining to Eq.(7) in Ref.[17], and a recent estimate in Ref. [1S] ).
We note that the order of magnitude of the renormalon ambiguity we find (IS) is close to the
higher-twist calculation (19).
Since the full QeD prediction for any physical quantity must be unique, the renor-
malon ambiguity must be canceled by a corresponding ambiguity in the definition of the
higher-twist term. For this reason we do not interpret the renormalon ambiguity as leading
directly to an ambiguity in as, but rather use the uncertainty in the higher twist term.
5 EXTRACTION OF as FROM BJORKEN SUM RULE DATA
We conclude this paper by extracting'7 a,(M/) from data on the Bjorken sum rule,
including a discussion of theoretical errors. We combine the available experimental
evaluations of rt" (Q2) to obtain
rt (3GeV2) - ri" (3GeV2) = 0.164 0.011 (20)
where we have evolved all the experimental results'9,20,2' to the common reference scale
(j = 3 GeV2. Evolution of the proton data from 10 to 3 GeV2 is based on the analysis in
Ref. [20]. The corresponding evolution of the deuteron data is based on the proton data,
combined with the expected (j dependence of the Bjorken sum rule, with as (Q2) needed
as in,put determined via iteration, in a self-consistent way. Evolution to 3 GeV2 from values
of (! different from lOGeV2 was treated in a linearly approximation to the full dependence
of the data on 1Ilog(j), which is adequate for this purpose.
We use the [2/2] PS (12) to obtain: a ,,(3 GeV2) = 0.32S~~:~;~. We evolve this up to
M; by numerical integration ofthe three-loop ~ function2, locating the b-quark threshold in
the MS scheme at mb = 4.3 0.2 GeV 22, and using the three-loop matching condition of
ref.[23], to fiind
(21)
315
where the ... in (21) recalls that theoretical errors remain to be assigned.
There is a theoretical error associated with the spread in the different evaluation
procedures shown in Fig.2, which we estimate from the difference between the [2/2] and
[112], [211] PS' s to be L'i proP s (3 Ge y2) = 0.014 , corresponding to
L'i proca s(M;) = 0.002 (22)
Another way oflooking at the theoretical error uses the ~-dependence of the [212] PS shown
in Fig.3 to estimate L'i ~a s (3 Ge y2) = 0.009 , corresponding to
(23)
from varying ~ between QI2 and 2Q. Both (22) and (23) are estimates of the uncertainty in
a s (M;) due to our uncertainty in the functional form of the QCD correction factor fix), so
that it could be regarded as double counting to include them both. Nevertheless, to be
conservative we will add them in quadrature'.
We also include a shift and error in the determination of aJQ2) infrared from the
estimated range (19) of the higher-twist correction: L'i HTa s (3 Ge y2) = -0.024 0.014,
corresponding to
L'iHTa s(Mz 2) = -0.003 0.002 (24)
Combining (21), (22), (23) ,(24), we extract
(25)
where the first errors are the experimental errors in (21), and the second errors are the sums
in quadrature of the errors in (22), (23), (24).
Our final value of as (Mi) (25) is compatible with the central value extracted from
compilations of previous measurements 24 , and has an error which is competitive. As well as
the experimental error, we have included motivated estimates of a number of theoreti-cal
errors, using information obtained from our study of Pad6 approximants. We believe that
this exercise demonstrates the utility of Pad6 Approximants in QCD, and the value of
polarized structure function data for determining a s (M; ) .
ACKNOWLEDGMENTS
We thank David Atwood, Bill Bardeen, Stan Brodsky, Georges Grunberg, Sasha
Migdal and Al Mueller for useful discussions. This research was supported in part by the
Israel Science Foundation administered by the Israel Academy of Science and Humanities,
and by a Grant from the G.LF., the German-Israeli Foundation for Science Research and
Development. It was also in part supported by the US Department of Ener-gy under Grant
No. DF-FG02-94-ER40852 .
We have also considered possible systematic theoretical errors due to uncertainty in the evolution of as up to
Mz , including unkwown higher-order terms in the QeD p-function, the uncertainty in mb, and freedom in
treating the heavy threshold 23 . They contribute an error in a sCMz,2) which is much less than 0.00 I.
316
REFERENCES
1. S.A. Larin, F.V. Tkachev and lAM. Vermaseren, Phys. Rev. Lett. 66(1991)862; S.A.
Larin and J.A.M. vermaseren, Phys. Lett. B259(1991)345.
2. S.G. Gorishny, AL. Kataev and S.A. Larin, Phys. Lett. B259(1991)144;L.R.
Surguladze and M.A Samuel, Phys. Rev. lett. 66(1991)560; S.G. Gorishny,
AL.Kataev and S.ALarin, Standard Modei and Beyond: from LEP to UNK and
LHC, ed. by S. Dubnicka et aI. (World Scientific, Singapore, 1991) p.299; O.V.
Tarasov, AA. Vladimirov and AYu. Zharkov, Phys. Lett. B93 (1980) 429.
3. A Mueller Nucl. Phys. B250(1985)327; AI.Vainshtein and V.l. Zakharov, Phys.Rev.
Lett. 73(1994)1207.
4. M.A. Samuel, l Ellis and M. Karliner, Phys. Rev. Lett. 74(1995)4380.
5. G. Grunberg, Phys. Lett. B95(1980)70, E - ibid. BIlO(1982)501; Phys. Rev. D29
(1984) 2315; P.M. Stevenson, Phys. Rev. D23(1981)2916.
6. A.L. Kataev and V.V. Starshenko, Mod. Phys. Lett. AlO(1995)235.
7. SJ. Brodsky, G.P. Lepage and P.M. Mackenzie, Phys. Rev. D51(1995)3652; see also
SJ. Brodsky and HJ. Lu, Precision Tests of Quantum Chromodynamics and the
Standard modei, SLAC-PUB-95-6937, hep-ph/9506322, and references therein.
8. M.A. Samuel, G. Li and E. Steinfelds, Phys. Rev. E51(1995)3911; M.A. Samuel and
S.D. Druger, IntI. J Th. Phys. 34(1995)903.
9. C.N. Lovett-Turner and C.l Maxwell, Nucl. Phys. B432(1994)147.
10. G. Grunberg, Phys. Lett. B325(1994)441; see also V.A. Fateev, V.A Kazakov and P.
B. Wiegmann, Nucl. Phys. B424(1994)505 for a 2D example where this pres-
cription is exact.
11. C.M. Bender and S.A Orszag, Advanced Mathematicai Method for Scientists and
Engineers, McGraw-Hill, 1978, sec 8.6 and Prob. 8.59.
12. J. Ellis, E.Gardi, M. Karliner and M. Samuel, to be published.
13. lBjorken, Phys. Rev. 148(1966)1467; Phys.Rev. D1(1970)1367.
14. C.N. Lovett-Turner and C.J. Maxwell, All orders Renormaion Resummations for
Some QCD Observabies, Durham preprint DTP-95-36, hep-ph/9505224.
15. P.A. Raczka, Z. Phys. C65(1995)481, hep-ph/9506462.
16. I.I. Balitsky, V.M. Braun and AV. Kolesnichenko, Phys. Lett. B242(1990)245;
erratum: ibid, B318(1993)648. B. Ehrnsperger, A Schaefer and L.Mankiewicz,
Phys. Lett. B323(1994)439; G.G Ross and R.G. Roberts, Phys. Lett. B322
(1994) 425; E. Stein et ai., Phys. Lett. B343(1995)369; E. Stein et ai., Phys.
Lett. B353(1995)107.
17. J. Ellis and M. Karliner, Phys. Lett. 341(1995)397.
18. V.M. Braun, QCD Renormaions and Higher Twist Effects, Proc. Moriond 1995, hep-
ph/9505317.
19. SLAC-Yale E80 ColI., MJ. Alguard et aI., Phys. Rev. Lett. 37(1976)1261;41(1978)70;
SLAC-Yale ColI., G. Baum et ai., Phys. Rev. Lett. 45(1980)2000; SLAC-Yale E130
ColI., G. Baum et ai., Phys. Rev. Lett. 51 (1983) 1135; EMC ColI., J.Ashman et ai.,
Phys. Lett. B206(1988)364, Nucl. Phys. 32(1989)1; SMC ColI., B. Adeva et ai., Phys.
Lett. B302(1993)533; E142 ColI., P.L. Anthony et ai., Phys. Rev. Lett. 71(1993)959;
SMC ColI., D. Adams et ai., Phys. Lett. B329(1994)399; E 143 ColI., K. Abe et ai.,
Phys. Rev. Lett. 74(1995)346, Phys. Rev. Lett. 75(1995)25; SMC ColI., D. Adams et
ai., A New Measurement of the Spin Department Structure Function gj(x) of the
Deuteron, CERN-PPE-95-097 (June 1995).
20. P. Grenier, Ph.D. dissertation (1995) and private communication.
21. Y. Roblin, Ph.D. dissertation (1995).
22. Particle Data Group, Review of Particle Properties, Phys. Rev. D50(1994) 1173.
317
23. W. Bernreuther, Z. Phys. C20(1983)331.
24. S. Bethke, Aachen preprint PITHA-95-14 (1995), to appear in Froc. 30-th Rencontre
deMoriond, QeD and High-Energy Hadronic Interations.
318
HADRON SUPERSYMMETRY AND RELATIONS BETWEEN
MESON AND BARYON MASSES
D. B. Lichtenberg
Physics Department
Indiana University
Bloomington, IN 47405
INTRODUCTION
Supersymmetry is a symmetry between bosons a~d fermions. To my knowledge,

Hironari Miyazawa was the first person to apply the idea of supersymmetry to parti-
cle physics. As early as 1966 he published a paperl in which the abstract states that
"both baryons and mesons are grouped together in ... a supermultiplet." Miyazawa in-
troduced his scheme with the aim of obtaining relations between properties of baryons
and mesons. He wrote a second paper on the subject two years later.2
Miyazawa's original work was five years earlier than that of Gol'fond and Likht-
man 3 and eight years earlier than that of Wess and Zumin04 on supersymmetry. Of
course, Miyazawa's hadron supersymmetry is quite different from the supersymmetry
of the 1970's, which requires that there exist elementary supersymmetric partners to
the elementary fermions and bosons of the standard model.
Catto and Giirsey 5,6 brought the work of Miyazawa into the framework of QCD.
They showed that meson-baryon supersymmetry arises from an underlying super-
symmetry of an anti quark and a diquark (a correlated two-quark system in a baryon)
because both are antitriplet states of color. Somewhat earlier, Gao and H0 7 ,8 inde-
pendently developed similar ideas, but their work is not widely known.
In Miyazawa's scheme, a pion and a nucleon belong to the same supermulti-
plet, and therefore, in the absence of symmetry breaking, should have the same mass.
Because the pion and nucleon have very different masses, it is clear that hadron super-
symmetry is badly broken. The breaking arises because a diquark and an antiquark
have different masses, sizes, and spins. However, despite the symmetry breaking,
it turns out that hadron supersymmetry can be useful to obtain information about
baryon masses from meson masses.,
In 1990 I suggested9 that hadron supersymmetry might be more applicable to
hadrons containing one heavy quark than to hadrons containing only light quarks.
During the same year a similar idea was discussed by Georgi and Wise lo and Sav-
age and Wise l l within the context of heavy quark effective theory,l2 These latter
319
authors called the symmetry "superflavor symmetry" in order to distinguish it from
supersymmetric extensions of the standard model. However, I shall continue to use
the term hadron supersymmetry to refer to meson-baryon or antiquark-diquark su-
persymmetry. A brief discussion of hadron super symmetry has appeared in a recent
review of diquarks. 13
SUM RULES
In this talk I use hadron supersymmetry to obtain three sum rules relating the
masses of mesons to those of baryons. Equivalent sum rules have recently been ob-
tained in another context. 14 ,15
In what follows I neglect the effect of the size difference between an antiquark
and a diquark. This is not as bad an approximation as it might seem. Although an
elementary, or current, quark is pointlike, a constituent quark is clothed with a sea
of gluons and quark-antiquark pairs. Such an object may not be much smaller than
a two-quark pair.
In order to minimize the effect of mass differences between antiquarks and di-
quarks, I consider only sum rules which equate differences of meson masses to dif-
ferences in baryon masses. Also, in replacing an antiquark Q by a diquark Qq, I
minimize the mass difference by restricting myself to the case in which q is a light
quark, namely, either u or d. (I neglect the mass difference between the u and d quark
as well as all electromagnetic effects.) Therefore, I consider only the replacements
b -> bq, c -> cq, S -> sq, if. -> qq. (1)
To mmlmlze the effect of the different spins of an antiquark and a diquark,

I consider only ground-state hadrons, i.e., hadrons which are neither radially nor
orbitally excited. This restriction perturbatively eliminates the effects of spin-orbit
and tensor forces, leaving only the colormagnetic (or spin-spin) force to contend with.
However, it is possible in some instances to average the mass of a hadron with a
given quark content over its possible spin states,16 thereby eliminating the effect of
the colormagnetic force as well.
The spin-averaging process depends on the fact that in QeD perturbation theory,
the colormagnetic interaction energy has the form
(Vern) = - L(O'i' O'j)(Ai' Aj)(f(1'ij))/(mim j), (2)

i<j
where the O'i are Pauli spin matrices, the Ai are Gell-Mann matrices, and f(rij) is a
positive-definite operator which depends on the separation rij between quarks i and
j . In the Fermi-Breit approximation, f (r) is given by17
f(r) = 7w s 8(r)/6, (3)
where Os is the strong-interaction coupling constant. The expectation values of the

spin and color operators have been worked out a long time ago; see, for exmuple, ref.
16. It is not necessary to know the expectation value of f(rij), but just to assume
that it is independent of spin for all ground-state hadrons with a given quark con-
tent. Then, when the appropriate spin average is taken, the colormagnetic interaction
energy cancels.
320
Substituting a diquark for an antiquark according to the prescription of Eq. (1)
and taking the appropriate spin averages, I can write down the following sum rule:
(4)
where Q2 and Q1 denote two different quarks. Here, M stands for the mass of a
ground-state meson and B for the mass of a ground-state baryon, both averaged over
spm.
In the following, Q2 denotes s, c, or b, Q1 denotes q, s, or c, and the symbol for
a hadron denotes its mass. Then I obtain three independent sum rules from Eq. (4):
(3K* + K)/4 - (3p + 7r)/4 = (2~* + ~ + A)/4 - (N + .6.)/2, (5)
(3D*+ D)/4 - (3K* + K)/4 = (2~~ + ~c + Ac)/4 - (2~* + ~ + A)/4, (6)

(3B* + B)/4 - (3D* + D)/4 = (2~b + ~b + Ab)/4 - (2~; + ~c + Ac )/4, (7)
where I have averaged the masses over spin states according to the prescription given
previously.16
All the hadron masses appearing in Eqs. (5-7) have been measured,18,19 although
some of the measurements are preliminary.19 In Eq. (5), the left and right hand sides
are respectively
183 1 MeV, 184 1 MeV. (8)
In Eq. (6) the left and right hand sides are
1178 1 MeV, 1180 7 MeV. (9)
Lastly, in Eq. (7), the numbers are
3341 2 MeV, 3333 20 MeV. (10)
Agreement with experiment is excellent.

I can trivially rewrite Eq. (7) in a way which brings out that it also follows from
heavy quark effective theory.12 Putting all terms containing a c quark on the left and
all terms containing a b quark on the right, I obtain
(2~~ + ~c + Ac)/4 - (3D* + D)/4 = (2~b + ~b + Ab)/4 - (3B* + B)/4. (11)
From this equation, one easily sees that the right side is obtained from the left by the
replacement
c ----+ b. (12)
In the limit of heavy quark effective theory,12 making this replacement should not
change the value of the energy, as the heavy quark mass cancels out. However, heavy
quark theory cannot be invoked to explain why Eqs. (5) and (6) hold so well. For
that I need to use hadron supersymmetry.
In conclusion, although hadron supersymmetry is badly broken, if one uses it in
the right way, one can obtain relations between meson and baryon masses which agree
remarkably well with experiment.
321
Acknowledgements
This talk is based in part on work done with Enrico Predazzi and Renato
Roncaglia. Partial support has come from the U.S. Department of Energy.
REFERENCES
1. H. Miyazawa, Baryon number changing currents, Prog. Theor. Phys. 36:1266 (1966).
2. H. Miyazawa, Spinor currents and symmetries of baryons and mesons, Phys. Rev.
170:1586 (1968).
3. Yu. A. Gol'fond and E.P. Likhtman, Extension of the algebra of Poincare group generators
and violation of P invariance, JETP Lett. 13:323 (1971).
4. J. Wess and B. Zumino, Supergauge transformations in four dimensions, Nucl. Phys.
B70:39 (1974).
5. S. Catto and F. Giirsey, Algebraic treatment of effective supersymmetry, Nuovo Cimento
86:201 (1985).
6. S. Catto and F. Giirsey, New realizations of hadronic supersymmetry, Nuovo Cimento
99:685 (1988).
7. C.-s. Gao and T.-h. Ho, Baryonium and possible supersymmetry between quarks and
antidiquarks, Commun. Theor. Phys. (Beijing) 1:761 (1982).
8. C.-s. Gao and T.-h. Ho, A possible fireball theory with supersymmetry between quark
and antidiquark, Commun. Theor. Phys. (Beijing) 2:1045 (1983).
9. D.B. Lichtenberg, Dynamical supersymmetry in hadrons containing both heavy and light
quarks, J. Phys. G: Nucl. Part. Phys. 16:1599 (1990); D.B. Lichtenberg, Baryon
masses from meson masses in a model with broken supersymmetry, J. Phys. G: Nucl.
Part. Phys. 19:1257 (1993).
10. H. Georgi and M.B. Wise, Superflavor symmetry for heavy particles, Phys. Lett. B
243:279 (1990).
11. M.J. Savage and M.B. Wise, Spectrum of baryons with two heavy quarks, Phys. Lett.
B 248:177 (1990).
12. N. Isgur and M.B. Wise, Weak decays of heavy mesons in the static quark approximation,
Phys. Lett. B 232:113 (1989); 237:527 (1990).
13. M. Anselmino, E. Predazzi, S. Ekelin, S. Fredriksson, and D.B. Lichtenberg, Diquarks,
Rev. Mod. Phys. 65:1199 (1993).
14. D.B. Lichtenberg and R. Roncaglia, New formulas relating the masses of some baryons
and mesons, Phys. Lett. B 358:106 (1995).
15. D.B. Lichtenberg and R. Roncaglia, Properties of exotic mesons in a diquark model,
Indiana University report WHET 303 (1995), unpublished.
16. M. Anselmino, D.B. Lichtenberg, and Ff. Predazzi, Quark color-hyperfine interactions in
baryons, Z. Phys. C 48:605 (1990).
17. A. De Rujula, H. Georgi, S.L. Glashow, Hadron masses in a gauge theory, Phys. Rev.
D 12:147 (1975).
18. 1. Montanet et al., Review of particle properties, Phys. Rev. D 50:1173 (1994).
19. P. Jarry, B spectroscopy, in: XV International Conference on Physics in Collision,
Cracow, Poland (June 8-10, 1995).
322
CONSTRAINING THE QCD COUPLING FROM THE SUPERSTRING
HIDDEN SECTOR
Paul H. Frampton
Institute of Field Physics, Department of Physics and Astronomy

University of North Carolina, Chapel Hill, NC 27599-3255
INTRODUCTION
Superstrings are presently the main (only) hope of a mathematical underpinning for
both quantum gravity and its unification with particle theory.
A testable prediction remains elusive.
We can think of:
(i) Quantum gravity predictions: too small to be measurable in the forseeable future.
(ii) Calculation of a parameter in the Standard Model: I cannot fault that.
Here we imagine a third scenario:
(iii) The hidden (E~) sector impacts on the visible sector, and a connecting bridge
at Mstring provides a consistency check.
The work I shall describe was done in collaboration with T. Kephart at Vanderbilt
and with B. Keszthelyi and B. Wright at Chapel Hill[l].
Most attempts at superstring phenomenology over the last ten years have been very
speculative and, in any case, have concentrated on the link to grand unification in the
visible sector. By this we mean that in the heterotic E(8) X E(8)' superstring only, say,
the first E(8) is involved. Here we use recent results for superstring thresholds effects
to link the visible sector to the hidden sector (E(8)') and the implications of hidden-
sector supersymmetry breaking for the value of the QCD coupling constant as(Mz ),
and for cosmology. On the hidden side we necessarily follow a top-down approach
because nothing is known empirically. The treatment of the visible sector is partially
bottom-up since we input the low-energy data. Of course, the eventual goal is to have
sufficient measurements in both sectors to allow consistency checks at the string-scale
bridge.
SUPERSTRING THRESHOLDS AND UNIFICATION
In the visible sector, the standard model of particle theory is well established up to 100
GeV and a supersymmetric unification at MCUT rv 2 X 10 16 GeV has been advocated[2,
3,4] (see, however, Ref. [5]). In a superstring the relevant scale is denoted Mstring which
is related to Newton's gravitational constant (MPlanck or the reduced form MPlanck =
323
MPlanck /.j8;) through the string coupling constant gstring in a calculable way. Here
we shall assume an orbifold compactification with, below Mstring, minimal MSSM fields
having standard U(l)y normalization. Mstring lies somewhat higher (typically Mstring
= a few times 1017 GeV) than MauT so that successful minimal unification at the higher
scale necessitates sizeable superstring threshold corrections.
The hidden sector is often assumed to generate supersymmetry breaking via a gluino
condensate for ahidden(flo) - t 00 at flo '" 10 13 GeV[6]. At the same time this may gener-
ate Cold Dark Matter(CDM) and have consequences for inflation and the formation of
structure in the visible sector. Big Bang Nucleosynthesis(BBN) constrains the massless
degrees of freedom in the hidden sector and thereby, in principle, the possible breakings
of the hidden E(8)'; however, this will depend on the ratio (I') of the temperatures of
the hidden to visible sectors.
We assume no matter fields are present in the hidden sector, so there the superstring
thresholds vanish. Thus, the renormalization group equation for the coupling constant
corresponding to any subgroup (G') of E(8)' is:
-1( ) _
aa' fl -
-1
astrmg
+ (b2a ')[ n (Mstrin g ) (1)
7r fl
The value of fl = flo where aN(flo) = 00 is thence given by:
flo = (Mstring)exp [ (astring)

-1
b27r]
a
, (GeV). (2)
To obtain the hidden-sector gaugino (x)condensate at e.g. flo '" 1013 Ge V (correspond-
ing to visible SUSY breaking at '" 1 TeV) we consider -~ba' = C(G'), the quadratic
Casimir, ranging from 2 (for SU(2)) to 30 (for E(8)); recall that C(G') = N for SU(N)
and 12,18,30 for E 6 ,7,8 respectively. This procedure fixes Mstring. The scale flo assumes
the gravitino (\fI) mass arises from an effective coupling W\fI < XX > /M~lanck> i.e.
IlU Mf,lanck = 1OOGeV '" 1TeV. This gives the range flo = 0-
4) X 10 13 Ge V by allow-
ing either MPlanck or M Planck in the denominator. The results for Mstring are displayed
in Fig. 1.
In the visible sector the standard SU(3)c x SU(2)L x U(l)y theory is well established
up to 100 GeV. By assuming supersymmetry which is broken at", ITeV in the visible
sector, the three running couplings ai(fl) (i = 1,2,3) unify at MauT '" 2 x 10 16 GeV. In
the superstring there is another scale, Mstring, which takes the value[7, 8]:
Mstring = (5.3 X 10 17 Ge V) X gstring (3)
where gstring is a dimensionless coupling constant related to the real part of the dilaton
superfield expectation value by g;t;ing = ReS. According to general arguments[9, 10],
one expects the superstring to be sufficiently strongly coupled that astring = g;tring /47r
should satisfy e.g. astring > 0.01 which implies gstring > 0.1; we shall see that phe-
nomenological consistency also demands that we are well above these values. At the
same time one requires, for a perturbative low-energy theory, that astring < 1, i. e.
gstring < 3.6. For all these cases we see numerically that Mstring lies above the old GUT
scale ( and becomes the new, higher, effective GUT scale) while lying safely below
MPlanck, even below the reduced MPlanck : MauT < Mstring < MPlanck.
The renormalization group equations in the visible sector of the superstring have
the form for fl :s; Mstring:
324
1'0 = I.OxI0 13GeV -
I'o=O.4xlO'3GeV .
1'o=4.OxI013GeV ----.
o 5 10 20 25 30
Figure 1. M.tring as a function of the hidden sector gauge group subject to the gaugino
condensation condition. Solid lines correspond to f..lo = 10 13 GeV while short(long) dashed
lines are for f..lo = 0.4(4.0) X 1013 GeV.
1 _ ka ba I M;tring 1 ~
~() - - 2 - -
9 a f..l 9.tring
+ 167r 2 n--
2 -
f..l
+ 167r 2 a (4)
Here ba are the renormalization group f3 functions for the gauge groups (a = 1,2,3);
and ~a represent the corresponding superstring thresholds[ll). In orbifold models,
~a = - L b~(i)~(i) where ~ is a common factor for the orbifold subplanes i=1,2,3 and
b~(i) is given by[12, 13):
(5)
while, summing over the three subplanes gives:
(6)
in which T(Ra) are the Dynkin indices and nRm are the modular weights of the light
matter fields in irreducible representations Ra.
If the (1,1) moduli Ti are either all equal T = Tl = T2 = T3 or one of them is by
far the largest T = Tl ~ T2, T3 then we may write ~a = b~~ where b~ = Li b~(i) for
the former case and b~ = b~1 for the latter case. In either case, ~ is a common function
whose explicit form in the large T limit is given in [13) for simple cases by
(7)
325
where 'T/ is the Dedekind function.
Elimination of 9string gives the generalization of the GQW equations[14]:
sinZ(}w(Mz) = _k_z__ _ k_l_aem(Mz) [Aln (M;trin g ) _ A/~] (8)

kl + kz kl + kz 411" M~ ,
-1(M ) _ k3 [-I(M) 1 1 (M:tring) 1 (9)

z - k1 + kz a em z - 411" B n M~ + 411" B~
I ]
as ,
where A = fbI - bj , B = b1 + bz - klt,k. b3 and A', B' are obtained from A,B by the
exchange ba {=:} ba . Note that kl = 3' kz = k3 = 1. For the supersymmetric standard
I 5
model one has the values A = -'

B = 20.
Given Mstring from the hidden sector analysis, and a given value of , = B' / A', we
can now determine from Eqs. (8,9) the value of A/~ and as(Mz) respectively, given
the range of allowed empirical values for sinZ(}(Mz) and aem(Mz) which,taking into
account low-energy MSSM, are currently[15]:
sinz(}w(Mz) = 0.2313 0.0003
a;;;,(Mz ) = 128.09 0.09

Since Eqs. (8,9) determine as(Mz) only in terms of the ratio, we must try to
determine its value. As a first example we use the compactification of the heterotic
string on the Z~ orbifold, as discussed in[12, 13]. In that particular example, for the
three complex planes corresponding to the three two-dimensional subtori within the orb-
ifold, the MSSM states were assigned modular weights as follows: nQ".,3 = (0, -1, 0);
nD,,2,3 = (-1,0,0); nU, = (0, -!, -!); nU2,3 = (-~, -Jf, -V; nL, = (-Jt, -~, -~);
nL2,3 = (-Jt, -~, -~); nE, = (-1,0,0); nE2,3 = (-~, -Jf, -~); nH = (-!, -~, -~);
nIl = (-', -~, -~). Using the definitions given above for the separate orbifold sub- .
Planes i -- I '23 this gives b1,z,3 = _l!i. _145 ~ yielding b' = _~. bl ,z,3 - _2. -~ +1
" I Z' 2' 2 I 3' Z - 2' 4' 4
yielding b; = -8; and finally b~,2,3 = -~, -~, +~ yielding b; = -7. In this case, the
values of A', B' defined above are A' = -2 and B' = -6 giving, = +3. For the other
viable orbifolds cited in [13] the results are similar. By rescaling ~, all that matters is
, which takes the value, = +3 in Z6 and Zz x Zz orbifold models, just as in the Z~
case.
Of course, a range of , is possible for a specific construction. But for all these
special cases one has, = +3 so in Fig. (2) we show the relationship between Cz( G' )
and as(Mz) given, = +3 and fJo = 1013 GeV in Fig. (1). In any specific superstring
unification, the three quantities " fJo and C2 (G ' ) determine as(Mz).
The range of as(Mz ), if we assume Cz(SU(2)) = 2 ~ C2 (G ' ) ~ C2 (E(7)) = 18, to
allow the CDM hidden photino discussed below, is seen from Fig. (2) to be:
0.1215 ~ as(Mz) ~ 0.1270
This is to be compared to the allowed LEP range as(Mz) = 0.121 0.005[15].

In the above, we h~ve taken the particular value I = +3 which occurs in three
simple examples given in [13] seriously. If we instead regard, as an arbitrary rational
number and input fJo = 1013 GeV and the full allowed ranges of aem(Mz),sinZ(}(Mz)
and as(Mz) = 0.121 0.005 [15J then we find 2.62 ~ I ~ 3.73 for 2 ~ C 2 (G ' ) ~ 18.
This is illustrated in Fig. 3 where the three cases, = 2.62,3.00 and 3.73 are shown,
each for the central values of aem(Mz) and sin 2 (}(Mz) and for the extremes of the
allowed ranges. Note that as(Mz) is maximized for aem(Mz) maximized and sin 2 (Mz)
326
0.13 .-------r-----r---,---....-----r-----,
y=3.0 ~o=l.OXlOl3 -
~=O.4XlOl3 -----.
110 = 4.OX10 13 ----.
0.128
0.126
0.124
0.122
0.12 '--_ _--'-_ _----''--_ _.....L-_ _----L_ _ _..J....-_ _---'

o 5 10 20 25 30
Figure 2. as(Mz) as a function of the quadratic Casimir for the highest rank hidden sector
gauge group for an orbifold model with 'Y = 3 and taking Po = 1013 GeV (solid curves). The
upper and lower solid curves take into account the uncertainties in sin 2 Ow and a em . The
shifted short(long) dashed curves indicate the effect of changing Po to 0.4(4.0) X 10 13 GeV.
minimized, and vice versa. These results are relatively insensitive to po. Fig. 3 provides
constraints on" and C2 (G'), for orbifold constructions to be consistent with LEP data,
the gaugino condensate idea, and the hidden photino CDM candidate disussed below.
For a specific orbifold construction which can realize minmal superstring unification
with a definite hidden-sector gauge group, the range of allowed as(Mz) is even more
restricted.
All these results depend on being able to obtain the necessary value for the moduli-
dependent superstring threshold correction .6. determined by Eqs. (8,9) in a given
orbifold construction. For the combination (A'.6.), the solution is mostly sensitive to
C2 (G'), or equivalently Mstring; e.g. for 2 ~ C2 (G') ~ 18, we find 38 ;::: (A'.6.) ;::: 25,
independently of " for the cases we have considered in Figs. 2 and 3. Given the value
of A', there may be naturalness constraints on the size of 1.6.1 coming from the values of
the moduli VEVs by which it is determined, so one might be restricted to large hidden-
sector gauge groups. However, as shown in [16], it is possible to obtain sufficiently large
1.6.1 in a Z~ orbifold with A' = -2 for natural values of the moduli VEVs by including
continuous Wilson lines; the latter are, in any case, generally necessary for the required
symmetry-breaking pattern in both the visible and hidden sectors.
327
0.13 ,------.:"""T""""":""--...-------,-----,-,-------,,-----,
y~3.00 -
--- .. y . 373 . ... .

---._ 1: _2 6~ _:~-- '
-
--- ... - ... -
--~- - ...
0 .125
N
~ 0.12
0.115
0.11 ~--~---~----~--~--~----~------~
o 5 10 20 25 30
Figure 3. O'.s(Mz) as a function of the quadratic Casimir for the highest rank hidden sector
gauge group for three values of I (2.6,3.0,3.7) and taking ,",,0 = 10 13 GeV. The horizontal lines
give the current LEP limits on O'.s( Mz) assuming the MSSM. The range 2.62 < I < 3.73
is that which just allows at least one U(1) component in the hidden sector gauge group.
Uncertainties in sin 2 Ow and O'.em are indicated by the shaded regions.
COSMOLOGY
Now we turn to the cosmological aspects and ramifications for the hidden sector. From
the above, an illustrative scenario is where 9string ~ 0.7 and the condensate occurs
in an SU(5) gauged subgroup of E(8)'; for example, the breaking of E(8)' by Wilson
lines could give the rank-8 subgroup SU(5) X SU(4) X U(l). The SU(4) condensates
are sufficiently heavy to decay gravitationally before BBN[17] and the hidden photino
associated with the U(l) will have a mass comparable to the visible supersymmetry
breaking scale of '" 1 TeV. Let us now pursue the general idea that such a shadow
photino is the origin of some, or all, cosmological dark matter.
A serious constraint on the number of hidden sector massless degrees of freedom
arises from the agreement of the visible standard model with BBN. The hidden sector
photons change the effective number of degrees of freedom according to:
4

ge!! =

9 visible +

9hidden r;-
( Thidden )
(10)
where 9~isible is the visible-sector degrees of freedom at nucleosynthesis (10.75 for e [3.5],
lIi[5.25] and I [2.0]) while 9hidden = 2n:y,hidden for n:y,hidden hidden photons. The upper
328
limit on 9:ff is rv 11.45[18] and so there cannot be any shadow photino unless Thidden <
T"{: if Thidden = T,,{, no such extra massless state is permissable. (Note that in the older
literature[19] the upper limit on 9:ff was weaker: 9:11 < 13).
As far as Eq. (12) and BBN are concerned, a key issue is the ratio of hidden to
visible temperatures at the BBN era. If we adopt a hidden photino mass in the range
100GeV - 1 TeV, and assume that it forms all the dark matter (D;y rv 1), then the
temperature ratio (Thidden/T"{) rv 10-3 .3 to 10- 3 . 7 , since the number density of hidden
photinos n;y,hidden rv TKidden just as n"{ rv T; and we assume n;y,hidden = n"{ at a very
early era when T = T"{ = Thidden close to T = MPlanck.
Consider a model with normal sector gauge group G and hidden sector gauge group
G*, whose origins are both in our heterotic superstring theory. The symmetry breaking
patterns, and herefore the phase transition structure of such a model, can be quite
complicated, and the temperatures T"{ and Thidden of the two sectors can evolve very
differently, but are typically intertwined by inflation.
As an example, suppose that a dilaton field begins to develop a VEV and acts as
an inflaton affecting equally the visible and hidden sectors. The universe supercools
exponentially since the scale factor increases exponentially and RT = constant. Most
of the required e-foldings may be accomplished in this way during a dilaton-induced
inflationary phase. The final, say, rv 8 e-foldings may arise from an inflation induced, for
example, by a superstring modulus field which couples to the visible matter and not to
the hidden sector. Reheating can then occur in the visible sector back to approximately
the critical temperature, while no reheating is possible in the hidden sector. This leads
to a temperature ratio between hidden and visible sectors e- 8 rv 10- 3 .5 as required. This
is just one possible scenario to illustrate how enough dark matter could be generated
to make D = 1.
Finally, we come to the testability of our dark matter proposal. The hidden photinos
have a Jeans mass, MJeanS) given by MJeans = M],lanck/(m"{,hidden)2 rv 10 51 GeV rv
1O- 6 M 0 for m"{,hidden = 1TeV; for the lower value m"hidden = 100GeV, MJeans rv
10- 4 M 0 . Since this is CDM we expect it to have clumped gravitationally at this scale
and so our galactic halo will be comprised of these hidden objects. The accretion
of hidden photinos into such MACHOs will cause gravitational microlensing of distant
stars, and detection of the resultant temporary achromatic amplification is a possibility.
The method[20, 21] is practical for dark matter objects in the range 10- 7 M0 to 10 2 M 0 .
The duration of a microlensing event'scales as M~, and since for rv 10- 1 M 0 , the
observed event durations are a few weeks, the shadow-photino MACHOs should have
microlensing event durations of order a few hours or days. Detection of such events
by dedicated searches would support the interpretation suggested here as the origin of
MACHOs in the required mass range. Although the details are model dependent, such
U(l) factors in the hidden sector are generic in superstring theories and hence give rise
to such CDM candidates.
SUMMARY OF MAIN POINTS
Will the first testable prediction of the superstring be in quantum gravity? calculation
of a parameter in the SM? OR hidden sector impact on the visible sector?
Assuming that the hidden sector provides supersymmetry breaking and a CDM
candidate gives constraints on the visible sector. For a class of simple orbifold models
we found e.g. 0.121 ~ as(Mz) ~ 0.127 for string unification to be consistent.
This work was supported in part by the U.S. Department of Energy under Grant
DE-FG05-85ER-40219, Task B.
329
REFERENCES
[1] P. H. Frampton, T. W. Kephart, B. Keszthelyi and B. D. Wright, UNC-Chapel

Hill Report IFP-714-UNC (1995).
[2] U. Amaldi, W. De Boer and H. Furstenau, Phys. Lett. B260, 447 (1991).
[3] P. Langacker and M. Luo, Phys. Rev. D44, 817 (1991).
[4] J. Ellis, S. Kelley and D. V. Nanopoulos, Phys. Lett. B249, 441 (1990).
[5] U. Amaldi, W. De Boer, P. H. Frampton, H. Furstenau and J. T. Liu, Phys. Lett.

B281, 374 (1992).
[6] J. R. Derendinger, L. E. Ibanez and H. P. Nilles, Phys. Lett. 155B, 65 (1985)

M. Dine, R. Rohm, N. Seiberg and E. Witten, Phys. Lett. B156, 55 (1985)
[7] V. Kaplunovsky, Nucl. Phys. B307, 145 (1988); Erratum: ibid, B382 436 (1992).
[8] V. Kaplunovsky and J. Louis, Nucl. Phys. B444, 191 (1995).
[9] M. Dine and N. Seiberg, Phys. Lett. 162B, 299 (1985).

[10] T. Banks and M. Dine, Phys. Rev. D50, 7454 (1994).
[11] L. Dixon, V. Kaplunovsky and J. Louis, Nucl. Phys. 355, 649 (1991)
G. L. Cardoso and B. Ovrut, Nucl. Phys. B369 351 (1992); Nucl. Phys. B392,
315 (1993).
S. Ferrara, C. Kounnas, D. Lust and F. Zwirner, Nucl. Phys. B365, 431 (1991)
J. P. Derendinger, S. Ferrara, C. Kounnas and F. Zwirner, Nucl. Phys. B372, 145
(1992).
[12] L. E. Ibanez, D. Lust and G. G. Ross, Phys. Lett. B272, 251 (1991).
[13] 1. E. Ibanez and D. Lust, Nucl. Phys. B382, 305 (1992).
[14] H. Georgi, H. R. Quinn amd S. Weinberg, Phys. Rev. Lett. 33,451 (1974).
[15] P. Langacker, talk at SUSY-95, Palaise~u, France. ITP Santa Barbara / Univer-
sity of Pennsylvania Report NSF-ITP-95-140 / UPR-0683T (October 1995) hep-
ph/9511207.
[16] H. P. Nilles and S. Stieberger, Report TUM-HEP-225/95 SFB-375/22 (October

1995) hep-th/951 0009.
[17] J. E. Kim and H.P. Nilles, Phys. Lett. B263, 79 (1991).
[18] C. J. Copi, D. N. Schramm and M. S. Turner, FERMILAB-PUB-95-104-A (June

1995) astro-ph/9506084 and FERMILAB-PUB-g5-257-A (August 1995) astro-
ph/9508029.
[19] E. W. Kolb, D. Seckel and M. S. Turner, Nature, 314, 415 (1985).
[20] B. Paczynski, Ap. J. 304, 1 (1986).
[21] K. Greist, Ap. J. 366,412(1991).
330
WEAK INTERACTIONS WITH ELECTRON MACHINES
A SURVEY OF POSSIBLE PROCESSES
S.L. Mintz, M.A. Barnett, and G.M. Gerstner
Physics Department
Florida International University
Miami, Florida 33199
M. Pourkaviani
MP Consulting Associates
714 Fox Valley Dr.
Longwood, Florida 32701
INTRODUCTION
The field of weak interactions is now a reasonably old one. The first beta decays
were observed in the first part of the present century. By the 1930's, a theory of beta
decay of the current-current form had already been proposed by Fermi. By ,the 1950's
this work was extended by Feynman and Gell-Mann in the form of the V - A theory
which was applicable to all weak first order processes. On a fundamental level in the
next two decades the weak and electromagnetic interactions were combined and can
be said to be understood on the quark level. This work was initiated by Weinberg,
Salam, and Glashow and along with quantum chromo dynamics forms the basis of
what is called the standard model. This model, while clearly not fundamental, 'has
done very well at meeting many tests and is the present established model.
However it has proved very difficult to study the connection between the funda-
mental level and the nuclear level where experiments are actually carried out. This
is particularly true for the range of momentum-transfer-squared, q2, from the few
Me V 2 range to the few Ge V 2 range. In this energy range, the running coupling con-
stant, 0:, is large and perturbation methods are not useable. Lattice gauge methods,
while showing great potential, can not yet be used to tackle these problems. At the
moment it would therefore be highly desirable to have experimental guidance to help
in the application of phenomenological methods towards understanding the nuclear
weak hadronic current.
One of the difficulties associated with trying to understand even simple semi-
331
leptonic weak nuclear processes is the lack of experimental evidence to provide guid-
ance. At very low momentum transfer (q2 ~ 0) there exist a wide range of beta
decays. At q2 ~ -m~ there exist a number of partial muon capture rates. Above this
q2 range there exists little experimental evidence. There are now a few 1 ,2 neutrino
experiments over a range of q2 around that for muon capture. In principle neutrino
reactions, both charged current and neutral current would be a very useful tool for
studying weak semi-Ieptonic nuclear reactions. However neutrino experiments are
difficult to perform 3 and in most cases one must deal with a spectrum of neutrino
energies which tends to wash out details of interest. At present error bars on these
experiments are still above the thirty percent range making detailed analysis impos-
sible.
Furthermore the future of intermediate energy neutrino physics is very much in
doubt. The premier facility for such measurements in the United States is the Los
Alamos Meson Physics Facility,LAMPF. However its future is very uncertain and it
is the only facility presently able to do measurements with decay-in-flight neutrinos.
Thus we consider the possibility of using electron reactions in nuclei to study the
weak semi-Ieptonic nuclear interaction.
A weak electron induced nuclear process offers a variety of advantages and diffi-
culties. On the positive side the energy of the incoming electron can be varied through
a number of fixed and definite values. The angle of observation can also be adjusted.
Thus it should be possible by varying both of the quantities to obtain a fixed q2 for
different kinematical conditions. This would be of great use in isolating contributions
from the individual nuclear form factors. There are difficulties, however, associated
with reactions of the kind that we are considering. The most obvious difficulty is that
for a reaction such as e - + 3 He -+ Ve + 3 H ,for example, which we shall consider in
detail later the outgoing lepton is not observable. Thus it will be necessary to observe
the final state nucleus. Such observations are difficult because the momentum of the
nucleus may be low, the final state nucleus may be difficult to identify and there may
be competing processes which produce it. We shall discuss these points later.
There is another class of measurements which are possible and which do not
suffer from the difficulty of not having an observable lepton. These are the polarized
parity violating electron experiments. For facilities such as CEBAF, which will have
polarized electron beams, the difference between the cross sections for right and
left handed electrons on nuclei will depend necesarily upon the weak part of the
interaction. However, as we shall see, the dependence upon the weak hadronic current
for these processes is not strong and may be difficult to observe. They may be
useful,nevertheless, for determining more accurately fundamental constants such as
the Weinberg angle.
We will therefore examine the possibility of using electrons both for polarized
parity violating experiments amd for direct electron induced weak scattering processes
sometimes known as inverse beta decays. We will restrict ourselves to exclusive final
states for both types of processes. Inclusive processes are certainly possible but at
present they are much more difficult to analyze than processes to specific final states.
In fact it has been noted for neutrino interactions that theory and experiment are in
good agreement for exclusive processes but experiments disagree among themselves
by factors of three as do the various theoretical calculations for inclusive processes.
Nonetheless in parity violating polarized electron scattering processes we have the
possibility of examining the weak neutral current and for inverse beta decay we look
at charged weak current. We now proceed to consider these cases separately.
332
POLARIZED PARITY VIOLATING ELECTRON SCATTERING
Weak electron neutraJ current electron scattering from nuclei and nucleons is im-
possible to observe under ordinary conditions because it is overwhelmed by ordinary
electromagnetic scattering. However if one looks at the difference between the cross
sections4 taken first with left handed polarized electrons and then with right handed
polarized electrons one obtains and expression which depends upon the weak neutral
current. The quantity normally calculated is called A, the asymmetry 5 and is given
by:
M(L) _ du(R)
A- dO dO (1)
- M(L) +
du(R)
dO dO
where the denominator of this expression is essentially the electromagnetic scattering
cross section as the weak contribution is negligible. The transition matrix element
for this neutral electroweak process may be written as:
e2 I (12
Mki = 2"U 'Yp,U< klJtm 0)1 C> + .GIOU
2
'Yp,[Yv + gA'Y5]U< kIJweak(O) 1 C>
I P, 12
(2)
q v
where k is the final state of interest and where we have chosen a 12C ground state
as an example. Because only the weak-electromagnetic interference terms survive in
the numerator of Eq.(l), when we square Mki only four terms are relevant and they
are of the form:
J.emJ.At
p, v
Jp, em Jvt
V
+ J.AJ.emt
p,v
Jp, JV
V
emt + JoVJoVt
p,v
Jp, em Jvt
A
+ JoVJ.em
p,v A
Jp, JV emt (3)
where lower case j refers to the lepton current and upper case J refers to the hadron
current.
After some calculation, the numerator of Eq. (1) is seen to be proportional to the
quantity, IMkil~v where PV stands for parity violating. This quantity is given by:
IMI2PV = 4V2mmi 1T2mUGq2cos 00 (_ q2/2)F(3)

M q
( 2)
e P
X {9V(P Q + p' . Q]F13} (q2) (4)
+gA[( _q2 /2)Q2 + (Q. p)2 + (Q. P')2][1 _ 2 sin 2 Ow] Fl;) (q2) }
4mimp
with the form factorsF13 ) and Fi.:) defined by the expressions for the axial and vector
currents for these transitions, namely:
and
F(3) ( 2)
< 12C*Iv:(3)112C
v
> = -iV2m-f qp,cPQu M q
vp,pu." 2mi 2mp 0
(5b)
In Eq.(4), the quantities gv and gA are given by:
gv = -1 + 4sin2 Ow (6a)
333
and
(6b)
where Ow is the Weinberg angle,sin 2 Ow = 0.2259 .0046. Due to this value for
the Weinberg angle, gv is quite small,about .0964. This strongly suppresses the
dependence of A, the asymmetry, on FA as can be seen from Eq.(4). Because the
numerator and the denominator depend only on F~) if the gv term is ignored, all
dependence on the hadronic current disappears from A. Thus it might be possible to
investigate the fundamental couplings from a reaction such as this but it is doubtful
that much information about about the hadronic current will be found.
THE EXCLUSIVE CASE
It is possible however to look at reactions of the form of e- + 3 He --+ Ve + 3 H.

This reaction is an inverse beta decay and has a matrix element similar both to that of
muon capture and that for a neutrino reaction. Thus it would be possible to study the
weak hadronic current in nuclei with a reaction such as this one, at least in principle.
Therefore we shall take a closer look at this process and shall discuss a few related
reactions.
The process we are considering can be described as a first order weak interaction.
Although values of q2 as high as 30 GeV 2 /c 2 are possible for the energies being
considered here, and we are assuming that an experiment of this type would be
performed at CEBAF where electron energies of 4 Ge V are presently available and
6 Ge V will be available shortly. This is still small compared to the mass squared of
the intermediate vector boson, approximately 6400 MeV 2 , and so we are justified in
writing the interaction as:
If the quantity <3 HlJl (OW He> were known it would be immediately possible to
obtain a differential cross section. We note that here the hadronic current,JI-'(O) is
written as:
(8)
where VI-' and AI-' are the vector and axial vector parts of the weak nuclear current.
We make use of the elementary particle model6 ,7 to describe the matrix elements
of the nuclear vector and axial vector currents in a Lorentz invariant way in terms
of spinors for the spin one-half nuclei 3 He and 3 H, linearly independent vectors, and
form factors. These matrix elements are well known 7 and may be written as:
(9a)
and
(9b)
where i is the initial nucleus and f is the final nucleus, 3 He and 3 H respectively. The
structure of the nucleus is contained in the four form factors,Fv(q2),FM(q2),FA(q2),
334
and Fp(q2). Thus if we are able to determine these four form factors we can write
and evaluate a transition matrix element for the 3 He +-+3 H transition.
The form factors of the vector current matrix element may be evaluated in
a straight forward way by making use of the conserved vector current hypothesis.
Making use of the standard current commutation relations, [I;, Jjl = if;jkJt we may
write:
(10)
where VI' = Ji - iJ!f and I = h + iI2 Taking into account that 3 Hand 3 He are
mirror nuclei with Iz = -~, ~ respectively, we have:
(11)
which leads to:
The matrix elements of the electromagnetic currents for the nuclei 3 He and 3 H may
be written as:
. v F2( 2)
< klJ;mlk >= Uk[F;(q 2 hl' + ZO"l'v~ k q lUk (13)
mp
where k stands for either 3 H or 3 He. From this we obtain the relations:
FV(q2) = FlHe(q2) - FlH(q2) (14a)
FM(q2) = F;He(q2) - F]H(q2). (14b)

Electron scattering data exists 8 - 16 for determining the form factors,F 1 and F2, for
both 3 He and 3 H from q2 ::: 0 to q2::: - 50m;. This yields the following results for
Fv and FM :
(15a)
for Iq21 :S 24.5m;

2 2
Fv(q2) = Fv(O) cos 2 ( 14.~~M; )(1 - 4.3~m;' )-2 (15b)
for Iq21 > 24.5m; and
(16a)
(16b)
for Iq21 > 43.0m;.

We note that at low values of Iq21 , the cosine terms in eqs. (15a),(15b),(16a),
and (16b) are approximately one, yielding the dipole form factors familiar from earlier
work 17 ,18. At higher Iq21, the cosine terms model the diffraction type minimum noted
in the electromagnetic form factors. We shall discuss this point later.
335
It is still necessary to determine the axial current form factors. For the pro-
cess which we are considering here, all of the terms of the transition matrix element
squared which contain a factor of Fp(q2) also contain the lepton mass squared as
a factor. Because the lepton here is an electron, this factor,m~, is very small com-
pared to all other energies, momenta, and masses which appear in the treatment of
this process. Thus terms proportional to Fp are suppressed and do not contribute
in a measurable way to the transition matrix squared and we may ignore them.
We still must determine F A (q2) however which does contribute substantially to the
transition matrix squared. The quantity FA(O) can be determined 18 from the beta
decay, 3 H ---+3 He + e- + i), which takes place at q2 c:::: O. There is no direct way to
determine the q2 dependence of FA. However a result by Kim and Primakoff6 based
upon the impulse approximation but not making use of the actual form of the nuclear
wave functions yields:
(17)
This relation works well for muon capture for our reaction 16 ,17 as well as for many
others 13 as well as for neutrino reaction,v + 12C ---+ e- +12 Ngs over 1 ,2 the Michel
spectrum. It has however never been tested at high Iq21. We shall make use of it here
with the expectation that it should work well at Iq21 up to m~ and should serve as as
a first approximation above this range.
We may now proceed with our calculation. We note that the values for the
necessary form factors at q2 = 0 have been previously determined 16 ,17 and are given
by:
Fv(O) = 1.0 (I8a)
FM(O) = -5.44 .0015 (I8b)
and
FA(O) = -1.212 .004 (I8c)
where eqs. (I8a) and (I8b) are determined by eve from electron scattering data and
eq.(I8c) is determined from beta decay data as has already been noted. We note that
the errors associated with the form factors at q2 = 0 are quite small and in the case
of Fv(O) would be zero because it is just the difference in charge between the initial
and final state nucleus. At larger Iq21 the errors are mostly due to an uncertainty in
the mass in the dipole part of the form factors. This gives rise to uncertainties of the
order of 11 percent at q2 ~ -m~. These uncertainties increase at higher Iq21.
The matrix element squared for this process is readily calculated from the first
336
order matrix element, eq.(7). The result is:
(19)
We note that every term in this matrix element squared is proportional to the neutrino
and to the electron energy. The differential cross section can now be calculated by
standard methods. The result is:
da
(20)
do'
where Pf and E f are here the magnitude of the three momentum and the energy
of the the final state nucleus respectively. We note here the presence of E in the
denominator of eq.(20) which cancels most of the direct dependence of the matrix
element squared ,eq.(19), on the incoming electron energy.
We are now able to evaluate eq.(20) and we do so for electron energies of 0.1
GeV,O.5 GeV,l.O GeV,2.0 GeV,4.0 GeV and 6.0 GeV. The results are shown in figure
1. In figure 2 we plot the results for the 0.1 Ge V case so that the peaking is more
visible. In figure 3 we plot the results for the 0.5 Ge V and the 2.0 Ge V cases again
so that the structure of the peaks are more visible.
In figure 4 we plot the contributions of the form factors,FA,Fv , and FM to the
differential cross section for an incoming electron energy of 0.5 Ge V and in figure 5
we plot the contributions from the same form factors for an incoming electron energy
of 2.0 GeV. In figure 6 we plot the contributions of the form factors FM and Fv
to the differential cross section for an incoming electron energy of 2.0 GeV over the
lower angular range of the final state nucleus so that the structure is more apparent.
Finally in figure 7 we show Iq21 as a function of outgoing nucleus angle for incoming
electron energies of 0.5 GeV and 2.0 GeV.
Three considerations lead to the shapes of the differential cross sections observed
in fig. 1. First the form factors which are functions of q2 occur to the second power
in eq.(13). Because each form factor contains a factor of the type:
(21)
337
10.--------,--------,--------,---------,--------,--------,
da)dn
(10- 40 cm 2 ;,,)
.. J
--
:I
~ =--
40 50 60 70 80 90
e (deg)
Fig. 1 Differential cross section for reaction C + 3 He ---+ Ve + 3 H as a function of

outgoing nucleus laboratory angle. The solid, dense dotted, solid, dotted, thin solid,
and solid curves are for incoming electron energies of 0.1 GeV,0.5 GeV,l.O GeV, 2.0
GeV, 4.0 GeV, and 6.0 GeV respectively in order of increasing peak height.
0.65 , . - - - - - - - - , - - - -__- - , . - - - - - - - - , . - - - - - - - , , . - - - - - - - , , . - - - - - - - , - - - - - - - ,
0.6
0.55
do/dO.
(10- 40 cm 2 ;,,)
0.5
0.45
0.4
0.35 L -______'---______'---______' - -____---'______---'______---'______--.J

o W W W ~ ~ w ro
e (deg)
Fig. 2 Differential cross section for the reaction e- + 3 He ---+ Ve + 3 H as a function

of outgoing nucleus laboratory angle showing the maximum clearly for an incoming
electron energy of 0.1 GeV.
338
4
da/dfJ
(10- 40 cm2 / ar)
oL-____~__====~====~~~____~~~~_L _ _ _ _ _ _~
30 40 50 60 70 80 90
e (dog)
Fig. 3 Differential cross section for the reaction e- + 3 He ----t Ve + 3 H as a function
of the outgoing nucleus laboratory angle showing the maxima clearly. The solid and
dotted curves are for incident electron energies of 0.5 GeV and 2.0 Gev respectively.
1.2,----,----,---,-----,-------,------,-----,------,---,------,
0.8
da/dfJ
(10- 40 cm2 /BI)
0.6
0.4
0.2
O~~~~~~~~~~--~----~---L--~--~~--~
w ro n w
e (dog)
~ ~ ~ ~ ~ ~ ~
Fig. 4 Contributions of the various form factors to the differential cross section
as a function of outgoing nucleus angle for an incident electron energy of 0.5 GeV.
The solid,dense dotted and dotted curves are the contributions of FA,Fv , and FM
respectively. The curves are obtained by setting all form factors but one at zero.
339
3.5 , - - - - - - , - - - - - , - - - - - - - - - - , , - - - - - - - - , - - - - . . , . - - - - - - - ,
2.5
da/dfl
(10- 40 em 2 Isr)
1.5
0.5
OL-______ _________ L_ _ _ _ _ _ _ _ _____ __
~ " ro n w " ~
e (deg)
Fig. 5 Contributions of the various form factors to the differential cross section
as a function of outgoing nucleus angle for an incident electron energy of 2.0 GeV.
The solid,dense dotted, and dotted curves are the contributions of FA,Fv, and FM
respectively. The curves are obtained by setting all the form factors but one at zero.
0.00014 ,-------,-----,----------,,------,----..,.-----;;f\,...---;
0.00012
0.0001
Be-OS
da/drl
(10- 40 cm 2 / sr ) 6e-05
4e-05
e (deg)
Fig. 6 Contribution of the form factor, Fv, to the differential cross section as a
function of outgoing nucleus angle for an incident electron energy of 2.0 Ge V for an
angular range below the peak values.
340
O~__- L_ _ _ _~_ _~_ _ _ _- L_ _ _ _l -_ _~====~==~~~~
o 10 20 30 40 50 60 70 80 90
e (deg)
Fig. 7 Plot of Iq21 as a function of outgoing nucleus laboratory angle. The solid and
dotted curves are for 0.5 Ge V and 2.0 Ge V incident electron energies respectively.
14,-------~--------,-------_.--------_,~~~--._------_.
12
10
da/dO
(10- 40 cm 2 /sr)
o~======~~~
W
e
W 00
(deg)
m W 00
Fig. 8 Differential cross section for the reaction e- + p --+ Ve + n as a function

of outgoing nucleus laboratory angle. The solid, dense dotted,dotted,dark solid,light
solid, and dark solid curves are for incoming electron energies Of 0.1 GeV, 0.5 GeV,
1.0 GeV, 2.0 GeV, 4.0 GeV, and 6.0 GeV respectively in order of increasing maximal
values.
341
i.e. a dipole factor, these form factors for Iq21 large compared to M2 rapidly
suppress the differential cross section. Secondly, IMl 2,the matrix element,eq.(13)
as noted is proportional to v, the outgoing neutrino energy. For low Iq21, because
qiJ. = eiJ. - viJ., the neutrino energy approaches the magnitude of the incoming electron
energy. This tends to drive up the cross section at low Iq21. Moreover the low Iq21
case corresponds to a large final state nucleus scattering angle as can be seen in fig.
7. Thus at low Iq21, this effect along with the reduced suppression caused by the form
factors, leads to the increases in the differential cross observed in figures 1, 2, and
3. The third factor which contributes to the shape of the differential cross section is
the factor,PI, which occurs in eq.(20). As Iq21 falls, most of the energy brought to
the process by the incoming electron is carried away by the outgoing neutrino. Thus
the energy carried by the final state nucleus approaches its mass and its momentum
approaches zero. This causes the rapid fall in the differential cross section at high
angle. Thus the gross features of the differential cross section are easily explained.
Moreover, a number of other features of the differential cross sections can be
understood by these same considerations. For low Iq21 i.e for the 0.1 GeV case where
Iq21 remains small compared to M2 (see eq.(15)),and where M2 c:: (4 - 6)m;, over
the entire range of outgoing nucleus angle , the form factors are almost constant
leaving the differential cross section relatively fiat, until it is suppressed by the falling
momentum, PI. By an incident electron energy of 0.5 GeV,lq21 is large enough to
suppress the differential cross section except at large angles so that a broad peak is
easily discernible, as can be seen in figures 1 and 3. As the incoming electron energy
is increased, the peak widths at half maximum shrink because higher values of Iq21
are achieved at increasingly lower angles as can be seen from fig. 7, thus suppressing
the form factors. However the increasing neutrino energy as Iq21 increases drives up
the maxima so that at 6.0 GeV, the differential cross section at maximum is about
15 times that at 0.1 GeV. Naively it might be expected to be 60 times larger but the
decrease in PI is also faster.
Thus kinematics determines the shape of the cross sections particularly at low
Iq21. Above this range the situation is unclear. Electron scattering data as previously
noted determined Fv and FM and data was available for the range of q2 to around
-50m;. The eve hypothesis was used to obtain these form factors. It is not
known if whole nucleus eve is appropriate at these energies however if the initial
and final state nuclei have definite isospin it seems likely that this would be the
case from the derivations given in eqs.(4) to (8b). However the situation for FA
is very different. The form for FA given by eq.(ll) was obtained originally from
a nucleons only impulse approximation but is extendable to include some meson
exchange current contributions 19 . As remarked,it does appear to work well for muon
capture but there is no reason to suppose that it should work for higher energy,
independently of whether eve remains good. In fact it would be very useful to
determine experimentally when eq.(ll) breaks down. However this seems unlikely
because at high Iq21, the cross sections are so small. From figs.3 and 5 it is seen that
away from the peak values the cross sections even for the .5 GeV case rapidly fall by
2 to 3 orders of magnitude making experimental observation exceedingly difficult.
We note that an earlier calcuiation 20 making use of shell model and impulse ap-
proximation techniques was performed for energies of 0.88 and 3.52 GeV. The results
are similar to those obtained here but are approximately 20 to 30 percent lower than
the results obtained here. This difference is not large for a weak interaction calcula-
tion of this type but we would remark that the elementary particle model treatment
used here has produced numbers for muon capture rates for the 3 H +--+3 He transition
342
in very close 18 agreement with experiment whereas an impulse approximation treat-
ment is lower than experimental values by 10 to 15 percent. This effect was noted
in reference 18 and attributed to smaller weak magnetic type couplings, FA and FM,
used in the impulse approximation calculation. Details of the calculation for reference
20 were not given in the paper but we expect that the reasons for the difference would
be similar.
We are interested in the possibility of actually determining the form factors Fv,
FM, and FA from experiments that might in the future be performed. As can be seen
from fig. 5, the contributions of Fv and FA to the differential cross section at low
Iq21 are of similar magnitude but that of FM is small. Thus if it proved possible to
perform an experiment near the peak of the differential cross section for energies in
the few GeV range at an electron accelerator facility, it might be possible to determine
Fv and FA simultaneously. This could be done by varying E, the incoming electron
energy and () f the outgoing nucleus angle so that q2 would remain fixed. The form
factor FM, could not be determined by this procedure, but if Fv is correctly given by
its CVC value it would seem likely that FM would also be correctly given by evc.
Accurate muon capture measurements might then make it possible to determine Fp,
the pseudoscalar form factor. It has proved difficult to experimentally determine
Fp but it would be very useful since Fp is predicted by various formulations of the
partially conserved axial current hypothesis,PCAC. At higher values of Iq21, as can
be seen from from fig. 6, the contribution of FM becomes substantial. However as .
we remarked above, the cross sections are probably too small to observe at this time.
We note at this point an interesting feature of the form factors at higher Iq21. In
fig. 6 it is seen that Fv has an oscillatory behavior for large Iq21. This is due to the
cosine factor in the form factors given in eqs.(15) and (16). The cosine factors are
present to fit the diffractive minimum noted in the region of q2 ~ -25m!. It would
be interesting to note if the succeeding minima are also present but the cross sections
are probably too small for any direct experimental observations.
In general this process,e- + 3He ----t v e + 3H, presents some real difficulties for
experimentalists. Although the cross sections as expected for this mirror transition
are relatively large, in fact approximately an order of magnitude larger than those 21
for 12C to 12 Bgs,the peaks are exceedingly narrow, particularly as E (and therefore the
cross sections) increases. Thus a small change in the angle of observation corresponds
to a large change in q2. Furthermore, the cross sections are largest where the energy of
the recoil nucleus is not particularly great compounding the observational difficulty.
Finally we note that although the 3 H nucleus has a definite signature via its beta
decay, it is not hard to think of other processes which also produce the same final state
nucleus such as pion electroproduction,e- +3 He ----t e- +7r+ +3 H and so it might be
necessary to observe the other outgoing particles or to find a kinematic region where
e- + 3 He ----t Ve + 3 H dominates. Another pair of reactions also produces the same
final state triton, namely e- +3 He ----t p + P + n followed by n +3 He ----t 3 H + p.
Here again it would probably be necessary to observe the proton to distinguish this
contribution.
We conclude this discussion by remarking that in the 12C ++12 Bgs transition 2 t,
we noted that there was a maximal scattering angle in general less than 90 degrees
for the outgoing nucleus and that the denominator in eq.(14) became infinite at that
value of Of necessitating the use of a wave packet to describe the outgoing nucleus.
For this process the effect is greatly diminished. This is because the expression for
343
the maximal 21 angle is given by:
sin (J f = 1 - ~ - i. (22)
Mi E
where (j = Mf - Mi. For our case delta is exceedingly small, namely .53 MeV instead
of the 17 MeV for the 12C case. Thus (J f is approximately 90 degrees and there is
need for the more elaborate treatment used in the earlier paper only for the 4 Ge V
and 6 Ge V cases. For these two cases we use a Gaussian wave packet given by:
(23)
to represent the final state nucleus where all notation and treatment is precisely as
in reference 19.
Thus the reaction e- + 3 He ~ Ve + 3 H, while presenting some advantages
such as a relatively large differential cross at its maximum also presents a host of
observational problems. Ultimately the experimentalists will have to decide if it is
presently practical to perform such an experiment.
It is possible to consider other reactions and we here briefly discuss two of them.
The first is the same reaction as we have just discussed but done directly on the
proton,Le e- + p ~ Ve + n. This reaction at first looks impossible because the final
state contains two neutral particles. However we have been assured that it is now
much more feasible to measure the neutron than would have been true formerly. The
matrix element is essentially the same as that for the 3 He case so that eq. (7) and
following hold except for the masses of the hadrons and for the precise form of the
form factors. We note that both of these processes are spin one-half spin one-half
transitions. Because neutrino scattering has been experimentally observed for this
system 2\FA(q2) is known to q2 = 1 GeV 2 directly, unlike the case for 3 He. In any
event the form factors are well known and we do not include them here. In figure 8 we
show the differential cross section for the inverse beta decay on a proton for the same
energies as that for the 3 He case. We see that the peaks are in general higher and
broader than those for the 3 He case. Thus the proton would be a promising target if
it were truly possible to observe the final state neutron. Calculations for the reaction
e- + 12C ~12 Bgs + Ve have also been undertaken 21 . However the differential cross
sections are tightly peaked and background is still a major problem. Finally, it has
been suggested 22 that e- + p ~ Ve + A or 23 e- + p ~ Ve + ~ might be possible
reactions in which to observe the weak hadronic current. These reactions have fewer
competing background processes. In the case of the former, A beta decays are well
measured and there is information on the electromagnetic form factors. However
there is a strong suppression from the Cabibo angle because the above reaction is
strangeness changing. In the latter case, a model independent calculation is not at
present possible.
Thus weak electron scattering in nuclei and nucleons has the prospect of en-
abling the weak hadronic current to be directly studied. However the problems are
formidable. They are mostly concerned with detecting the final state nucleus or other
hadron against the same states produced via competing background processes. In ad-
dition these cross sections for even light nuclei are severely peaked near the forward
direction leading to experimental difficulties at present. Nonetheless these processes
are worthy of consideration. This is particularly true because without electron beam
machine results, there will likely be very little new data on the weak nuclear hadronic
current. Ideally neutrino beams would induce the same or related transitions as those
344
induced by electron machines with far fewer background problems because the final
state lepton is observed. However the outlook for both new neutrino machines and
even for the old ones is bleak. Thus we must hope for improved technology to enable
this work to be carried out at electron facilities.
REFERENCES
1 R.C. Allen, Phys.Rev.Letters64, 1871(1990).

2 B.Zeitnitz,Prog. in Part and Nucl. Phys. 13,445(1995).
3 S.L. Mintz and M. Pourkaviani,Neutrino Reactions in Nuclei in the Large and
in the Small, Proceedings of the 1995 Coral Gables Conference on Symmetry in
the Small and in the Large Editors B. Kursunoglu, S.L. Mintz,and A. Perlmut-
ter,Plenum Press, New York,p.151(1996).
4 G. Feinberg,Phys. Rev. DI2,3575(1975).
5 D.H. Beck,Phys. Rev. D39,3248(1989).
6 C.W.Kim and H. Primakoff,Phys. Rev. 139,B1447(1965).
7 C.W.Kim and H. Primakoff,Phys. Rev. 140,B566(1965).
8 H. Collard et al.,Phys. Rev.138,B57(1965).
9 M. Bernheim et al.,Lett. Nuov.5,431(1972).
10 Z. Szalata et al.,Phys. Rev.CI5,1200(1977).
11 J. McCarthy,I. Sick, and RR Whitney,Phys. Rev.CI5, 1396(1977).
12 R Arnold et al.,Phys. Rev. LettAO,1429(1978).
13 P. Dunn et al.,Phys. Rev.C27,71(1983).
14 G. Retzlaff and D.M. Skopik,Phys. Rev.C29,1194(1984).
15 D. Beck et al.,Phys. Rev.C30,1403(1984).
16 F. P. Juster et al.,Phys. Rev. Lett.55,2261(1985).
17 J. Frazier and C.W. Kim,Phys. Rev. 177,2568(1968).
18 J.G. Congleton and H.W. Fearing, Nucl. Phys. A552, 534(1993).
19 C.W. Kim and H. Primakoff,Mesons in Nuclei,edited by M. Rho and D.H.
Wilkinson,p.68,North Holland, Amsterdam(1979).
20 T.W. Donnelly,E.L. Kronenberg, and B.E. Norum,Research Program at CEBAF,
Report of the 1985 Summer Study Group,CEBAF ,Newport News,Virginia,pp.12-
14(1986).
21 S.L. Mintz and M. Pourkaviani,Phys. Rev. C37,2249(1988).
22 M. Finn, private communication.
23 E. Brash,private communication.
345
SECTION VIII - EXACTLY SOLUBLE QUANTUM
MODELS
MATRIX ELEMENTS OF LOCAL FIELDS IN INTEGRABLE QFTI
G. DELFINO
Theoretical Physics, University of Oxford
1 Keble Road, Oxford OX1 3NP, United Kingdom
and
G. MUSSARDO
International School for Advanced Studies and,
Istituto Nazionale di Fisica Nucleare
34014 Trieste, Italy
1. Introduction
In the recent years there has been a remarkable progress in the computation of
correlation functions of two-dimensional statistical models nearby the critical points. It
is well known that correlation functions of local fields have a quite distinct behaviour
at and away from criticality. While at the critical points the correlation functions
of the statistical models fall into a scale-invariant regime and their computation may
be achieved by solving the linear differential equations obtained by the representation
theory of the infinite-dimensional conformal symmetry [2], away from criticality their
determination may be quite difficult for the presence of mass scales which destroy the
conformal invariance of the critical point.
Away from criticality, the very first problem to solve is to determine the spectrum
of the massive excitations. As shown originally by Zamolodchikov [3], there are special
cases where we can have an exact answer to this question. This occurs for Renor-
malization Group trajectories originating from the fixed points which are associated
to relativistic integrable field theories. Due to an infinite number of conservation laws
present in these field theories, the scattering processes in which the massive excitations
lTalk presented by G. Mussardo
349
are involved are completly elastic and factorizable. Moreover, they satisfy the so-called
bootstrap principle which permits to consider asymptotic states as well as bound states
on the same footing. In many cases of physical relevance, this statement acts as a dy-
namical principle and by using it, one is often able to determine completly the spectrum
of the theory and the scattering amplitudes. An explicit example will be provided in
the sequel.
Once we have obtained the exact spectrum and the exact scattering matrix of the
field theory under consideration, we can proceed further and compute the correlation
functions of its local operators. This goal can be achieved by using spectral represen-
tation methods based on the determination of exact matrix elements of local operators
[5, 6]. Several QFTs have been solved through this approach and some examples may
be found in the references [7, 8, 9,10, 11, 12]. One of the main reasons of the success of
this approach is the fast rate of convergence shown by the spectral series. This impor-
tant quality of the approach allows us to extract the exact value of quantities related
to the large distance properties as well as to control their ultraviolet limit.
It is worth to notice that one of the first exact computation of correlation functions,
the spin-spin correlator in the thermal deformed Ising model (T #- Te , h = 0) obtained
by a tour-de-force calculation in the classical papers [14, 15, 16, 17], can be elegantly
recovered by a Form Factor approach, as shown in the references [5, 8, 18, 19]. In
this talk we are concerned with the other deformation of the Ising model, namely with
the Ising model at T = Te but at h #- O. The calculation of the spin-spin correlation
function G(x) =< a(x)a(O) > of the two-dimensional Ising Model in a magnetic field
has been a long-standing problem of statistical mechanics, recently solved in [1]. In this
talk we will illustrate the main features of the solution and we will also comment on
some interesting aspects of the method we used.
The first thing to do is to discuss the on-shell solution obtained by Zamolodchikov
[3].
2. Scattering theory of Ising Model in a Magnetic Field
For small values of the magnetic field h, the system is still at T = Te although the
coupling to the magnetic field induces a mass scale M (h) in the problem and destroys
the long-range fluctuations of the critical point. The system is integrable though and
its properties are ruled by the root system of the E8 Dynkin diagram [3, 4]. In fact,
the spin of its conserved charges are s = 1,7,11,13,17,19,23,29 (mod 30), which are
nothing else but the Coxeter exponents of E 8 . Moreover, the spectrum is given by eight
different species of self-conjugated particles A a , a = 1, ... ,8 whose mass values are
proportional to the components of the Perron-Frobenius eigenvector of the E8 algebra.
The scattering processes in which the eight particles are involved are completely elastic
(the final state contains exactly the same particles as the initial one with unchanged
momenta) and, due to the factorization of multi particle scattering amplitudes induced
by integrability, they are entirely characterised by the two-particle amplitudes Sab.
These are functions of the relativistic invz,riant Mandelstam variable s = (Pa + Pb)2
or, equivalently, of u = (Pa - Pb)2. Sab has a branch root singularity in the variable s
at the threshold s = (rna + rnb)2. By crossing symmetry, an analogous branch point
350
also appears at the threshold of the u-channel, namely at s = (rna - rnb)2. Those
are the only branch cuts of the S-matrix, due to the elastic nature of the scattering
processes. The other possible singularities of the scattering amplitudes Sab are simple
and higher-order poles in the interval (rna - rnb)2 < s < (rna +rnb)2 which are related to
the bound state structure. An important simplification in the analysis of the analytic
structure of the S-matrix comes from the parameterization of the external momenta in
terms of the rapidity variable (), i.e. P~ = rna cosh ()a, p! = rna sinh ()a' The mapping
S(()ab) = m~ + m~ + 2mamb cosh ()ab, where ()ab = ()a - ()b transforms the amplitudes Sab
into meromorphic functions Sab(()ab), which satisfy the equations
(1)
(2)
expressing the unitarity and the crossing symmetry of the theory, respectively. The
simple poles of Sab( ()) with positive residues are related to bound state propagation in
the s-channel. Since the bootstrap principle gives the possibility to consider the bound
states on the same footing as the asymptotic states, the amplitudes Sab are related each
other by the functional equations Si/( ()) = Sij( ()+iuj/) Sik( ()-iufk) , (u~b == 1r -u~b)' For
the IMMF, the expression of the elastic S-matrix that satisfies all the above constraints
can be cast in the form
Sab(()) = II (Jc.(())Y" (3)
c.EAab
.
where !c.(()) == ::::ti:~:::~ The sets of numbers Aab and their multiplicities Pc., spec-
ifying the amplitudes (3), can be found, for instance, in [1]. The functions !c.(()) have
two poles located at the crossing symmetrical positions () = i1r0: and () = i1r(1 - 0:). In
addition to simple poles, the S-matrix of the IMMF presents higher-order poles due to
multi-scattering processes (for their detailed interpretation see [20, 21, 22]). The odd
order poles correspond to bound state poles while those of even order do not. Their
appearance is an unavoidable consequence of the iterative application of the bootstrap
equations.
3. Correlation Functions and Form Factors
There is a very efficient way to compute correlation functions in integrable models.
Once the spectrum and the S-matrix are known, this is provided by the spectral rep-
resentation methods, i.e. by the possibility to express the correlation functions as an
infinite series over multi-particle intermediate states
(4)
Basic quantities of this approach are the Form Factors (FF), i.e. the matrix elements
of the local operators on the asymptotic states defined as
(5)
351
Since the spectral representations are based only on the completeness of the asymptotic
states, they are general expressions for any QFT. However, for integrable models, they
become quite effective because the exact computation of the form factors reduces to
finding a solution of a finite set of functional equations. There are other advantages in
using the spectral representation method. In fact, since it deals with physical quanti-
ties, there is no need of renormalization and the coupling constant dependance is taken
into account at all orders by a closed expression of the Form Factors. Secondly, as we
mentioned in the introduction, the above representation (4) present a very fast rate of
convergence for all values of the scaling variable (mr). This is of course expected for
large values of (mr) which are dominated by the lowest massive state but the surprising
and pleasant result is that in many cases the series is saturated by the lowest terms
also for small values of (mr). This may be interpreted as a threshold suppression phe-
nomenon discussed in [7] but we should stress that the same fast convergent behaviour
is also observed in massless theories [13]. Finally, let us mention that the two-point
correlation functions (4) have the form of a Grand-Canonical Partition Function of a
one-dimensional fictitious gas (with coordinate position (Ji) 2(mr) = E}V=Q zN ZN(mr)
although with a coordinate-dependent activity zi(mir, (Ji) = 2~ e-m;rcosh,8; . This ob-
servation is extremely useful to recover ultraviolet data of the theory, as the anomalous
dimensions of the fields, in terms of massive quantities [8, 9].
Let us discuss now the main properties of the Form Factors for 2-D Integrable Massive
Field Theories which are the crucial quantities entering the spectral representation of
correlation functions. For a scalar operator (x), relativistic invariance requires that
its form factors depend only on the rapidity differences Oi - OJ. The elasticity of the
scattering processes, together with the crossing symmetry and the completeness relation
of the asymptotical states, permit to derive the following monodromy equations satisfied
by the FF
Apart from these monodromy properties, the FF are expected to have poles induced
by the singularities of the S-matrix, among them the simple poles associated to the
kinematic singularities with residue given by [6]
and those due to the bound state singularities with residue [5, 6]
where Oe = (OaUbe + Obrt:.e)/U~b' In general, the FF may also present simple poles which
do not fall into the two classes above. In addition, they may also have higher-order
poles induced by the multi-scattering processes and indeed, their analytic structure may
352
be quite complicated. Their analysis has been pursued in [1] and we refer the reader to
this paper for a detailed account on this aspect.
4. Form Factors of the Magnetization Operator a(x) in the Ising Model
in a Magnetic Field
In the Form Factor approach to integrable theories, the two-particle FF play a
particularly important role, both from a theoretical and from a practical point of view.
From a theoretical point of view, they provide the initial conditions which are needed for
solving the recursive equations relative to the residue conditions (7) and (8). Moreover,
they also encode all the basic properties that the matrix elements with higher number
of particles inherit by factorization, namely the asymptotic behaviour and the analytic
structure. In other words, once the two-particle FF of the considered operator have been
given, the determination of all other matrix elements is simply reduced to solve a well-
defined mathematical problem. From a practical point of view, the truncation of the
spectral series at the two-particle level usually provides a very accurate approximation
of the correlation function, which goes even further than the crossover region. This
section is mainly devoted to the discussion of the basic features of the two-particle FF
in the IMMF.
The FF Fa~( B) of a local field <1>( x) must be a meromorphic function of the rapidity
difference defined in the strip 1mB E (0,271-). Its monodromy properties are dictated
by the general equations (6), once specialized to the case n = 2 Thus, denoting by
Fabin(B) a solution of those equations free of poles and zeros in the physical strip and
also requiring asymptotic power boundness in momenta, we conclude that Fj,(B) must
be equal to Fa'b in (B) times a rational function of cosh B. The poles of this extra function
are determined by the singularity structure of the scattering amplitude Sab( B).
According to the analysis of the paper [1], a convenient parametrization of the two-
particle FF is given by
F4>(B) = Q~b(B) Fmin(B) (9)
ab Dab( B) ab ,
where Dab(8) and Q~b(8) are polynomials in cosh 8: the former is fixed by the singularity
structure of Sab( B) while the latter carries the whole information about the operator
<1>(x).
To further constrain the order of the polynomial Q~b(B), a bound extremely useful
has been derived in [1]. Briefly stated, the argument consists in looking at the large
energy limit of the FF and relating it to the conformal properties of the corresponding
operator <1>( x). Denoted by .0.4> the conformal weight of the operator <1>( x) and by Y4>
the real quantity defined by
lim Fa4> a (B ... , Bn) rv ey~IOd

10,1 ..... 00 }, , n ll
we have the bound

(10)
Taking into account the degree of the polynomial Dab(B) in the denominator of the
two-particle FF (9), it is easy to translate this inequality into an upper bound on the
degree of the polynomial Q~b(B).
353
Let us now apply the aforementioned considerations to the case of the IMMF, con-
sidering in order the different terms entering (9). For the minimal FF, we have
(11)
where
_ {laoodtCOSh(o:-t)t. 2 (i7r-8)t}
Gex (8) - exp 2 - h sm . (12)
t
t.
cos 2" smh t 27r
For large values of the rapidity Gex (8) rv exp(181/2) ,181 -+ 00, independent of the
index 0:. Concerning the pole terms in eq. (9) this could be written as
Dab( 8) = II (Pex( 8) )ia (Pl - ex ( 8) )ia , (13)

exEAab
where
iex =n +1 , jQ =n if PQ = 2n + 1 (14)
lex =n , JQ = n , if Pex = 2n ,
and we have introduced the notation
PQ(8) == cos 7r0: - cosh 8 . (15)
2 cos 2 1r ex
2
Both F:J,in( 8) and Dab( 8) have been normalized to 1 in 8 = i7r.

Finally, let us turn our attention to the determination of the polynomials Q~b( 8) for
the specific operator we are interested in, namely the magnetization field 0"( x). In view
of the relation
8(x) = 27r h(2 - 26. q ) O"(x) , (16)
which relates 0"( x) to the trace 8( x) of the stress-energy tensor, we can concentrate
our attention on the latter operator. This is particularly useful. In fact, the conserva-
tion equation ol"TI"V = 0 implies the following relations among the FF of the different
components of the energy-momentum tensor
~: F~, ... ,aJ8I, ... , 8n) ; (17)

p-
p+ F~, ... ,aJ8l, ... ,8n)' (18)
where x = Xo Xl are the light-cone coordinates and p == 2::'=1 P;'. The requirement
that all the components of the energy-momentum tensor must exhibit the same singu-
larity structure, leads to conclude that the FF of 8( x) must contain a factor p+ P-.
However, the case n = 2 is special because, if the two particles have equal masses, the
mismatch of the singularities disappears in eqs. (18) and no factorisation takes place 2
From this analysis, we conclude that for our model we can write
(19)
2These properties of the matrix elements of the trace of the stress-energy tensor may
be explained quite naturally by considering special limit of non-integrable quantum field
theories [24].
354
0.14
0.12
0.1
0.08
0.06
0.04
0.02
0
0 5 10 15 20 25 30
Figure 1. Correlation function < 0" (x)<r (0) > < versus lattice space distances. The points of the graph represent the
numerical data, as extracted from ref. [23], while the continuum curve is the theoretical estimate with only the
fIrst three Form Factors.
0.14
0.12
0.1
0.08
0.06
0.04
0.02
0
0 5 10 15 20 25 30
Figure 2. Correlation function < a(x)<r(O) > < versus lattice space distances. The points on the graph represent
the numerical data, as extracted from ref. [23], while the continuum curve is the theoretical estimate obtained with
the first eight terms of the spectral series.
355
where
Nab
Pab (()) == L a~b cosh k () (20)

k=O
The degree Nab of the polynomials Pab (()) can be severely constrained by using eq. (10).
Additional conditions for these polynomials are provided by the normalization of the
operator 8(x), that for the diagonal elements F!, reads
(21)
Using all the information above, in [1] we have computed the first FF of the trace
8(x). Their explicit expressions may be found in our original paper. Rather than
providing them here, we would like instead to present the final result of our computation,
namely the graphs relative to the two-point function G( x) =< 0"( x )0"(0) > compared
with a high-precision numerical determination of this quantity obtained by Montecarlo
simulations [23]. In Fig. 1 we have only included the first three terms of G( x) (those
relative to the form factors of the one-particle states AI, A 2 , A3)' As shown in this
figure, they can reproduce correctly the behaviour of the correlation function on the
whole infrared and crossover regions. A slight deviation of the theoretical curve from
the numerical values is only observed for the first points of the ultraviolet region, where
a better approximation can be obtained by including more terms in the form factor
series. This is shown in Figure 2, where five more contributions (those relative to form
factors up to state AIA3) have been added to the series. As far as the Form Factor
approach is concerned, this comparison constitutes one of the most remarkable success
obtained by means of these techniques.
Let us also mention that analogous calculations have been recently performed for
other two statistical models closely related to the Ising model in a magnetic field.
These are the thermal deformations of the Tricritical Ising model and the three-state
Potts model [25]. For these two models, it would be highly interesting to compare the
theoretical predictions of [25] with their numerical or experimental determination.
References
[1] G. Delfino and G. Mussardo, Nucl. Phys. B 455 (1995), 724.
[2] A.A. Belavin, A.M. Polyakov and A.B. Zamolodchikov, Nucl. Phys. B
241 (1984), 333.
[3] A.B. Zamolodchikov, m Advanced Studies in Pure Mathematics 19

(1989), 641; Int. J. Mod. Phys. A 3 (1988), 743.
[4] V.A. Fateev and A.B. Zamolodchikov, Int. Joum. of Mod. Phys. A 5
(1990), 1025.
[5] B. Berg, M. Karowski, P. Weisz, Phys. Rev. D 19 (1979), 2477; M.

Karowski, P. Weisz, Nucl. Phys. B 139 (1978), 445; M. Karowski, Phys.
Rep. 49 (1979), 229;
356
[6] F.A. Smirnov, Form Factors in Completely Integrable Models of Quan-
tum Field Theory (World Scientific) 1992, and references therein.
[7] J.1. Cardy and G. Mussardo, Nucl. Phys. B 410 [FS] (1993),451.
[8] V.P. Yurov and Al.B. Zamolodchikov, Int. J. Mod. Phys. A 6 (1991),
3419.
[9] J.1. Cardy and G. Mussardo, Nucl. Phys. B 340 (1990), 387.
[10] Al.B. Zamolodchikov, Nucl. Phys. B 348 (1991), 619.
[11] G. Delfino and G. Mussardo, Phys. Lett. B 324 (1994), 40; J. Balog,
Phys. Lett. B 300 (1993), 145.
[12] A. Fring, G. Mussardo and P. Simonetti, Nucl. Phys. B 393 (1993), 413;
G. Mussardo and P. Simonetti, Int. J. Mod. Phys. A 9 (1994), 3307.
[13] G. Delfino, G. Mussardo and P. Simonetti, Phys. Rev. D 51 (1995),

6620.
[14] T.T. Wu, B.M. McCoy, C.A. Tracy and E. Barouch, Phys. Rev. B 13
(1978), 316.
[15] B.M. McCoy and T.T. Wu, The Two-Dimensional Ising Model (Harvard
University Press, Cambridge, 1973).
[16] B.M. McCoy, C.A. Tracy and T.T. Wu, Jour. Math. Phys. 18 (1977),
1058.
[17] M. Sato, T. Miwa and M. Jimbo, Publ. RIMS, Kyoto Univ. 14 (1978),
223.
[18] J. Palmer and C.A. Tracy, Adv. in Applied Math. 2 (1981) 329;
[19] O. Babelon and D. Bernard, Phys. Lett. B 288 (1992), 113.
[20] S. Coleman and H.J. Thun, Commun. Math. Phys. 61 (1978), 31; C.J.
Goebel, Prog. Theor. Phys. Supplement 86 (1986), 261.
[21] H.W. Braden, E. Corrigan, P.E. Dorey and R. Sasaki, Nucl. Phys. B
338 (1990), 689; Nucl. Phys. B 356 (1991), 469.
[22] P. Christe and G. Mussardo, Nucl. Phys. B 330 (1990), 465; Int. J.
Mod. Phys. A 5 (1990), 4581.
[23] P.G. Lauwers and V. Rittenberg, Numerical Estimates of the Spin-Spin

Correlation Function for the Critical 2-D Ising Model in a Magnetic
Field, Bonn-HE-89-11.
357
[24] G. Delfino, G. Mussardo and P. Simonetti, Non-integrable Quan-
tum Field Theories as Perturbations of Certain Integrable Models,
ISAS/EP /96/23.
[25] C. Acerbi, G. Mussardo and A. Valleriani, Form Factors and Correlation

Functions of the Stress-Energy Tensor in Massive Deformation of the
Minimal Models (Enh (8) (Enh/(Enh, ISAS/EP /95-161.
358
BOUNDARY S MATRIX FOR THE BOUNDARY SINE-GORDON MODEL
FROM FRACTIONAL-SPIN INTEGRALS OF MOTION
Luca Mezincescu 1 ,2 and Rafael I. Nepomechie 2
1 Physics Department, Bonn University
Nussallee 12, D-53115 Bonn, Germany
2 Physics Department, University of Miami

Coral Gables, FL 33124 USA (permanent address)
Abstract: We construct integrals of motion(IM) for the sine-Gordon model with bound-
ary at the free Fermion point UP = 47r) which have a simple form. These 1M determine
the boundary S matrix. We also provide an Ansatz for the fractional-spin 1M away
from the free Fermion point ((J2 i- 47r).
INTRODUCTION
Much is known about trigonometric solutions of the Yang-Baxter equations, pri-

marily because the algebraic structure underlying these equations has been identified
and elucidated - namely, quantum groups [1]. In contrast, much less is known about
corresponding solutions of the boundary Yang-Baxter equation [2], because the rele-
vant algebraic structure (some sort of "boundary quantum group") has not yet been
formulated.
Motivated in part by such considerations, we have undertaken together with A.
B. Zamolodchikov a project [3] to construct fractional-spin integrals of motion [4] -
[6] of the sine-Gordon model with boundary [7], which should generate precisely such
an algebraic structure. A further motivation for this work is to determine the exact
relation between the parameters of the action and the parameters of the boundary S
matrix given in [7].
Some preliminary results were reported in [8]. Specifically, certain integrals of
motion (1M) for the sine-Gordon model with boundary at the free Fermion point UP =
47r) were constructed. Even though these 1M correctly determine the boundary S
matrix, they are rather complicated (nonlocal in time).
We report here some further progress. We construct new 1M for the sine-Gordon
359
model with boundary at the free Fermion point, which also correctly determine the
boundary S matrix, but which have a much simpler form. The two sets of 1M have
distinct "flow" properties: as the "boundary magnetic field" h is varied from h = 0 to
h --+ 00, the new 1M flow to the "topological charge", while the old 1M flow to a set of
charges which are also conserved for "fixed" boundary conditions. Moreover, we provide
an Ansatz for the fractional-spin 1M away from the free Fermion point UP =I- 471").
FRACTIONAL-SPIN INTEGRALS OF MOTION

FOR THE BULK SINE-GORDON THEORY
Let us recall some of the basic properties of the bulk sine-Gordon field theory. (For
more details, see [9] and references therein.) The Lagrangian density is given by
(1)
where (x, t) is a real scalar field, mo has dimensions of mass, and (3 is a dimensionless
coupling constant. For 471" :s: (32 :s: 871", the particle spectrum consists only of solitons
and antisolitons, with equal masses m, and with "topological charge"
T = - (31
271" -00
00
f)
dx~(x,
uX
t) (2)
equal to +1 or -1, respectively. The particles' two-momenta PI" are conveniently pa-
rameterized in terms of their rapidities ():
Po = m cosh () , PI = m sinh () . (3)
This theory is integrable; i.e., it possesses an infinite number of integer-spin in-

tegrals of motion. Consequently, the scattering of solitons is factorizable. For the
case of two particles with rapidities ()I and ()2, the two-particle S matrix S( ()) (where
() = ()I - ()2) is defined by
(4)
where A (()) t are "particle-creation operators". The S matrix obeys the Yang-Baxter
equation, and is given by
a(()) o 0
b(()) c(()) o
S(O) ~ p(O) ( ~ c(()) b(()) oo ) . (5)
o 0 a(())
where
a(()) sin[>'(71" - u)] (6)

b(()) sin(>.u) (7)
c(()) sin( >.71") (8)
and
871"
u = -i(), >'=(p-1. (9)
360
The unitarization factor p(O) can be found in Ref [9].
In addition to having an infinite number of integrals of motion (1M) of integer
spin, the sine-Gordon model also has [6] fractional-spin integrals of motion Q and
Q. In the ultraviolet limit mo -+ 0, the corresponding densities are given by certain
exponentials of
</J(x, t) ~ (<p(x,t) + [Xoo dy Ot<p(y,t)) , (10)
(/J(x, t) ~ (<p(x,t) - [Xoo dy Ot<p(y,t)) , (11)
respectively. These 1M obey the so-called q-deformed twisted affine sl(2) algebra (slq(2))
with zero center, where q = e- i7r (A-1). Moreover, these 1M have nontrivial commutation
relations with the particle-creation operators, which in terms of the matrix notation
t
A ( 0) =
(A+( O)t )
A_(O)t
. b
are gIVen y
QA(O)t qU3A( O)t Q + q-t eA8 17'fA( O)t , (12)

QA(O)t q'fU3 A(O)tQ + qte-A817'fA(O)t, (13)
T A(O)t A( 0) tT + 173 A (0) t . (14)
Associativity of the products in QA a .(Od t A a2 (02)tI0) and invariance of the vacuum
QIO) = 0 (where Q = Q, Q, T); or, equivalently,
[S,~(Q)l =0 (15)
(where S is the "S operator" and ~ is the comultiplication) leads to the S matrix
(5), up to the unitarization factor. This is a realization in quantum field theory of the
general program [1] of linearizing the problem of finding trigonometric solutions of the
Yang-Baxter equations.
SINE-GORDON WITH BOUNDARY
Following Ghoshal and Zamolodchikov [7), we consider the sine-Gordon field theory
on the half-line x E (-00,0], with action
S= foo dt {fO
-00 -00
dx 0 + M cos ~(<P - <Po)i _ } ,
2 X_O
(16)
where M and <Po are parameters which specify the boundary conditions. There is
compelling evidence that this choice of boundary conditions preserves the integrability
of the bulk theory, and hence, the scattering of the solitons remains factorizable. The
boundary S matrix R(O), which is defined by [7]
(17)
where B is the boundary operator, describes the scattering of solitons and antisolitons
off the boundary. The boundary S matrix must satisfy the boundary Yang-Baxter
equation [2], and therefore has the form [7],[10]
R(O) = (0) ( c?sh()'O + iO -~k+ sinh(2)'O) ) (18)

r -~k_ sinh(2)'O) cosh()'O - i~) .
361
Evidently, this matrix depends on the three parameters ~, k+, k_; however, by a suitable
gauge transformation, the off-diagonal elements can be made to coincide. Thus, the
boundary S matrix in fact depends on only two parameters, as does the action (16).
The relation between these two sets of parameters is not known for general values of {3.
As mentioned in the Introduction, we would like to construct fractional-spin in-
tegrals of motion for the sine-Gordon field theory with boundary. As a warm-up, we
first consider the special case {32 = 47r which corresponds to q = 1 (i.e., undeformed
algebra).
Free Fermion point ({32 = 47r)
As is well known [11], in the bulk, the sine-Gordon field theory (1) is equivalent to
a free massive Dirac field theory for {32 = 47r, with corresponding Lagrangian densityl
C[F = ~I]/ ~ IlJ - iml]/llJ. (19)
The Dirac spinor field has two complex components, which we denote by IlJ = ( ~: ),
IlJt = ( ~= ). Evidently, the Lagrangian has a U(l) symmetry; 1/J+, 1;)+, have charge
+1 and 1/J-, 1;)- have charge -l.
The sine-Gordon field theory with boundary (16) for (32 = 47r has been argued
[8],[12] to be equivalent to
(20)
where Lboundary is given by
Lboundary = ~ [ei(+'Pl1;)+ 1/J+ + e- i(+'Pl1;)_1/J_ + aa

- h (e i'P1;)+ + e- i'P1;)_ + ei 1/J+ + e- i 1/J_) a] Ix=o ' (21)
where a(t) is a Fermionic boundary degree of freedom, and h, </J, r.p are parameters
which specify the boundary conditions. This action is a generalization of the one for the
off-critical Ising field theory (free massive Majorana field) with a boundary magnetic
field [7]. The action given in [12] differs slightly from (21); specifically, it involves
two Fermionic boundary degrees of freedom instead of one. However, the boundary
conditions on the spinor field which follow by the variational principle from the two
actions are identical (apart from minor differences in notation).
We make use of the mode expansions (following the conventions of [7])
(22)
(23)
where p. x = Pot + PIX, and w = eirr / 4 ,w* = e- irr / 4 Canonical quantization implies that
the only nonvanishing anticommutation relations for the modes are
(24)
lOur conventions are ryOO = -1 = _ry11; ~ = \jIIT,O; ,a = -ia2; , I = aI, where ai are the Pauli
matrices.
362
We regard A(f1)t as creation operators.
Computing R(B) according to the definition (17), we obtain the result (18), with
..\ = 1 and
(25)
ei(rI>-cp) - ft '
me'fi(Hcp)
hJl - 'f!l cos( - cp) + 'f1; ,

2
(26)
r( B) =
)1- ~~ cos( - cp) + r;::
(27)
cosh( - cp) - i sinh B - ft cosh 2 B .
The relation between the parameters of the Bosonic and Fermionic actions is discussed
in [12].
Define the charges
0 .
Q =] -00 dx 1/J1/J, (28)
(29)
where the dot (.) denotes differentiation with respect to time. (Note Q~ = -Q_, Q~ =
-Q _ , Tt = T.) Evidently, these charges are of the form J~oo dx J O, where JO is the time
component of a two-component current Jp, which is conserved (op,JP, = 0). Nevertheless,
in general these charges (unlike J~oo dx JO) are not separately conserved, since J1lx=o "#
O.
For "fixed" boundary conditions (h -+ 00), there are three linearly independent
combinations of the above charges (28), (29) which are conserved: T, Qfixed, Q}ixed,
where
(30)
For "free'" boundary conditions (h = 0), there are only two linearly independent com-
binations which are conserved: Qfree, Qjree, where
(31)
A principal new result of this paper is that for general values of h, the charges
and Qt are conserved. Note that as h varies from 0 to 00, Q interpolates between Qfree
and T:
Q ~ Qfree (33)
h2e- 2iCP T .
~
m
There are also higher-derivative generalizations of these charges which are conserved.
363
In a previous report of our work [81, we exhibited a different set of charges Qold,
Q:ld which interpolate instead between Qfree and Qfixed:
~ Q' free
----. (34)
-e -4i'PQ'tfixed.
The old charges have the drawback of being "nonlocal" in time; indeed, they are con-
structed in terms of the quantities C("p+) and C(1f+), where
C("p) = \ a (ei'P"p + e-i'P"pt) . (35)

1 + h28t
Moreover, we remark that it is possible (by working in a suitable way with time deriva-
tives of C("p)) to construct generalizations of Qold which are not "nonlocal" in time,
but which interpolate between higher-derivative generalizations of Q free and Q fixed.
(The higher-derivative generalizations of Q free and Q fixed are discussed in more de-
tail in the Appendix.) Indeed, it is by trying to close the algebra generated by these
higher-derivative generalizations of Qold that we were led to the new charge (32).
The charges Q (given by Eq. (32)) and Qt determine the correct boundary S
matrix. Indeed, after using (22), (23) to express the charges in terms of the modes 2 ,
it becomes a simple matter to determine the commutation relations of Q and Qt with
A(O)t, which we encode in the 2 x 2 matrices Z(O) and Z'(O):
[Q,A(O)t] Z(O) A(O)t, (36)

[Qt,A(O)t] = Z'(O) A(O)t. (37)
Finally, the conditions
Z(O) R(O) R(O) Z( -0), (38)

Z'(O) R(O) R(O) Z'( -0), (39)
(which follow from relating QA(O)tIO)B to A( -O)tIO)B in two different ways; here, Q =
Q, Qt) lead directly to the boundary S matrix given by (18), (25), (26), up to the
unitarization factor.
Away from the free Fermion point UP =I- 471")

We would like to explicitly construct generalizations of the integrals of motion Q
(32) and Qt for the sine-Gordon model with boundary for (32 =I- 471".
As a preliminary step in this direction, we make the following observation . Define
the quantities Q and Q' by
q-"2Q+
1 _ lk+
+ q"2-Q_ e-
- 2i-
ie (l_ q- T )
(40)
k_ k_ q - q-l
k_ eie (qT - 1 )
q-"2Q_
1 _
+ q"2 -Q+
1
+ 2i- ----1 . (41)
k+ k+ q- q
Assuming that Q, Q, and T have commutation relations with A(O)t given by (12) -
(14), then relations analogous to (38), (39) lead to the boundary S matrix R(O) given
2We assume that we can make the replacement J~oo dx -t ~ J~ dx in Eqs. (28), (29).
364
by (18). Note that for q -+ 1, the charges Qand QI reduce to (32) and its Hermitian
conjugate, respectively.
It remains to construct Qand QI explicitly in terms of the local sine-Gordon field
<J>(x, t), to demonstrate their conservation, and to identify their algebra.
ACKNOWLEDGMENTS
The work presented here was initiated by A. B. Zamolodchikov, and represents a

preliminary account of a joint ongoing investigation with him. We are grateful to him
for his invaluable advice, kind hospitality, and for giving us the permission to present
some of our joint results here. We also thank O. Alvarez, D. Bernard, A. LeClair, and
V. Rittenberg for discussions; and we acknowledge the hospitality at Bonn University,
Rutgers University, and the Benasque Center for Physics, where part of this work was
performed. This work was supported in part by DFG Ri 317/13-1, and by the National
Science Foundation under Grant PHY-9507829.
APPENDIX: ALGEBRA AT THE FREE FERMION POINT
We briefly discuss the algebra of the sine-Gordon integrals of motion for the special
case (32 = 411'. The basic strategy to identify a basis is to use the Fermionic formulation
of the model to construct integrals of motion in coordinate space; then use (22), (23)
to express the integrals of motion in terms of Fourier modes A(8); and finally, to form
linear combinations which (in Fourier space) are particularly simple.
For the bulk theory (19), such an analysis has been performed by LeClair [13J. The
result is as follows (our notation differs slightly from his):
Q~ = r~ocod8 e- nO A+(8)tA_(8) }
n odd, (42)
Q:;; = J~co d8 e- nO A_( 8)t A+( 8) ,
Tn i: d() e- nO [A+(())tA+(()) - A_(())tA_(())]
1:
n even, (43)
Pn d8 e- nO [A+(8)tA+(8) + A_(8)tA_(8)] n odd, (44)
(Q~t = Q:;; ,T~ = Tn, P~ = Pn) and the algebra is given by

[Tn' Q!] (45)
[Q~ ,Q;;;] (46)
and all other commutators vanish. 3

For the theory with boundary (20), (21), evidently the integrals of motion are
linear combinations of (42) - (44), and the algebra of these 1M is a subalgebra of (45),
(46). For "fixed" boundary conditions (h -+ (0), we obtain the integrals of motion
Q
1
+ ei2(1)-<p)Q -1 , (47)
Q
n
+ ei2(1)-<p)Qn-2 + Q2-n + ei2(1)-<p)Q-n' n odd;:::: 3, (48)
3The relation of the charges (28), (29) to (42), (43) is given (up to constants) by Q ex: Q~l'
-
Q ex: Q 1 ,T ex: To
365
To To, (49)
Tn Tn + T-n' n even 2: 2, (50)
Fn Pn + P-n , n odd 2: 1, (51)
which obey the algebra
' + Qm-n
2 (Qm+n ') , (52)
Tn+m + Tn+m- 4 + T2+n-m + T2- n+m
+2cos(2(<fJ - cp)) (Tn-m + Tn+m- 2) . (53)
For "free" boundary conditions (h = 0), we obtain the integrals of motion
Qt + e- 2i (Hip)Q:::1 ~ ie-i(Hip)To , (54)

Q n+ + Q+-n+2 + e- 2i (Hip) (Q-n-2-n
+ Q- )
_ie-i(Hip) (Tn- 1 + T-n+d , n odd 2: 3, (55)
T2 - T-2 - 2ie i(Hip) (Qt + Q!1) + 2ie- i(Hip) (Q~ + Q:::1) , (56)
Tn - T_ n + T-nH - Tn- 4
-2ie i(Hip) (Q+n-1 + Q+n-3 + Q+1-n + Q+3-n )
+ 2ze -i(Hip) (Q-n-1 + Q-n-3 + Q-1-n + Q-3-n'

) n even 2: 4, (57)
n odd 2: 1, (58)
together with Q~. Notice that the expressions (55), (57) for Qn and Tn are invariant
under n -+ 2 - nand n -+ 4 - n, respectively; and T~ = Tn. The algebra is given by
2 (Qm+n + 2Qm+n-2 + Qm+n-4

+Qm-n + 2Qm-n+2 + Qm-n H ) , (59)
't 't 't
-2 ( Qm+n + 2Qm+n-2 + Qm+n-4
't
+Qm-n + 2Qm-n+2
't
+ Qm-nH
't)
, (60)
Tn+m + Tn- m+2 , (61)
and all other commutators vanish.
For general values of h, the integrals of motion are certain "deformations" of those
for free boundary conditions (i.e., the expressions reduce to (54) - (58) for h = 0). For
example,
' 1_- Q' 1free + Z. h 2 e -2iipT,0,

Q (62)
m
where Qt ee is given by (54). The complete set of generators and their algebra will be
presented elsewhere.
REFERENCES
[lJ P.P. Kulish and N.Yu. Reshetikhin, J. Sov. Math. 23 (1983) 2435; V.G. Drinfel'd,
J. Sov. Math. 41 (1988) 898; M. Jimbo, Int. J. Mod. Phys. A4 (1989) 3759.
366
[2] LV. Cherednik, Theor. Math. Phys. 61 (1984) 977.
[3] L. Mezincescu, R.L Nepomechie and A.B. Zamolodchikov, in preparation.
[4] M. Luscher, Nucl. Phys. B135 (1978) 1.
[5] A.B. Zamolodchikov, "Fractional-spin integrals of motion in perturbed conformal

field theory," in Fields, Strings and Quantum Gravity, eds. H. Guo, Z. Qiu and
H. Tye, (Gordon and Breach, 1989).
[6] D. Bernard and A. LeClair, Commun. Math. Phys. 142 (1991) 99.
[7] S. Ghoshal and A.B. Zamolodchikov, Int. J. Mod. Phys. A9 (1994) 3841.
[8] L. Mezincescu and R.L Nepomechie, "Integrals of motion for the sine-Gordon
model with boundary at the free Fermion point," in Unified Symmetry in the
Small and in the Large 2, eds. B.N. Kursunoglu, S. Mintz and A. Perlmutter,
(Plenum, 1995).
[9] A.B. Zamolodchikov and Al.B. Zamolodchikov, Ann. Phys. 120 (1979) 253;
[10] H.J. de Vega and A. Gonzalez-Ruiz, J. Phys. A26 (1993) L519.
[11] S. Coleman, Phys. Rev. D11 (1975) 2088.
[12] M. Ameduri, R. Konik and A. LeClair, Phys. Lett. B354 (1995) 376.
[13] A. LeClair, Nucl. Phys. B415 (1994) 734.
367
SECTION IX - EPILOGUE
SONOLUMINESCENCE AND THE HEIMLICH EFFECT
AlanChodos
Center for Theoretical Physics

Yale University
P.O. Box 208120
New Haven, Connecticut 06520
The phenomenon of sonoluminescence (SL), originally observed some sixty years ago, 1
has recently become the focus of renewed interest,2-9 particularly with the discovery that one
can trap a single bubble and induce it to exhibit SL stably over a large number of acoustical
cycles.2
In a typical experimental situation, a bubble of gas (usually just air, possibly doped with
a noble gas) in a liquid (usually just water) is made to expand and then to contract violently
under the influence of an applied acoustic field. During this motion,6 the bubble emits a very
sharp pulse of light, after which it expands again and oscillates about its equilibrium radius,
until stability is regained. The process then reoccurs in the next cycle.
Some features to note4 are: (1) the pulse is extremely narrow, probably not more than 10
ps and possibly much less, whereas the acoustic frequency is on the order of 30 kHz and the
relevant scale for bubble collapse is perhaps 10 ns; (2) the photon energies are typically at
least of order 1 eV, and may be greater. (The water is opaque to photons with energies
beyond 1 eV); (3) The intensity of SL varies considerably depending on a number of
parameters (intensity of the sound field, temperature of the water, composition of the bubble,
etc.) but under optimal conditions pulses with several million photons are routinely
achieved;5,8 (4) SL represents a remarkable concentration of energy: the acoustic energy per
atom is typically eleven or twelve orders of magnitUde less than the energies of the individual
photons that are emitted.
On the theoretical side, the problem of understanding SL resolves itself into three coupled
components: the dynamics of the bubble driven by the sound wave; the dynamics of the gas
within the bubble; and the radiative process that produces the photons. The first and second
of these2. 7 would appear to be classical problems governed by well-known equations (which
is not to say that everything has been understood), whereas the third is undoubtedly a
quantum phenomenon whose origin is still very much in dispute.
371
In this work we shall explore a version of the provocative suggestion lO put forward by
Schwinger: the mechanism responsible for the radiation in SL is a dynamic generalization of
the Casimir effect. It has been known since Casimir's original work in 1948 11 that the zero-
point energy of quantum fields can be modified by the presence of boundaries, and that these
modifications generate observable effects. For example, in Casimir's original work, the
quantum fluctuations of the electromagnetic field in the presence of a pair of uncharged,
parallel, perfectly conducting plates were shown to give rise to an attractive force between the
plates.
Schwinger invites us to consider the situation in which the boundary is that between a
dielectric medium (the water) and, essentially, the vacuum (the gas inside the bubble). Here,
of course, the geometry is spherical, which already makes the computation more difficult,
and an additional but clearly crucial complication is that the location of the boundary depends
on time. Under these circumstances, one may expect that instead of (or perhaps in addition
to) the static Casimir force, one will observe the radiation of the quanta of the electromagnetic
field, which will constitute the sonoluminescence pulse.
The challenge is to present a calculation of this effect that is simple enough to be tractable,
and yet captures the essential physics. Schwinger struggled with this problem over the
course of seven telegraphic communications 10,12 to the Proceedings of the National
Academy. Eberlein13 has done a computation based on an analogy with the Unruh effect,14
in which the adiabatic approximation is used to permit quantization in a bubble of fixed
radius, and then the photon emission amplitude is computed to lowest order in the velocity of
the bubble. Her results illustrate an inherent problem with the dynamic Casimir effect:
Casimir energies tend to be quite small, and in order to reproduce the observed pulse
intensity one must invoke bubble velocities that are rather higher than seem physically
reasonable, even perhaps exceeding that of light Milton 15 has done a careful investigation of
the static situation and has pointed out similar difficulties. Related computations, in a
simplified dynamical model, have been performed by Sassaroli, Srivastava and Widom. 16
In this work we consider a model that neglects the volume effect, i.e. the fact that light
propagates with different velocities in the bubble and in the medium, and concentrates on the
surface effect, i.e. the fact that a boundary condition must be imposed on the photon field at
the surface of the bubble. Furthermore, we adopt the attitude that, at least for the purposes of
an initial investigation, it should not matter precisely which boundary condition is chosen.
Thus the choice of a particular form for the term that enforces the boundary condition will be
motivated more by computational convenience than by an appeal to underlying physical
principles, leaving open the future possibility of finding a more realistic choice.
Thus we consider the action
(1)
Here Rtv =i E ~vpa Fpa ' F~v =a~Av - avA~ and f(x) is a dimensionless function that
represents the coupling of the photon to the boundary of the bubble, located at r = R(t),
where R(t) is an externally prescribed function. We note that
372
FJ.!-v A1v = 2d~ [e ~vpa AvopAa] so the second term of S may be written
!Jd4x o~f(x)e ~vpaAvOpAa. IT we choose f(x) =fo O(R(t) - r), we shall obtain a strictly
local coupling of the photon to the surface. Classically S describes a system obeying the
equations of motion
o~F~v + ()J1f F~v =0 , (2)
which, for our choice of f, is solved by a freely propagating electromagnetic wave subject to
the boundary condition
(3)
on the surface, where n~ is the four-dimensional normal, n~ =..j(R. r)2

I-a
At the quantum level, the simplest thing to do is to treat the ~ term in S as a
perturbation, that is, to assume that fo is a small parameter. Whether this is physically
reasonable can only be checked a posteriori, by fitting fo to the data and seeing if it is indeed
small.
In this note, we shall follow this approach, and compute some relevant amplitudes in
lowest order perturbation theory. Diagrammatically, the basic vertex is
(4)
where the cross represents the action of the external source f(x). It is actually more
convenient to work in momentum space, and therefore we require the Fourier transform
f(p) =_1_
(2'11;)4
Jd4 x eipx f(x) . (5)
=
When f(x) fo O(R(t) - r), we fmd (assuming that R(t) =R(-t), which is true in Eberlein's
model but which does not really fit the data)
f(p~ = ~!- [!Img(p, po)] (6)
where g(p, PO) =[0000 dt eipR(t) eiPol; and p =IPI.

As a simple example where we shall be able to evaluate g explicitly, we may consider
=
R(t) Ro, Itl > T .
R(t) =Ro + v (itl - 1') ,ltI < T . (7)
373
This then yields
1m = v [COS (P~ + Po1) + cos (PRo - Po1) + 2cos (PRo - pv:!?j (8)
g P Po O+pv) Po (PO-pv) p2vLpO
Among the quantities of physical interest that we may compute is the average number of
photons that are produced due to the action of the source:
(N) = 1t3 I d 4 P 9(Po) 9(P~ - p2) If(P, Po)12 (Po - p2)2 . (9)
The challenge is to see whether the left-hand side can be of order 1()6, even though f6, which
appears on the right-hand side, is a small number. Unfortunately, one sees that since ffalls
off as p~ for large PO, the right-hand side actually diverges. To ameliorate this, one should
include some or all of the following effects:
(a) instead of a sharp boundary function, 9(R(t)-r) one should presumably smooth the
boundary over a small distance 11. This will result in a modification of f:
f(p, Po) ~ f(p, PO) e-pa (10)
which will not directly solve the large Po divergence problem for (N), but will insure that the
p integral remains finite in this and other expressions;
(b) As Casimir pointed out in his original work,ll the boundary that is represented by
f(x) essentially disappears for high frequencies, because high energy photons do not interact
with the boundary as a whole, but only, if at all, with the individual constituents. Thus one
should insert a high frequency cutoff on physical grounds;
(c) The experimental data are cut off by the fact that water absorbs all photons with
energies greater than about I eV. Thus the experimentally measured (N) is only for photons
with energies less than I eV, and hence one should integrate only up to Po - 2 eV or so (the
Po in the integral represents the energy of a pair). Of course, one does not want the water to
absorb large numbers of high energy photons: these would have observable effects that are
not seen.
In addition to the total number of produced photons, it is also possible to compute other
quantities of interest in the same approximation, such as the spectrum of produced particles.
It seems wise, however, to concentrate first on (N), since too large an fo will vitiate the
approximation. This and related computations are currently in progress.17
We turn next to the consequences of a point we remarked upon earlier: independent of
whatever the radiation mechanism turns out to be, SL can be viewed as the conversion of
acoustic energy, which is distributed diffusely throughout the liquid, into a burst of
electromagnetic energy which is concentrated spatially into a small region, perhaps a few
microns, at the center of the bubble, and which invests a typical photon with an energy of at
least an e V, eleven or twelve orders of magnitude more than the acoustic energy of an atom in
the liquid. Furthermore, this process is reasonably effIcient, in that the energy carried off by
374
the photons is comparable to the energy required to combat the shear viscosity of the fluid. 3
These features suggest that SL might, with a lot of technological development, be a
candidate for a mechanism of particle acceleration. Mter all, the one indispensable property
of an accelerator is not just the total energy, but rather the ability to transfer energy that is
macroscopically generated to individual microscopic particles. Whatever the mechanism, this
is an ability that SL is observed to possess.
To pursue this idea within the framework of the kind of model that we have been
discussing, we must add to the action the terms, familiar from QED, describing the electron
and its interaction with the electromagnetic field:
(11)
Then, combining the QED interaction with the photon-bubble interaction, we shall have
diagrams describing the acceleration of the electron. To lowest order in the electron-photon
interaction, we have
~,
(12)
p------~--~~-------r
in which the bubble creates a pair and one of the photons then is absorbed by the electron.
The probability of starting with an electron of momentum p (say, at rest) and measuring an
electron of momentum p', together with a photon of momentum q, is given by
(13)
where the bar indicates an average over the spin of the electron and the polarization of the
photon, and K is a kinematical factor,
\
K= (~ - 1) p . q)2 + (p\ . q)2) _ [q. (p\ _ p)]2 . (14)
To obtain the total amplitude for acceleration, one should then integrate this expression over
q.
We refer to this process as the "Heimlich effect", because, at a rather different length
scale, it produces the same result as the well-known Heimlich maneuver: 18 a bubble is
squeezed, and a particle pops out
We note that the SLC accelerator at SLAC imparts an energy of about 15 eV/micron to the
electrons. Since the size of the sonoluminescent region is of order a micron, and since the
energies are in the 1-10 eV range, we appear to have the potential to achieve similar results.
Of course, SLAC is 2 miles long whereas so far SL has been commed to one micron-sized
bubble at a time. It is premature to speculate on how difficult it might be to improve this
situation.
375
In summary, we have presented a phenomenological model that we believe captures the
essence of Schwinger's suggestion about the mechanism behind the SL radiation process.
To confront this model with the data, one must achieve reliable numerical estimates first of
(N) and then of other quantities such as the photon spectrum, not only for the simple R(t)
chosen here but for more realistic choices as well. Furthermore, one should be prepared to
extend the analysis beyond lowest order perturbation theory, and to take advantage of the
quadratic nature of the photon-bubble interaction in order perhaps to obtain non-perturbative
results that will more stringently test the model. 17
We have also suggested, regardless of the validity of our model, that the phenomenon of
sonoluminescence may provide the tentative first step toward a new method of particle
acceleration that may be of increasing relevance as the currently dominant species of
accelerator faces inevitable decline and perhaps even extinction in the twenty-first century.
References
1. H. Frenkel and H. Schultes, Z. Phys. Chern. B27:421 (1934); see also N. Marinesco
and J.-J. Trillat, C.R. Acad. Sci. Paris, 196:858 (1933).
2. D.F. Gaitan, L.A Crum, C.C. Church and R.A. Roy, J. Acoust. Soc. Am. 91:3166
(1992).
3. R. Hiller, S.L. Putterman, and B.P. Barber, Phys. Rev. Lett. 69:1183 (1992).
4. L.A Crum and R.A Roy, Science 226:233 (1994).
5. R. Hiller, K. Weninger, S.J. Putterman and B.P. Barber, Science 226:248 (1994).
6. B.P. Barber and S.J. Putterman, Phys. Rev. Lett. 69:3839 (1992).
7. C.C. Wu and P.H. Roberts, Phys. Rev. Lett. 70:3424 (1993).
8. B.P. Barber, C.C. Wu, R. Lofstedt, P.H. Roberts and S.J. Putterman, Phys. Rev.
Lett. 72:1380 (1994).
9. L. Frommhold and AA Atchley, Phys. Rev. Lett. 73:2883 (1994).
10: J. Schwinger, PNAS 89:4091 (1992).
11. H.B.G. Casimir, Proc. K. Ned. Akad. Wet. 51:783 (1948).
12. J. Schwinger, PNAS 89:11118 (1992); PNAS 90:958, 2105, 4505, 7285 (1993);
PNAS 91:6473 (1994).
13. C. Eberlein, quant-ph 9506023, quant-ph 9506024.
14. W.G. Unruh, Phys. Rev. 014:870 (1976).
15. K. Milton, hep-th 9510091.
16. E. Sassaroli, Y. Srivastava and A Widom, Nucl. Phys. B (Proc. Suppl.) 33C:209
(1993); Phys. Rev. A, 50:1027 (1994).
17. A Chodos and A Nyffeler, work in progress.
18. See, e.g. Stedman's Medical Dictionary, 25th ed., pages 688 and 918.
376
BOUNDARY CONDITIONS FOR SCHWINGER-DYSON
EQUATIONS AND VACUUM SELECTION
Zachary Guralnik
Princeton University
Princeton, NJ 08544
INTRODUCTION
When attempting to solve a field theory or matrix model by using Schwinger-Dyson

equations, one must address the problem that these equations do not possess a unique
solution. This problem came to our attention when trying to numerically solve the
equations for certain lattice field theories using a technique known as Source Galerkin
[1], which is discussed elsewhere in these proceedings. It was found that to make the
numerical method stable, one had to find some way of selecting the right boundary
conditions. This led us to the more general question of how the boundary conditions are
related to the phase diagram of a field theory or a matrix model. We shall summarize
our work on this relation in this paper. Most of the details will appear elsewhere.
The naive resolution of the problem of how to select the boundary conditions is
to simply pick the solution which corresponds to the path integral over real fields.
However, within certain phases of many theories, it can be shown that the path integral
solution is actually not the physical one. Furthermore in some matrix models the
integral over real eigenvalues is not even convergent because of negative couplings.
This forces the consideration of "exotic" solutions of the Schwinger-Dyson equations
which have integral representations involving sums of integrals of the fields over various
inequivalent complex contours. For theories with a local order parameter, symmetry
breaking solutions are generated naturally by choosing a symmetry breaking set of
contours. In the conventional approach to obtaining the broken phase, the real contour
is chosen but a small symmetry breaking term is added to the action. This term is
removed only after taking a thermodynamic limit, in which the number of degrees of
freedom becomes infinite. In fact one can also obtain the broken phase by choosing a
symmetry breaking boundary condition (contour) and then taking the thermodynamic
limit directly. This is a simple example showing that the exotic solutions are not
necessarily unphysical. We conjecture that this is true even for theories with a nonlocal
order parameter, although this has yet to be demonstrated.
The difficulty in choosing the correct boundary conditiOIls lies in the fact that there
are so many of them. Furthermore since the Schwinger-Dyson equations satisfied by
377
the partition function are linear, there naively appears to be a continuum of mixed
phases, which does not make physical sense. Most of this problem is resolved by taking
the thermodynamic limit. In this limit the solutions associated with many boundary
conditions coalesce, while other boundary conditions do not lead to solutions with
a thermodynamic limit. This often leaves a countable number of discrete solutions,
although sometimes a continuum of distinct solutions survives, yielding something
analogous to the theta vacua in Yang Mills theories. The set of solutions which survives
in the thermodynamic limit may include both the physical vacua, and false vacua with
a complex free energy. It can be shown that the set of boundary conditions for which a
thermodynamic limit exists varies along paths in the space of coupling constants. This
set may vary smoothly or it may be discontinuous at certain points along the path.
We have found it easiest to study the behavior of this set in the context of one matrix
models, while field theories have proven less tractable. However we suspect that our
conclusions are quite general. One important conclusion is that the set of boundary
conditions with a thermodynamic limit changes discontinuously as one crosses a phase
boundary.
BOUNDARY CONDITIONS IN ZERO DIMENSION
To illustrate the multiple solutions of the Schwinger-Dyson equations we begin

with the simple example of a zero dimensional theory with the polynomial action
S(</J) = (1/n)gn</J n. The generating functional of disconnected Green's functions, Z(J),
satisfies the Schwinger-Dyson equation
(1)
The order of this equation is determined by the highest order term in the polynomial
action. If the highest order term is </Jk then there is an k - 1 parameter class of
solutions. One of these parameters is just the overall normalization of Z, so there
is a k - 2 parameter class of solutions with distinct Green's functions. An integral
representation of the solutions is
(2)
It is easy to show that this satisfies the Schwinger-Dyson equation provided that
e-S(q,)+Jq,1 = 0 (3)
ar
This is true if ReS(</J) ---+ +00 asymptotically on the contour r. For the polynomial
action this condition becomes Re gk</Jk ---+ +00. Therefore there are k wedge shaped
domains in the complex plane in which the contour can run off to infinity. There
are k - 1 independent contours satisfying (3), which is consistent with the order of
the Schwinger-Dyson equation. In general it is the behavior of the action for large
fields which controls the solution set, even for many degrees of freedom. The condition
(3) can be used to construct the solution set even when the action is non-polynomial
and the order of the Schwinger-Dyson equations is unclear. For example, the action
378
s = f3 cos J (one plaquette QED) can be shown by this method to yield a two parameter
class of solutions for Z. A basis set of solutions is given by the contours
r1 = [-ioo, +ioo] (4)
and
r2 = [-ioo,O] + [0, 21l'] + [21l', 21l' + ioo] (5)
The difference between these two solutions is the usual solution in which the contour
runs from 0 to 21l'. Note that in general the exotic solutions for a gauge theory cor-
respond to a complexification of the gauge group. In this simple case it is possible to
verify the counting of solutions by coupling sources J and J to the loop variables eit/>
and e-it/>. The Schwinger-Dyson equation is then,
(6)
which when combined with the constraint,
(7)
yields a two parameter class of solutions for Z.
BOUNDARY CONDITIONS IN THE GENERAL CASE
Let us consider the solution set for a lattice field theory. For theories in which
the large field behavior of the action is dominated by a local term, such as gkJ k,
the construction of the solution set is a simple generalization of the zero dimensional
construction. One simply chooses one of the zero dimensional contours for each field at
each lattice site. An arbitrary solution is obtained by summing solutions with a definite
set of contours. The solution set is then somewhat reduced by imposing the lattice
symmetries, but is still very large. In an N x N matrix model, the lattice site label is
replaced by an eigenvalue label, and, since the interaction between eigenvalues is only
logarithmic, the highest order term in the potential determines the allowable contours
for each eigenvalue. For theories in which the large field behavior of the action is
not dominated by a local term, the construction of the solution set is somewhat more
complicated. This appears to be the situation in general for lattice theories with a
non-local order parameter.
For a generic action with a finite number of degrees of freedom, the number of in-
dependent solutions of the Schwinger-Dyson equations is exactly equal to the number
of classical solutions, including the complex solutions. The exception to this rule oc-
curs when the action has flat directions, or extrema with vanishing second derivative,
in which case every term in the perturbative expansion about the classical solution
diverges. For example in zero dimensions, the potential V = gJ4 has only one classical
solution, but there are three independent solutions of the Dyson Schwinger equations.
Assuming that we are not considering such an exceptional case, the Borel resummation
of the perturbation series about any classical solution can be shown to yield one or
379
more exact solutions of the Schwinger-Dyson equations with some exotic integral rep-
resentation. One can see to which contours these solutions must correspond by taking
the weak coupling limit. The contours, assumed to be of constant phase, must avoid
all the classical solutions of equal or lower action than the one whose perturbative
expansion is being Borel resummed. There are often many choices of such contours
which correspond to the various choices for avoiding positive real singularities in the
Borel variable. Thus the complete set of solutions of the classical equations yields an
over complete set of solutions of the Schwinger-Dyson equations. An arbitrary solution
is obtained by taking linear combinations of these solutions.
Thus, away from the thermodynamic limit, there are a continuum of phases, re-
sembling theta vacua. To complete the classification of these phases, one needs to give
a rule for identifying solutions in the same phase at different values of the coupling
constants. More precisely, one needs some set of first order differential equations in
the coupling constants, which the partition function in a given phase should satisfy. A
natural choice is given by the Schwinger Action Principle, which may be stated as,
(8)
or, for the example of the zero dimensional polynomial action, as
(9)
For a solution satisfying the action principle, it is possible to hold the associated
measure, or sum over contours, fixed locally as one moves about the space of coupling
constants. Note however that the action principle does not allow one to fix the measure
globally if there is a branch cut in any of the couplings. For instance in the zero
dimensional l4 theory, if one rotates the phase of the coupling constant by 271" in
accordance with the action principle, the contour of integration rotates by 71" /2 yielding
an inequivalent solution. One could have chosen another set of first order differential
equations instead of the action principle. However, the Schwinger Action operators
which annihilate the partition function commute with the Schwinger-Dyson operators
which annihilate the partition function. If in the thermodynamic limit a continuum of
solutions to the Schwinger-Dyson equations reduces to a discrete set, then in general
the action principle turns out to be automatically satisfied. For instance, in a matrix
model the action principle is satisfied order by order in the 1/N expansion without
having to be imposed by hand.
VACUUM SELECTION
Having classified the solutions far from the thermodynamic limit, one needs some
way of reducing the solution set. It is tempting to invoke such requirements as reality
and positivity. Reality however is not much of a constraint, because if all the couplings
of the theory are real, one can always take linear combinations of exotic solutions
such that all Green's functions become real. Thus this is not so strong a constraint
and also throws away false vacuum solutions which may be of physical interest. One
380
might also be tempted to invoke positivity, since only the integral over real fields is
obviously positive. However it is not in general necessary to throw out members of the
solution set by hand and it is dangerous to invoke reality and positivity before taking a
thermodynamic limit. The thermodynamic limit alone does most of the job of reducing
the solution set. This can seen very explicitly in matrix models.
Consider a model of an N x N hermitian matrix M,
z= J dM e-NtrV{M} (10)
M may be written as UAUt where A is diagonal, and then, if the integral over U is
performed, one obtains
z = JII dAn ~ 2[A]e-Nl:n V{An} (11)

n
where ~[A] is the Vandermonde determinant
~[A] = II (An - Am) (12)

n<m
One can separately choose a contour for each eigenvalue by the condition Re V(An) ~
+00 for large An. The boundary condition problem is then most easily understood if
one solves the model by the method of orthogonal polynomials. The reader is referred
elsewhere for details [2]. To summarize the method very briefly, polynomials Pn(>') =
>.n + ... are defined such that,
(13)
where the integral over>. is over some permissible complex contour or sum of complex
contours. The Pn have the property that
(14)
The partition function and the Green's functions may be written in terms of the coef-
ficients Rn and Sn. These coefficients satisfy a set of recursion relations, and a smooth
thermodynamic (large N planar) limit is obtained if they may be written, in this limit,
in terms of functions of x = n/N on the interval [0,1]. These functions are easily ob-
tained from the recursion relations. One can then look for the possible contours which
give such limiting functions, using the fact the Rn and Sn coefficients are related to the
Green's functions of the zero dimensional theory with the action NV(>') as N ~ 00.
In this way it is easy to see that the solutions associated with a very large number
of boundary conditions either coalesce or have no thermodynamic limit as N ~ 00
At certain critical values of the couplings, such as the points about which a double
scaling limit may be taken, the functions of x which one obtains in the planar limit
become complex within the interval [0,1]. This means that as one crosses these critical
domains, the boundary conditions which give real solutions no longer have a thermo-
dynamic limit. A change in the boundary conditions which lead to a thermodynamic
limit is a general phenomenon which occurs as one crosses a phase boundary.
381
As an interesting aside, we note that there are some very exotic boundary conditions
in matrix models which do have a thermodynamic limit. It has been shown that, in
the planar limit, matrix models with degenerate minima have a continuum of solutions
analogous to theta vacua [3]. One can find integral measures which account for some
of these solutions [4]. However there are too many such solutions to be accounted
for by integral measures which are the same for every eigenvalue. Furthermore these
solutions persist even when a small term is added to the action lifting the degeneracy.
Yet if one removes the degeneracy but restricts oneself to solutions with the same
integral measure at each eigenvalue, then the associated 5(0) and R(O) can only take
certain discrete values. This can be seen from the fact that 5(0) and R(O) are the
one-point and two-point connected Green's functions of the zero dimensional theory
with action NV('\) as N -+ 00. A discrete set of values for 5(0) and R(O) would
not permit a continuum of theta vacua. Consequently the theta vacua can only arise
from boundary conditions for which the measure is not a simple product of the same
measure at each eigenvalue. One could instead take sums of solutions with staggered
boundary conditions, in which the contours are different for different eigenvalues.
There are several phenomena which occur in the thermodynamic limit which appear
to be disparate. One is the collapse of the space of boundary conditions, due to the fact
that many boundary conditions do not lead to thermodynamic limit. The other is the
appearance of new nonanalyticity and phase boundaries in the coupling constants due
to the accumulation of Lee-Yang zeroes [5]. In fact these are not really disparate. We
will give a heuristic argument below. The collapse of the space of boundary conditions
means that as one varies the couplings, one must also vary the boundary conditions
(contours), since the set of boundary conditions with a thermodynamic limit depends
on where one is in the space of coupling constants. It is then possible that by following
a closed path in coupling constant space, one does not return to the boundary condition
with which one started. This would mean that the thermodynamic limit has introduced
a nonanalyticity in the coupling constants, which in turn requires an accumulation
of Lee-Yang zeroes. Note that there are simple zero dimensional analogues of this.
Consider the polynomial action 5 = L:~=l 9nn. The solutions are analytic in in all
but the highest coupling constant gk. If one follows a closed path in the complex
plane of any of the lower coupling constants, one does not have to change the contour
to maintain the convergence, and thus there is only a single Riemann sheet in the
lower coupling constants. If however one tunes gk to zero, then the space of boundary
conditions shrinks. If one now rotates the phase of gk-l by a large amount, then one
must also rotate the contour of integration, whereas the contour could be held fixed
for non-zero gk. Green's functions become multiply sheeted in gk-l
The conventional tool for studying the phase structure of theories with a local order
parameter is the effective potential. It is interesting to note the consequences of our
analysis of of boundary conditions on the form of the effective potential. In fact, far
from the thermodynamic limit there are many effective potentials. Consider again the
zero dimensional example. The function ( J) = (a j aJ) In Z (J), satisfies a nonlinear
differential Schwinger-Dyson equation of order k - 2. The effective potential f( <1 is
defined by the relation J = (dfjd. Any effective potential can only correspond to
a discrete number of solutions <1>( J) out of the full k - 2 parameter class. In fact in
zero dimensions there is no non-analyticity in J so there is only one solution associated
with each effective potential. Note that order by order in a loop expansion, (J) is
multiply sheeted and the effective potential appears to possess multiple minima which
correspond to different boundary conditions. This is a spurious feature of the loop
expansion in zero dimensions. Actually <1>( J) has an infinite tower of poles in J, rather
382
than branch cuts. In a thermodynamic limit these poles can coalesce, in which case
several boundary conditions become described by the same exact effective potential.
Many effective potentials associated with different boundary conditions coalesce in the
thermodynamic limit, while effective potentials associated with other boundary condi-
tions do not have a well defined thermodynamic limit.
CONCLUSIONS
The phase structure of field theories and matrix models appears to have a very
natural interpretation in terms of the boundary conditions of Schwinger-Dyson equa-
tions. There are several interesting open questions. One question concerns the relation
between the various boundary conditions and the phases of a theory with a nonlocal
order parameter. As yet we have not been able to construct such a relation. Another
interesting question is whether by using exotic contours one could formulate a practical
method to study theta vacua on the lattice.
ACKNOWLEDGEMENTS
This work developed from discussions with S. Garcia, G. Guralnik, and S. Hahn.
A detailed paper on the subject is in preparation.
REFERENCES
[1] S. Garcia, G. Guralnik and J. Lawson, Phys. Lett. B322 (1994) 119-124.
[2] See for example: P. Ginsparg and G. Moore, Boulder TASI Lectures (1992) 277-
470
[3] R. Brower, N. Deo, S. Jain and C.-I Tan, Nucl. Phys. B405 (1993) 166-190.
[4] F. David, Phys. Lett. B302 (1993) 403-410.
[5] C. N. Yang and T. D. Lee, Phys. Rev. 87 (1952) 404-410.
383
NUMERICAL QUANTUM FIELD THEORY
USING THE SOURCE GALERKIN METHOD*
G.S. Guralnik
Physics Department
Brown University
Providence RI 02912
INTRODUCTION
This talk describes a program, still in its early stages, to develop a new numerical
approach that we call the Source Galerkin method [1]. This method has little in
common with the Monte Carlo methods which have been successful in demonstrating
that numerical techniques can be an important tool for obtaining results from quantum
field theories. Source Galerkin appears to have strengths which will allow its use for
problems (such as scattering predictions) that are not readily accessible to conventional
methods with the currently available computers. It is, unlike Monte Carlo, equally
suited for studying fermionic or bosonic problems. Further, at least in lower dimensions,
a continuum formulation of Source Galerkin is straightforward. While we have made
considerable progress in understanding the issues in defining and applying these ideas,
we can not yet been able to complete enough calculations to evaluate the chances
that Source Galerkin could actually replace Monte Carlo techniques in any of their
traditional applications.
The basic idea of Source Galerkin is to formulate a field theory in the presence
of external sources and then to construct approximate solutions to the linear source
functional differential equations for the vacuum-vacuum amplitude Z of the theory.
These solutions are constructed by (enlightened) guessing. It is possible to utilize
* Supported in part by DOE Grant DE-FG09-91-ER-40588-Task D
385
internal (charge, statistical) and external (spatial) symmetries expected of the exact
answer in the construction of the approximate solutions. This possibility, is one of
the great advantages of our method. In contrast, it is difficult or impossible to insert
knowledge obtained from analytic techniques into the construction of solutions using
Monte Carlo techniques.
A straightforward way to make a Galerkin guess is to pick a linear combination

of a subset of a complete set of functions of the source variables appropriate for the
system being studied. This choice has the advantage of having a clear iteration path
defined by extending the number of functions from the complete set used in successive
approximations. This guess will not satisfy the original equations. The deviation from
the exact solution will be characterized by the numerical coefficients multiplying each
member of the complete set contained in the guess. The idea of Galerkin is to fix these
coefficients by adjusting them so that the guess is true as an appropriately defined
weighted average over some region of source space. The optimal definition of average
is part of the problem of defining a rapidly convergent approximation scheme. Such
a scheme allows the reduction of the error between the approximate solution and the
exact answer as the number of iterations is increased. Galerkin is a weighted residue
procedure. These methods guarantee convergence in a mean. That is, the integral of
the square of the local error vanishes in an appropriate limit. In this context, it should
be noted that the more familiar variational Hamiltonian approximation techniques are
in this same general class of approximation. However, the Source Galerkin methods
do not have a direct correspondence to such methods as they have been used in physics.
EXAMPLE OF GALERKIN METHOD:
Before we examine a field theory, we outline the application of Galerkin to a generic

linear differential equation
Lj=O
in some domain D (Xl, x2 - X N) with boundary conditions on a surface S (Xl - X N ).

Guess a solution:
M
f* = ~o (Xl - XN) +L aNi (Xl - XN)'
i=l
386
Assume the '-P are linearly independent and that they satisfy the appropriate
boundary conditions but are otherwise unrestricted.
Now make the definition
R(al,az - aM;Xl - SN) == L(1*)

M
= L ('-Po) + L a j L ( '-P j)
j=l
Next, pick M linearly independent test functions Pi (Xl - XN)' We do not confine our
considerations to the usual Galerkin choice of picking Pi = '-Pi. The coefficients al - aM
are fixed by requiring that
(R,pi) = 0 i = 1,2 .. M.
We define this generalized dot product as follows:
To make sure that this is sensible we choose d so it dampens integrals. Options

that we have used in field theory problems include restricting d to be composed of 8
functions, so as to confine the integral to a finite volume or choosing it to be of the
form P- L,.J x,A'Jxj .
After choosing d and performing the above integrations, we obtain M linear equa-
tions in M unknowns. If these are independent, we can solve for all of the unknowns. If
not, we must change our choice of test functions or increase their number. Under very
general circumstances, these approximations converge (weakly) to the exact answer as
M ---+ 00. Obviously the convergence is best if the '-Pi form a complete set.
SCALAR FIELD THEORY
Now we can look at the example of Euclidean quartic field theory coupled to a
source J (x) for the scalar field [1-3]. This source will be the variable used in our
Galerkin solution and hence the name Source Galerkin. The action is given by
which leads to the field equations
387
With the usual definition Z = (010), it follows that
The Galerkin method appears to be a self consistent method on the continuum

because of the regularization properties that can be directly built into the test functions
and guess solutions. However, for the very simple considerations of this paper we will
confine our examples to the lattice. On the lattice we have
The it sum is over nearest neighbors and the lattice spacing a is absorbed. Conse-
quently, the lattice equations for this model are
Note, this is a very large set of coupled differential equations, since there is one
equation for each point on the lattice. For the transition to a lattice to be a sensible
analogue to the continuum equations, we must define finite lattices at their boundary
points. In what follows, periodic boundary conditions will always be implicit.
In order to solve this set of equations, we make an initial Galerkin guess. An

uninspired (but, nevertheless, interesting) way to proceed is to just try a Taylor series
expansion. This has the advantage of being an expansion in a complete set of functions.
Z (JI - IN) = L Gmln2-nN JflJ;2 - Jit

ffll,n2- R N
Substituting this expansion into the differential equations above results in com-
plicated sets of recursion relations particularly for a large lattice. Realistically, the
appropriate application of boundary conditions and the size of the sets involved makes
direct solution impossible.
388
One alternative way to proceed is to truncate the series at some maximal power
of the sources. Truncated series are definitely not solutions to our equations. It is
reasonable to hope that for small sources that the contributions from higher terms in
the expansion decrease rapidly. In any event, we can minimize the error from truncation
if we determine the coefficients of the Taylor expansion by a Galerkin procedure in the
source co-ordinates. This means using the definition of Galerkin given above with the
identification J dX1 - dx N --+ J dJ1 ... dJN In low source dimensions this is using
excessive force for the outcome, but as the dimension increases the use of this method
to guarantee consistency is not trivial.
To solve the truncated equations we use lattice symmetry (translation invariance).

If, for example, we examine a 3 site lattice in 1 dimension keeping terms to quartic
polynomial order in lattice invariant polynomials the test solution is
Here, we have introduce the definitions:
P6(J) = Jf + Ji + Ji
P'f(J) = J1J2 + hJ3 + J3J1
pt(J) = Jt + Ji + Ji
Altogether there are six independent P~ to this order. We Galerkin the results of
inserting this guess into the lattice equations by using
as test functions. By lattice symmetry. all lattice points are equivalent. Consequently,
we need only examine equations for one lattice site. This problem can easily be solved
using a computer algebra program (we use Aljabr which is very robust and powerful for
this type of problem). We can easily compare our method with Monte Carlo Methods
on one dimensional systems without using any substantial amount of computer time.
In the following table we calculate the two point function as we increase the degree of
the polynomial and compare the results with Monte Carlo. We also include the results
of a Shanks interpolation of the Galerkin results. The coupling is relatively small but
389
even for much larger coupling the results remain good, despite the small number of
polynomials used. This reflects the fact that the Galerkin procedure forces the guess
to be reasonable in the average we have defined. The computer time required for the
Galerkin solution is orders of magnitude less than that required for the Monte Carlo
of comparable accuracy.
Table 1. I-dimensional two-point function: 11 site lattice; M = 1, g = 0.5
z-] J4 J6 J8 Shanks MC
0 0.326538 0.360178 0.347223 0.35082 0.35100.0005
1 0.094508 0.117673 0.108121 0.11091 0.111O0.0004
2 0.026768 0.038823 0.033422 0.03509 0.03510.0003
3 0.007454 0.012940 0.010269 0.01114 0.01120.0004

4 0.002075 0.004461 0.003197 0.00363 0.00360.0003
5 0.000694 0.001922 0.001218 0.00147 0.00110.0004
For the purposes of this talk, I have kept the examples very simple. We can calcu-
late much bigger systems than this tiny example even in 4 dimensions. However, the
number of representatives of an invariance class of polynomials on the lattice grows ex-
ponentially. Unless a low order polynomial gives a good approximation (which happens
in the example), simple Taylor series are not a great starting point and we should pick
other expansion functions which more closely approximate the actual answer. While,
we do not usually know what this is, we do know a lot about the structure of an exact
answer from symmetry and the spectral representations of quantum field theory. Con-
sequently, it is easy to make a better guess.
NON-LINEAR METHODS
Our initial examinations of the Galerkin method concentrated on expansions that

were simple for Z. However, because of the disconnected Greens functions content of
Z it is often much better to deal with In Z. The difficulty with this is that the resulting
equations are non-linear in the Galerkin parameters. For example, from the previous
equations, it is straightforward to show that the approximation for a solution to quartic
scalar field theory of the form
390
yields non-linear equations after being acted on by the source differential equations and
then Galerkin averaged. These can be handled numerically. It is possible to develop
iteration schemes based on the above guess which keep the problems associated with
non-linearity under control.
An effective way to construct leading approximations is to build Z in terms of

spectral functions (propagators). A good combination can be guessed by writing down
the perturbation theory or an effective theory for the action being studied. However,
the only information we use from these structures is the general spectral forms. The
masses and couplings are left arbitrary and set by Galerkin. Another effective and
rapidly convergent guess can be made using polynomials weighted with propagators.
There are a huge number of reasonable choices. In practice, we have found remarkable
stability (per computational time) between very different types of iterations schemes
when used on simple problems.
As a special case of the above non-linear form, we have been examining a spectral
iteration scheme that we would think has much of the structure of an actual solution.
The initial ansatz is the form
Jdwdx~J(W)G2(W'X)J(x)+
logZ =
J ~J(w)J(x)G:Cw,
dw dx dy dz x, y, z)J(y)J(z)
with
and
- - - B - -
G:(p, q, r) = G2(p)G2(q) (p + q)2 + p,/2 G2(r)G2(p + q - r).
We are generating results based on this starting form in 4 dimensions on fairly big
systems. We are finding that this method consumes small amounts of CPU (relatively)
but at the cost of large amounts of memory.
BOUNDARY CONDITIONS
The discussion up to this point has avoided a fundamental issue. One of the most
complicated and interesting problems associated with this type of numerical solution
391
lies in the analysis of the appropriate boundary conditions imposed on the source differ-
ential equations. If we look at the special case of zero space time dimensions (ultralocal
model) there is only one differential equation. It is a third order differential equation so
its solution involves the specification of three a-priori arbitrary constants. In particu-
lar, the normalization of Z, and the one and two field Green's functions are unspecified.
The situation for arbitrary space-time dimension is more complex with a correspond-
ingly greater number of unspecified constants. This situation is not entirely unfamiliar.
In this discussion we have excluded terms involving odd powers of the sources. This
constitutes a boundary condition on the first derivative of Z. We know that, if we
do not exclude odd terms, we can under the correct conditions write solutions that
have spontaneous symmetry breaking. Dealing with the boundary condition on the
second derivative terms is a much more complicated and a very interesting problem
that leads to an enhanced understanding of phase structure and the possibility of so-
lutions to quantum field theory which are extremely singular for small bare couplings.
We analyze this elsewhere [5]. I have avoided direct contact with this problem in the
example detailed here by truncating the Taylor series solution in Z. It is easy to see
that truncation forces the approximation to agree with the smooth 9 = 0 limit of the
theory that corresponds to perturbation theory. Therefore, truncation imposes an im-
plicit boundary condition on the second derivative terms.
FERMIONS
An interesting feature of this scheme is that fermions can be included in a natural

manner. Fermions present difficulty in Monte Carlo approaches for two reasons. The
first is because of the "fermion determinant". The problem occurs because after an
approximate bosonic configuration is generated, the fermionic information contained
in a determinant involving every point of the lattice system must be evaluated. This
evaluation requires a large amount of computer time and it must be done for many
bosonic configurations. There are of course many schemes to decrease the amount of
computation this requires. However, the difficulty is intrinsic to how path integration
forces us to deal with fermions. Source Galerkin avoids this difficulty because it allows
making approximations for fermions without first solving a bosonic problem. I will
outline how this is done in the following. The second problem is the well known
"fermion doubling problem". This difficulty has nothing to do with numerical methods,
but is an artifact of lattice formulation. As mentioned, it is possible that we can avoid
this as well. We have worked several fermionic problems in low dimensions on the
continuum. This has been easy because the divergences we dealt with were trivial. We
are currently working on a "proof" that Galerkin has enough flexibility to regularize
all calculations without being forced to resort to a lattice.
392
The key difficulty in developing a Source Galerkin procedure for fermions is that
the naive inner product definition is not appropriate because of the Grassmann nature
of the sources.
Direct integration of Grassmann polynomials is defined by
J de = 0, J dee= 1.
Fermionic polynomials are decimated if we base the inner product on this type of
integration. Instead we define a similar inner product to that developed for bosons.
For Majorana fermions, we take this to be
Since det f3q = 1, this leads us to the normal Gaussian integration, which in turn gives
us the Wick contraction in 1]. The formulation for Dirac fermions is
To test these definitions, we have studied simple quartic fermion interacting sys-
tems on 4 sites (one dimension) and compared our solutions to the exact result [3].
Conv~rgence is rapidly achieved by the Galerkin averaging.
SUMMARY
Previous published work in developing the Source Galerkin method has been fo-
cussed primarily on free scalars and </J4, in various spacetime dimensions [1-3). Current
work targets robust methods for gauge and fermionic theories. We are also attempting
to improve the convergence rate of our approximations by using the full Schwinger ac-
tion principle (this is somewhat like imposing renormalization group restrictions). We
have tested these ideas on the Thirring and Schwinger models and have shown that
nothing in pri~ciple precludes working with models with many degrees of freedom and
gauge structure. In these simple solvable models convergence to the exact answer is
almost instant. We have looked at leading order in Gross Neveu model and find results
which correspond to the leading order -Jv. As we develop our techniques we anticipate
to be able to answer questions about more realistic theories. At this stage, I am very
optimistic that the Source Galerkin model will be important in making a contribution
to our understanding of the solutions of quantum field theory.
393
Acknowledgement
Results presented here were obtained in collaborations with S. Garda (IBM), Z. Gu-
ralnik (Princeton), S. Hahn (Brown), J. Lawson (ICTP Trieste) and K. Platt (Brown).
REFERENCES
1. "A New Approach to Numerical Quantum Field Theory", (with S. Garcia and
J. Lawson),Phys. Lett. B22, 119 (1994).
2. "New Numerical Method for Fermion Field Theory (with John Lawson),to be
published (1995).
3. "Source Galerkin Calculations in Scalar Field Theory (with John Lawson),to be

published (1995).
4. "Multiple Solutions in Quantum Field Theory (with Santiago Garcia),to be pub-

lished (1995)
5. " Boundary conditions of the Schwinger Dyson equations and multiple solu-
tions of Quantum Field Theory" (with Santiago Garcia and Z. Guralnik), Brown
Preprint (1996).
394
INDEX
Active galactic nuclei, 121 Gravity gradients, 99

ATLAS, 43 GUT, 19,43, 303
Baryon masses, 319 Hadron, 154, 319

Bayesian analysis, 115 Heimlich effect, 371
Bjorken Sum Rule, 329 HERA,126
Black holes, 273 HERMES, 166
Borel Transformations, 329 Higgs Boson, 4, 287
Bose-Einstein condensate, 3, 4 Homestake, 116
BRS, 14 Hubble Parameter, 87
Canonical quantization, 137 Ising Model, 350

CEBAF,332
Chargino, 268 Kilc-Moody Alegbras, 33, 68
Charm, 163, 167 Kamiokande, 113, 116
Chirp mass, 90
CMS,45 LAMPF,332
Color confinement, 13, 14 Lattice Gauge Theory, 13,241
Color helicity, 218 Lee-Yang equation, 382
Color transparency, 162 LEP, 45, 177
Confinement, 3, 194 LHC, 303
Conformal field theory, 57, 63, 245 LHS, 45
Cosmological constant, 31 Light cone, 133, 153, 160, 183, 223
CP,31 LlGO, 73, 79, 95
Local fields, 349
Dark matter, 43, 253, 283, 287 LSND,103
DLCQ, 156, 205, 225
D0 detector, 265 Magnetic charge, 3
Mesons, 199
Electric charge, 3 Microscopic black holes, 273
Electroweak, 67 Monopoles, 3, 6
Exclusive processes, 334 MSSM, 19,32,44,266,326
Form factors, 173,340,351

Neutralino, 268
Freeze out, 291
Neutrino, 121, 122, 124
Neutrino mass, III, 112
Galerkin method, 385
Neutrino oscillations, 103
GALLEX, 115
Neutrino telescope, 127
Gauge theories, 143
Nuclei,l72
Gluinos, 226
Nucleons, 172
Gluon,44
Numerical quantum field theory, 385
Gluonium, 156
Gravitational collapse, 6
Gravitational wave, 73, 95 Pade-Approximants, 329
Gravitino, 39 Parity violating electron scattering, 333
395
PaTtons, 124 Standard model, 19, 31
Photinos, 287 Squark, 27, 266
Strings, 31, 245
QCD, 15, 134, 156, 187, 194,279,323 Supergravity, 58, 61, 324
QED, 13, 194 Superstrings, 52, 323
Quark mass, 65 Supersymmetry, 19,323
Quarkonium, 164, 166 SUSY-Dark matter, 49, 253, 265, 287. 319
Quarks, 13, 133 SUSY-GUTS, 31, 32, 35, 37, 43, 46, 112
Tamm-Dancoff, 192
R-parity, 19
Top quark, 26
Renormalization, 21, 211, 217
Top squark, 269
Running coupling constant, 94
Transverse lattice, 242
SAGE, 115 Unification, 301

Schwinger-Dyson equation, 377 Universality, 301
Sigma model, 60, 61
SINE-GORDON equation, 359, 360 VIRGO, 81
S-Matrix, 359
Sneutrino, 44 Weak Interactions, 331
Solar neutrinos, 115
Solitons, 57 Yang-Mills, 49, 58, 135,206
Sonoluminescence, 371 Yukawa Coupling, 19,20
396

(Behram N. Kursunoglu (Auth.), Behram N. Kursunogl

Hochgeladen von

Dokumentinformationen

Originaltitel

Copyright

Verfügbare Formate

Dieses Dokument teilen

Dokument teilen oder einbetten

Freigabeoptionen

Stufen Sie dieses Dokument als nützlich ein?

Sind diese Inhalte unangemessen?

Copyright:

Verfügbare Formate

(Behram N. Kursunoglu (Auth.), Behram N. Kursunogl

Hochgeladen von

Copyright:

Verfügbare Formate

Neutrino Mass, Dark Matter,

Springer Science+Business Media, LLC

Neutrino mass, dark natter, gravitational waves, monopole

ISBN 978-1-4899-1566-5 ISBN 978-1-4899-1564-1 (eBook)

All rights reserved

Global Foundation Board of Trustees

Unified Symmetry in the Small and in the Large

Unified Symmetry in the Small and in the Large 1

Unified Symmetry in the Small and in the Large 2

Global Energy Demand in Transition: The New Role of Electricity

Economics and Politics of Energy

Neutrino Mass, Dark Matter, Gravitational Waves, Condensation Of

Contributing Co-Sponsors of the Global Foundation

Gas Research Institute, Washington, DC

(NEUTRINO MASS, DARK MATTER, GRAVITATIONAL WAVES,

(24TH IN A SERIES OF CORAL GABLES CONFERENCES ON

JANUARY 25.28, 1996

EDEN Roc RESORT AND SPA

THURSDAY.January 25.1996 (Cotillion Ballroom)

1:30 PM SESSION I: PROLOGUE

3:30 PM Coffee Break

5:00 PM SESSION III: PROGRESS ON SOME NEW AND OLD IDEAS I

7:30 PM Orbis Scientiae adjourns for the day

FRIDAY, .January 26 1995 (Mona Lisa)

10:15 AM Coffee Break

10:30 AM ROUND TABLE DISCUSSION of Gravitational Waves

Noon Lunch Break

1:30 PM SESSION V: NEUTRINO MASSES

3:30 PM Coffee Break

5:30 PM SESSION VI: STRINGS AND FIELD THEORY

7:00 PM Orbis Scientiae adjourns for the day

SATURDAY, January 27,1996 (Mona Lisa)

STAN BRODSKY, SLAC

10: 15 AM Dissertator Presentations continue

1:00 PM SESSION VIII: THE MATTER OF DARK MATTER

3:30PM Coffee Break

3:45 PM SESSION IX: PROGRESS ON SOME NEW AND OLD IDEAS II

6:00 PM Orbis Scientiae adjourns for the day

7:30 PM Conference Banquet - MONA LISA ROOM

SUNDAY. January 28. 1996 (Key Biscayne Room)

10: 15 AM Coffee Break

10:30 AM SESSION XI: EXACTLY SOLVABLE QUANTUM MODELS

SECTION II - PROGRESS ON NEW AND OLD IDEAS - A

SECTION III - GRAVIT ATIONAL WAVES

SECTION V - DIRAC'S LEGACY: LIGHT-CONE QUANTIZATION

SECTION VI - THE MATTER OF DARK MATTER

SECTION VIII - EXACTLY SOLUBLE QUANTUM MODELS

Sonoluminescence and the Heimlich Effect.. ............................................ 371

The process of condensation of the monopoles carrying magnetic charges gn with n

The similarities of the Bose-Einstein condensation of a dense gas of atoms at a

The Bose-Einstein condensation, compared to other phase transitions governed by

A BRIEF OUTLINE OF THE GENERALIZED THEORY OF GRA VITA nON

The Largrangian of general relativity can be expressed in terms of the

ro2q 0 2= c4 /2G , (3)

The mass relation (5) \vhen written in the f0l111 '

Hence, if we consider the special case of Planck particle we obtain

ropo = Yzft , (8)

CREATION OF MATIER VIA MONOPOLE CONDENSATION

where e and 9 represent fundamental units of electric and magnetic charges,

of magnetic charge. where Ig

DISCRETENESS OF CONFINED ELECTRIC CHARGE DISTRIBUTION

If in the fundamental relation (13) we represent the fundamental length r0 in

N ~ 10n , n = 0, 1, 2, .... , (19)

oL-J.-LL....L---LL__...L..J._ _' - - - - - "_ _- ' -_ _-'------' o L-J.~ - L ~ ~~