Beruflich Dokumente
Kultur Dokumente
Edited by
All rights reserved. No part of this book may be reproduced or utilized in any form or by
any means, electronic or mechanical, including photocopying, recording, or by any in-
formation storage and retrieval system, without permission in writing from the Publishers.
(0206)
Seventy years ago when the fraternity of physicists was smaller than the
audience at a weekly physics colloquium in a major university, a J. Willard
Gibbs could, after ten years of thought, summarize his ideas on a subject in
a few monumental papers or in a classic treatise. His competition did not
intimidate him into a muddled correspondence with his favorite editor nor
did it occur to his colleagues that their own progress was retarded by his
leisurely publication schedule.
Today the dramatic phase of a new branch of physics spans less than a
decade and subsides before the definitive treatise is published. Moreover,
modern physics is an extremely interconnected discipline and the busy
practitioner of one of its branches must be kept aware of breakthroughs in
other areas. An expository literature which is clear and timely is needed to
relieve him of the burden of wading through tentative and hastily written
papers scattered in many journals.
To this end we have undertaken the editing of a new series, entitled Docu-
ments on Modern Physics, which will make available selected reviews, lecture
notes, conference proceedings, and important collections of papers in
branches of physics of special current interest. Complete coverage of a field
will not be a primary aim. Rather, we will emphasize readability, speed of
publication, and importance to students and research workers. The books
will appear in low-cost paper-covered editions, as well as in cloth covers. The
scope will be broad, the style informal.
vi EDITORS' PREFACE
From time to time, older branches of physics come alive again, and forgotten
writings acquire relevance to recent developments. We expect to make a
number of such works available by including them in this series along with
new works.
ELLIOTT W. MONTROLL
GEORGE H. VINEYARD
MAURICE LEVY
Contents
1. NONINTERACTING PARTICLES 1
1.1 Introduction
1.2 Sources 3
1.3 Spinless Particles 4
1.4 An Application. Stimulated Emission 10
1.5 Spin-1 Particles. The Photon 12
1.6 Massless Spin-2 Particles. The Graviton 16
1.7 Spin-(1/2) Particles 18
1.8 Multispinor Formulation of Arbitrary Spin 25
General Connection of Spin and Statistics
2. ELECTRODYNAMICS 29
1.1 Introduction
shares with field theory the physical emphasis upon space and time, but it is
not an operator theory. Like S-matrix theory, it is phenomenological in its
emphasis upon the actual physical system, but there is no reference to
analyticity in momentum space. The results of quantum electrodynamics are
reproduced without the irrelevance of divergences, or renormalizations. This
means that we remove the unnecessary field-theory hypothesis that detailed
space-time description is possible, down to distances smaller than those
involved in presently accessible phenomena. Characteristic S-matrix features
such as dispersion relations are deduced, not from abstract mathematical
requirements, but through intuitive space-time arguments. Current algebra
results, which have lent credence to detailed speculations concerning the
inner structures of particles, are recovered by purely phenomenological
procedures. Thus, a unified method is now available, one that is economical
in hypothesis and effective in application.
PARTICLES AND SOURCES 3
1.2 Sources
The concept of particle has been steadily extended in the last few
decades. From the stable electron and proton, to the very long-lived neutron,
to the short-lived n and A, to the highly unstable p and N * it has progressed
to more and more short-lived excitations. Thus, in general, particles must be
created in order to study them, since most of them are unstable. In a general
sense this is also true of high-energy stable particles, which must be created
in that situation by some device, i.e. an accelerator. One can regard all such
creation acts as collisions, in which the necessary properties are transferred
from other particles to the one of interest. There will be a variety of collisions
that can serve to create a particle with specified properties. The other
particles in the collision appear only to supply these attributes. They are,
in an abstract sense, the source of the particle in question. In other words,
the source concept is the abstraction of all possible dynamical mechanisms
whereby the particular particle can be produced. We try to represent this
abstraction of realistic processes numerically. The fact that the collision will
have some degree of space-time localizability indicates the utility of a
numerical function S(x) to measure this aspect. The effectiveness of the
collision in liberating various momenta can be measured by another function
S(p). The quantum-mechanical requirement of complementarity can then
be imposed, as in
which convey the idea that the source liberates or absorbs momentum p in
the respective processes. The square-root signs appear because the proba-
bilities are proportional to the momentum range (dp). The subscript on the
vacuum state indicates the time sense, IO_) is the vacuum state before the
source has operated. The factors of i are included here for later convenience.
There are probability requirements that express the weak source restriction:
(1.5)
p
these two conditions, (1.5) and (1.6), will be equivalent if K(x) is real. We
shall take this to be a general property of any source; a complex source is
regarded as a combination of two real sources which describes some physical
property of that multiplicity. We also notice, from the restriction (1.5) or
(1.6), that I (O+ IO_ )KI must differ from unity by terms of order K 2 • It is
consistent to assume that no terms of order K appear in (O+ IO_)\ as we
shall verify later.
PARTICLES AND SOURCES 5
Now consider a complete situation in which particles are created by K 2 ,
propagate in space and time, and then are detected by K 1 (Fig. 1.1). In this
FIGURE 1.1
--------------------- o_
case the causal situation is well-defined, the detection source K 1 (x) is cer-
tainly localized in time later than the emission source Ki(x). The overall
description is then expressed by the vacuum probability amplitude
p
;; 1 + O(K1 2 ) + O(K2 2 )
+ f dwPf (dx)(dx')iK1(x)eip(x-x')iK2 (x'). (1.7)
We regard K 1 (x) and K 2 (x) as two different manifestations of the same
physical mechanism, that is, they are the values of one general source in
different space-time regions. Therefore the only possible combination that
can occur is the total source
K=K1 +K2 •
This is a fundamental postulate, the principle of the unity of the source,
which embodies the idea of the uniformity of nature. Then the vacuum
amplitude must have the general q~adratic form
Since only the part of L1+(x-x') that is symmetrical in x and x' can con-
tribute to the integral, we are permitted to define
that obeys the boundary condition of outgoing waves in time, that is,
positive (negative) frequencies for positive (negative) time differences. It
should be remarked that we did not begin with these requirements, they
appear as a consequence of our theory. We record here the alternative
representation for L1 + (x- x'),
Ll+(x-x) =
,I (dp)
--4
(2n) p
eip(x-x')
2
+m
2 .
I
-u; s~+o
. (1.13)
The quadratic terms in K 1 and K 2 that appear in (O+ IO_ )Know reproduce
the structure of the product of ( 0 +IO_ )K 1 and ( 0 + IO_ )K2 • We can now
proceed to check the probability requirement, (1.5) or (1.6). From (1.8), we
get
I (O+ Io_ y 1
2
=1- f (dx)(dx')K(x) Re (1/i)L1 +(x-x')K(x') (1.14)
for a weak source. The notation "Re" means "the real part of". Now
Re(1/i)L1+(x-x') =Ref dwPeip(x-x'), (1.15)
which holds for all x-x', since the right-hand side is symmetrical in x and
x'. Thus
f (dx)(dx')K(x) Re (1/i)Ll+(x-x')K(x')
= Ref dwpK(-p)K(p)
FIGURE 1.2
occurs. Since
(O+ IO_ )K f
= exp[ (i/2) (dx)(dx')K(x)L1+(x-x')K(x')]. (1.21)
Conversely, under the condition {1.20) and that the sources are weak, (1.21)
reproduces (1.18), because
J
exp[ (i/2) (dx)(dx')K(x)Lf+(x-x')K(x')]
~ exp[(i/2)}2J (dx)(dx')Ka(x)Lf+(x-x')Ka{x')]
IX
= II exp[(i/2)f (dx)(dx')Ka(x)Lf+(x-x')Ka{x')]
IX
~ 11 [1+(i/2)J(dx)(dx')Ka(x)Lf+(x-x')Ka(x')].
IX
FIGURE 1.3
--------------------- O+
o_
interference effects come into play. Thus consider more generally multipar-
ticle exchanges between sources under the noninteracting cau~al condition
(Fig.1.3)
Then
(O+ Io_ )K = (O+ IO_ )K 1 exp(,if (dx)(dx')K1(x)Lf+(x-x')K2(x')]
x (O+ IO_ )K 2 ' (1.22)
0 0
where, as before, keep in mind that x > x ', and therefore
if (dx)(dx')K1(x)Lf+(x-x')K2 (x')
= f dwiK1(-p)iK2 (p)
= l)K1 P *iK2 P (1.23)
p
KP =.JdwPK(p). (1.24)
We now analyze the vacuum amplitude in terms of multiparticle states
exp(}:iK1P*iK2 P)
p
=II exp(iK1P*iK P)
p
2
(1.26)
Thus, we identify
(1.28)
p
<
(0-1 {n} >K = [ {n} Io_ )K] *.
Inserting ( 1.27) and its complex conjugate, we get
}:(0-1 {n})K({n} IO_)K = 1(0+ jO_)KI 2 IJ exp(IKPI 2 )
w p
and the completeness is verified. Notice how the structure of (O+ IO_ )K
enters in two different ways. On the one hand it gives 1(0+ IO_ )K 12 directly,
and on the other it is used to generate ({n}IO_)K / (O+IO_)K. The con-
sistency of these two procedures serves as a severe test of the formulation.
10 J. SCHWINGER
FIGURE 1.4
~K
FIGURE 1.5
o_
source K 2 , and the final particles be detected by a detection source K1 • The
well-defined causal arrangement is indicated in Fig. 1.5. The total source,
now denoted by (K), is the sum of its constituents
(K) = K 1 +K2 +K.
Now (1.21) gives
(o+ Io_ )<K> = (O+ Io_ )K' +K2(o+ Io_ )K
x exp{if (dx)(dx')[ K 1(x)L1+(x-x')K(x')
+ K(x)L1+(x-x')K2 (x') ]}
= }2(0+ I {n} )K {n} IO_ )K2(0+ IO_ )K
1
(
{n}
{n}
which follow from (1.27) and (1.28), so that the terms linear in Kare
{n} P
.J
+ (O+ I {n} )K 1 nP+ 1 iKP *( {n} + 1P IO_ )K 2 J.
On the other hand, we have the general relation
(1.36)
and
( {n} I {n} + 1P)K =.Jnp+ 1 iKP *. (1.37)
l
µ,v=k,l : <>kl
µ,v=O,O :0
PARTICLES AND SOURCES 13
and
Insistence upon the physical positiveness property has left us, for each
momentum in its rest frame, with three independent sources, which are
transformed into each other by spatial rotations. We are clearly describing
a spin-1 particle. In an arbitrary coordinate frame we have
(1.44)
Here
(1.45)
(1.46)
where ep;. are the polarization vectors associated with momentum pµ, which
may be taken to be real. As a check of dimensionality, setµ= v in (1.44) and
sum overµ. We get
4-1=3,
which gives the right account of multiplicity. With the definition
Jp;. =.Jdwpeµp;.*Jµ(p), (1.47)
we have
(1.48)
(1.49)
and
(1.50)
14 J. SCHWINGER
(1.51)
(1.52)
We are obviously describing the photon. Its masslessness demands that its
source be a divergenceless or conserved vector. Incidentally, if the deri-
vatives in (1.43) are transferred to the sources, the mass-dependent part has
the structure--;. <\Jµ(x)Lf + (x- x')8J 11(x'), which disappears if the source
m
is restricted to be a conserved vector. And then one has a smooth transition
from m=FO tom= 0.
To see that the condition (1.52) is sufficient as well as necessary for the
correct physical interpretation of
(1.55)
(1.56)
PARTICLES AND SOURCES 15
Hence
(p+ iJY'(p+ PY (p- p)µ(p- PY + ~ µ v *
(p+ p)2 + (p- p)2 .tft~ p;.e p;. '
or
~ µ p;.e vp;. * •
+ £..Je (1.57)
l=1,2
4=2+2.
When the representation for gµv' (1.57), is inserted into (1.55), and the
conditions pJµ(p) = pµJµ(-p) = 0 are used (notice that pµlµ(p)#O), we get
l(O+ IO-YI 2 =exp(- }:llp;.l 2)~1, (1.58)
p).
gµ).gv"-<112)gµvg)."__. I [eµaevpeA.ae"p-c112)eµaevae).pe",,J.
IX,P=1,2
Therefore, only two of the eµvaP, IX ,p = 1,2 are independent, and if one
defines
(1.65)
PARTICLES AND SOURCES 17
and
(1.66)
then
(1.67)
= exp(-_LITpA..1 )~1,
2
l(O+IO-Yl 2 (1.69)
p,A=1,2
(O+ IO_)" f 0
= exp[ (i/2) (dx)(dx')rJ(x)y Ll+(x-x')rJ(x') J
in order to describe spin-(1/2) particles? The objection here is that all four
components of rJ~(x) contribute, giving a multiplicity of four rather than the
multiplicity of two appropriate to spin-(1/2). This is overcome by writing
Then
if (dx)(dx')rJ 1(x)y°(m-y(1/i)8)L1 + (x-x')rJi(x')
(1.75)
since y0 is antisymmetrical and the y0 yµ are symmetrical (8' µ = -8µ). But then
if rJ(x) are ordinary numbers, the quadratic form in the exponential function
of (1.72) is zero identically! There are two possibilities to escape this con-
clusion. The first one is to introduce another matrix in G+(x-x'), which is
outside of space-time and therefore does not affect the spin description,
and is antisymmetrical. The simplest choice is
q = (~ -~} (1.78)
The source functions, 17(x), of course are now given an additional multi-
plicity of two. Keeping the previous definition of G+(x-x'), we then have
f
(O+ Io_)" = exp[ (i/2) (dx)(dx')rJ(x)y°qG+(x-x')rJ(x')]' (1.79)
f (dx)(dx')17(x)y q Re (1/i)G+(x-x')rJ(x')~O
0
(1.80)
'·"
= + f (dx)(dx'),L11,,(x')[y G+(x' -x)]n1h(x),
0
'·"
which is simply a relabelling of variables. Just this kind of anticommuting
number is well-known as the elements of a Grassman, or exterior algebra.
The physical properties of the system can now be studied by the
standard procedure. Consider an emission and absorption source
and deduce
J
x exp[i (dx)(dx')17 1(x)y 0 G+(x-x')11 2 (x') ](O+ I0_ )" 2 , (1.81)
(1.85)
where
(1.86)
PARTICLES AND SOURCES 21
and A. is a polarization or spin index. The dimensionality can be checked by
taking the trace of (1.85).
(1.87)
(1.90)
(1.91)
Hence
f
exp[i (dx)(dx')17 1(x)y 0 G+(x-x')17 2(x')]
= exp(~)1'/1p;. *i1'/2p;.)
p).
Since the sources do have simple algebraic properties, each term of this
structure can be factored into two parts, one on the left referring entirely to
the detection source and the other on the right referring entirely to the
emission source. To illustrate the procedure, consider the example of two
states, a and b,
and in general
(1.95)
where TI is some standard multiplication order and ITT is its inverse. The
factor np;. ! = 1 is not necessary, it is included only to show that the results
are universal-BE particles and FD particles are exactly on the same
footing, only the implicit algebraic property of the sources conveying the
nature of the particles.
It is important to check the completeness property of the multiparticle
states, as in
(1.96)
n n
1=l(O+10-)11 l 2 I [IT<11p;.r).]*IT<11p;.)"PA,
n p). p).
(1.98)
PARTICLES AND SOURCES 23
This is a general source property, since transposition is ineffective for BE
sources. We infer from (1.98) that, while the product of two real BE sources
is real, the product of two real FD sources is imaginary. Now
= I1T(17p;.*)"PA.I1(17p;,)"PA.
p). p).
= II <11p). *Y/p;.tpA.,
p).
which is thus valid for both statistics. Now, direct evaluation gives
m-y(1/i)o_.m-yp = 2m2:up;.up;,*y 0 •
).
Therefore,
(1.100)
But
(1.101)
Uq'*,Jdw;K(p) = Kpq''
= LiK1pq'*q'iK2pq'•
pq'
But
. [ ( 1 - q') ] ] 2
q'= [ e' - 4 - n: , (1.105)
only introduces a q' -dependent phase factor in the identification of multi-
particle states. These phase factors will be suppressed when we consider
the completeness requirement (1.29), since each matrix element is combined
with its complex conjugate. So it is required that
But since K(x)qK(x') is real, which follows from (1.98), we have directly
1(0+ 10-YI 2 =exp[ - f (dx)(dx')K(x)q Re (1/i)Lf+(x-x')K(x')]
=exp(- 2)'Kpq'*Kpq'), (1.107)
pq'
a=1 a=1
gives
f(dx)(dx')S (x) TI [r (m-yµ(1/i)8µ) ]aL1+(x-x')Si(x')
1
a
0
which is the projection matrix to select y0 ' = + 1 for each index a. Thus, in
the rest frame, the sources reduce effectively to Su 1u2 ••• a"' each a being a
two-valued spin index. Now suppose s, 1, 2 ••• ," has definite symmetry property.
Specifically, let us consider the totally symmetrical one. Then Sa 1 u 2 ••• un is
totally symmetrical. As we all have learned from elementary quantum
mechanics, this produces the state of maximum resultant spin, that is, the
sources are those of a particle of spin
s = (1/2)n, (1.112)
and
s = (1/2), 1, (3/2), ... '
as
n = 1, 2, 3, ....
26 J. SCHWINGER
Only s = 0 is missing in this series. For this one has to take n = 2 and use an
antisymmetrical spinor source s,,, 2
(x). Incidentally, for n~3, the indepen-
dent components of a totally antisymmetrical multispinor vanishes identi-
cally in the rest frame, since S 111112 ... 11" = O(n~3). Note that more complicated
symmetry patterns give equivalent descriptions, to the extent that a definite
spin appears in the rest frame. Consider, for example, n = 3 with the require-
ment of antisymmetry in a pair of Dirac indices. The latter contributes zero
spin in the rest frame and we have a possible description of an s = ( 1/2)
particle.
The symmetry property of the matrix propagation function is expressed by
= fI(-)[y 0 (m-yµ(1/i)8µ)]aL1-r-(x-x'),
lX
or
{IJ [y 0 (m-yµ(1/i)8'µ)]aL1+(x' -x)Y
a
(1.113)
If the algebraic properties of the sources are to match the symmetry pro-
perties of the propagation function, we learn that
J
n =even, s =integer: [ S(x), S(x') = 0, BE
be rewritten as
1= I
{n)
(0-l{n}/
(0- IO+)s
(1.116)
PARTICLES AND SOURCES 27
As in the discussion of the previous section for spin 0, eigenvalues q'
appear as phase factors in the identification of multiparticle states, since
q' = 1, or q' = - 1 = (i) 2 • These phase factors disappear on forming
x Re (1/i)Lf+(x-x')S(x')}, (1.118)
since S(x)q II y0 aS(x') is real with abnormal statistics, while S(x) II y0 aS(x')
IX IX
is real with normal statistics. In both cases, the reality property follows
simply from the Hermitian character of the matrices. Now the indefinite
eigenvalues of q are explicitly involved in (1.118), and the completeness
requirement is violated.
2. ELECTRODYNAMICS
These are the fields of the corresponding sources. They are numerical
quantities of the same type as the sources, namely, Aµ(x) are totally com-
mutative numbers, and IJl(X) are totally anticommutative. The auxiliary
quantities, fields, are very useful, since they summarize the effects, in the
region of interest, of sources which may be very far away. Now, w2 can be
represented alternatively as
In the latter form only fields appear. A third form combines the two repre-
sentations,
w2 = f (dx)[ 1J1(x)y017(x)-(1/2)1J1(x)y0 (y(1/i)8 + m)IJl(X)]
+ f (dx)[ Aµ(x)Jµ(x)-(1/2)Aµ(x)(-8 2 )Aµ(x)]. (2.6)
All the three forms are equivalent in virtue of the relations between fields
and sources, (2.3). But the last form has the advantage that these connections
can be derived from it instead of being stated independently. Let us examine
the dependence of w2 on the sources 17, Jµ and the fields IJI, Aµ,
Now, if we insist that w2 depends only on the sources, that is, that it be
stationary with respect to variations of the fields, the field differential
equations are recovered. In this form w2 is an action, and we write
(2.7)
(2.8)
FIGURE 2.1
£q J£:
-p2>m2
-p2=m2
FIGURE 2.2
e
J I/
FIGURE 2.3
,,
J
J
I/
FIGURE 2.4
J ,.,
I/
e - y scattering e - e scattering
(Fig. 2.4). This process can be continued. For example, from electron-
photon scattering, one gets an effective electron source of the form 11JJ.
Single electron exchange with an effective electron source of the form 17J
gives a new process which involves 1717JJJ, and so on. Thus, one gets a series
34 J, SCHWINGER
of dynamical processes involving 1717J, 17ytJJ, yt17JJJ, ... etc. This proliferation
introduces, at a first stage, an infinite variety of interactions, in skeletal
form, involving only single particle exchanges.
One can give an analytic expression to describe all these processes by
beginning with the primitive interaction term w3 which must have the form
f
w3 = (1/2) (dx)(dx')(d~)yt(x)y°G(xx', ~µyt(x')Aµ(~), (2.12)
o c>w'(A) = 0, (2.15)
µ c>Aµ(x)
or
c>w'
f (dx).A.(x)oµ c>Aµ(x) = o, (2.16)
(2.17)
which asserts that w'(A) must remain unchanged when Aµ(x) is displaced by
oµ.A.(x). In other words, w'(A) must be gauge invariant.
An elementary solution to both requirements is perfectly familiar. It is
obtained from w2 by the gauge-covariant substitution
aµ~aµ-ieqAµ
PARTICLES AND SOURCES 35
and
J
w = (dx)[ JµAµ-(1/4)PvFµv+(1/2)17y 01f1 A]
= f(dx)[JµAµ-(1/4)PvFµv]
0 0
+(1/2) f (dx)11Y 1/1+(1/2) f lf/Y eqyAlfl
+(1/2) f lflJJ°eqyAG+eqyAlfl+ .... (2.26)
The action principle also supplies the field equation for A (in the Lorentz
gauge, oµAµ = 0),
-o 2 Aµ =Jµ+(1/2)1f1y°eqyµlfl+.... (2.27)
Again, we use a special notation, At/t, for the vector potential under the
influence of 1f1 to distinguish it from the noninteracting field Aµ. Then
,.,
(2.28)
and
f
+(1/2) (dx)(dx')lfl(x)y 0 eqyA(x)G+(x-x')eqyA(x')lfl(x')
+ ... , (2.29)
in which the interactions are now made explicit. The first term describes the
noninteracting system, the second term is the primitive interaction, the
third term describes e- -e- ore--e+ scattering, and the fourth term des-
cribes electron-photon scattering or pair annihilation. The infinite series in
(2.29) is a sequence of increasingly elaborate interaction skeletons. Later
terms in this series do not contain modifications of earlier ones.
It should be emphasized that the iterated solution is a classification of
processes in terms of increasing degree of complexity. It is not a perturbation
expansion. The physical electron mass m, and the physical electron charge e,
which are identified originally under specific physical circumstances, will
never change their significance when the class of phenomena under exami-
nation is enlarged.
PARTICLES AND SOURCES 37
2.3 Calculation of Matrix Elements
FIGURE 2.5
,, 1
112
shown in Fig. 2.5. When the substitutions (2.30) are made, in (2.29), Comp-
ton scattering is described by terms involving l/lil// 2 A1 A2 , which are
f
w = (dx)(dx')[ 1J1 1(x)y 0 eqyA 1 (x)G+(x-x')eqyAi(x')'/1 2(x')
+ 1/1 i (x)y 0 eqyAi(x)G+(x-x')eqyA 1 (x')1J1 2(x')]. (2.31)
In these two terms, the roles of initial and final photon are reversed. Therefore
w is invariant under the interchange A 1 ~A 2 • This symmetry, so called
crossing symmetry, is automatically incorporated in the source theory, and
is simply a consequence of particle statistics.
Now, the interaction region is far away from the sources, and since the
interaction certainly occurs later in time than the emissions and earlier than
the detections, we have, in the interaction region,
1/12(x) = L .JdwP2m ueipxif/ puq'
paq
2
puq
Aiµ(x) = L il1u *.Jdwk eµu *e-ikx. (2.32)
kA.
(2.35)
Substituting the expressions (2.32) into (2.33), we identify, with the. aid of
(2.34) and (2.35),
(2.36)
Standard procedures can now be followed to calculate the differential cross
section, etc.
PARTICLES AND SOURCES 39
2.4 Two-Particle Exchange
FIGURE 2.6
+
,,
-p2 =m2 -p2>m2
will be modified to account for the fact that mass values other than -p 2 = m 2
can now be exchanged. The new propagation function G+(x-x') must
have the form
G+ = G++ ... ,
where the additional terms refer to higher mass exchanges. If the sources
cannot supply sufficient mass, then G+ effectively reduces to G+, as it
should, since this is the situation under which G+ was originally introduced.
The primitive interaction also contains the possibility that a photon source
with -k 2 >(2m) 2 creates an electron-positron pair. Two such sources interact
by exchanging a photon, or an electron-positron pair (Fig. 2. 7), and the
FIGURE 2.7
J J
J J
-k2~(2m)2
40 J. SCHWINGER
FIGURE 2.8
ordinary Coulomb scattering and the second term describes pair annihi-
lation and subsequent pair recreation. If now the electron and positron
created by an extended photon source are permitted to interact, one recog-
nizes that the second mechanism just described is the one that produces the
modification D+ -+D+. The dynamical details are suggested in Fig. 2.9(a).
The first mechanism, the ordinary Coulomb scattering [Fig. 2.9(b)] leads
to a new phenomenon. The quantitative theory shows that the effect is
represented by the alteration of the electromagnetic properties of the
electron, namely, the introduction of an electric form factor, and an addi-
tional magnetic moment, with its form factor.
PARTICLES AND SOURCES 41
In order to get a complete picture of the second stage of description, one
must also consider multiple two-particle exchanges. The pair creation from
an extended photon source is viewed as the conversion of a virtual photon
FIGURE 2.9
(a) (b)
FIGURE 2.10
f
which is to be added to (dx)(dx')J1 µ(x)D+(x-x')Jl'(x').
A more complicated tensor structure cannot appear, in virtue of the general
property of a photon source,
-.
Relativistic invariance requires that a(x-x') be a scalar function, and we
can write
a(x-x') =
f (dk)
- - 3e 1
(2n)
"k(
x-x
')
a(-k )
2
'
(2.38)
One recognizes
if dwkeik(x-x') = A+(x-x', M 2)
for the special circumstances x 0 > x 0 ' under consideration. Thus, inclusion
of multiparticle exchange gives the modified photon propagation function
D+(x-x') = D+(x-x')+ f dM 2 a(M 2 )L1 +(x-x', M 2
). (2.41)
D (x-x') =
+
f (dk) eik(x-x')"jj (k)
(2n) 4 + '
(2.42)
with
- 1
D+(k)= - 2- . +
f dM i a(M2)
.• (2.43)
k -ze k 2 +M 2 -ze
PARTICLES AND SOURCES 43
There are general physical restrictions on the function a(M 2 ) from vacuum
persistence probabiJity considerations. For a weak source, we have
f
(O+ Io_ y = 1+ (i/2) (dx)(dx')P(x)D+(x-x')Jµ(x')
(2.45)
Next, consider the situation in which the source contains only time-like
momenta. Since
(2.48)
and
I (dk)
-m5(k
4
(2n)
+M )=
2 2 I( dk) 1
- · -3 = f0 d w k
(2n) 2k '
(2.49)
we get
1(0+ 10_)11 2 = f
1- dwkJµ(k)*.fµ(k)/k2=0
- f dM 2
f
a(M 2 ) dwJµ(k)* Jµ(k)f-k2 =M2. (2.50)
In the rest frame of kµ, M#-0, we have J 0(k)=0 and Jµ(k)*Jµ(k)
= IJ(k) I 2 > 0. Therefore it is necessary that
(2.51)
(2.52)
44 J. SCHWINGER
if f
exp[ (dx)1f1(x)·J'°17(x) J~ -(1/2) (dx)(dx')lfl(x)y 0 17(x)
x 17(x')y0 lfl(x'), (2.53)
one identifies
(2.54)
where the left-hand side is regarded as a matrix. The consideration of two
such effective electron-positron sources give the following contribution to
the vacuum amplitude:
-(1/2) f (dx)(dx')A 1 µ(x)A 2"(x') Tr[ eqyµG+(x-x')eqyv
x G+(x' -x)], I"• (2.55)
when the substitution (2.54) is made for 17 1 (x)17 1 (x') and 17i(x)17 2 (x') in the
expression
k=p+p',
f
(2n) 3 dM 2dwk<5(p+ p' -k). (2.59)
Thus, (2.58) reduces to
f
-(e 2/2) dM dwkJ/(-k)Jz"(k)(1/M 2 ) Iµv(k),
2 2
(2.60)
PARTICLES AND SOURCES 45
where
Iµv(k) = (2n) 3 f dwPdwp·~(p + p' -k)
x Tr [qyµ(m-yp)qy"(-m-yp')]. (2.61)
A simple evaluation in the rest frame of k gives the result
(2.62)
in which the kµk" term will not contribute, since the current Jµ is conserved.
With the definition
a(M 2 ) =
3n M
1J1- (2m)
ex - · -
2
-
M
2
( 1+
M2 '
2
2m-) M>2m
'
(2.63)
e1
where ex = - is the physical fine-structure constant, we get the contribution
4n
to the vacuum amplitude
- f dM 2 a(M 2 )dwkJ1 µ(-k)J2 µ(k)
=if dM 2 a(M 2 ) f (dx)(dx')J1 µ(x)if dwkeik(x-x')J2µ(x'). (2.64)
Again we recognize that L1+(x-x', M 2 ) is the appropriate generalization
for the idwk integral. The modified photon propagation function appears as
- 1
D+(k)= - 2- - . +
f dM z a(M2)
• ,
k -ze k 2 +M 2 -ze
which is a particular realization of the general form (2.43). The function
a(M 2 ) given by (2.63) is real and non-negative. For M»2m, the integral
behaves like
dM 2 1
f Mz kz+M2
FIGURE 2.11
"'•
where A is the vector potential constructed from the photon source J with
the aid of the modified propagation function D+. The momentum space
form of the second term is
-
1
-
2
I--
4
(dk) dwPdwp.17(-p)y°(m-yp)eqKµ(p, p', k)
(2n)
x (-m-yp')11(-p')Aµ(k). (2.66)
therefore, M 2 is the only scalar that can be constructed from the momentum
vectors p, p' and k. The possible vectors are JI', kµ and (p- p')µ. Suppose
we use Lorentz gauge, so that
Furthermore,
(p-p')µ = - (1/2){yµ' yp-yp'}.
In virtue of the projection matrices (m-yp) and (-m-yp') in (2.66), when
yp appears as a left factor (denoted as Y]3), it can be replaced by -m, and
similarly a right factor of yp' (denoted as yp') can be replaced by + m. Then
yp = yk - yp' ~ yk - m,
"YP' = yk - yP ~ yk + m.
Therefore, the only possible structure for Kµ(p-p', k) is
or
(2.68)
Since there is no individual reference top and p', (2.66) can be put back in
coordinate space in the form
f
i(1/2) (dx)l/f(X)y 0 eqyl'l/f(X)eikxdM 2K 1(M)idwkAµ(k)
and
Fµv(k) = i[kµAv(k)-k"Aµ(k)] (2.71)
f 0
We are discussing an addition to (dx)(1/2)1/fy eqyµl/fAµ which contains the
primitive interaction and a description of how the gauge variance of A (or A)
is tied to l/f. Therefore, one requires all additional effects to be gauge
invariant. The second term in (2.69) is gauge invariant by itself, and the
first term can also be so written [kµAµ(k) =OJ,
(2.72)
The causal situation under consideration is clear cut. The source produces a
virtual photon which converts into an electron-positron pair close to the
source. Then various interactions take place at a later time. Thus > x 0 ', ;o
and again we can introduce Lf+(x-x', M 2 ) as the appropriate space-time
iJ
generalization of dwkeik(x-x').
The complete structure of (2.65) is then
(2.73)
where, using Fourier transforms,
and
Here F1(k) and F2 (k) are, respectively, the electric and magnetic form factor.
And µ' in (2. 73) is the anomalous magnetic moment.
Let us now carry out a calculation of a specific mechanism. Consider the
effect of the Coulomb interaction of the electron-positron pair emitted by
an extended photon source. As we have discussed [see (2.54) in particular],
the primitive interaction describes an extended photon source as an effective
electron-positron source. Inserting this effective two-particle source into
PARTICLES AND SOURCES 49
the electron-positron scattering description, one gets a contribution to the
vacuum amplitude given by
f
(1/2) (dx)(dx')(d()ip(x)JJ°eqyµG+(x-1;)
x eqyA(()G+(l;-x')eqyµip(x')D +(x-x'). (2.75)
FIGURE 2.12
The situation is described in Fig. 2.12. The causal arrangement is such that
x 0 > <; 0 , and x 0 ' > <; 0 • The evaluation is performed without difficulty in the
center-of-mass frame of the two-particle system. Only one special point
should be mentioned. It is the integration over all scattering angles. The
long-range Coulomb interaction gives infinity in the forward direction. This
difficulty can be removed by using a screened Coulomb potential, i.e.
(1/r)e-µr, which is equivalent to the unphysical use of a small photon massµ.
It is clear that this is simply a version of the infra-red problem. This problem
is easily resolved by asking the proper physical question, that is, soft photon
emission processes must also be considered. The result is expressed in the
general framework of (2.74) as
(2m)2 1
2
i/! (M)=A{ Ji-(2:;)" M>2m,
- 23+ (2m)
2
M ] , M>2m. (2.76)
The result that we have just obtained describes the electromagnetic properties
of a free electron. For the Lamb-shift calculation, on the other hand, one has
to deal with a bound electron with p 2 + m 2 ¥= 0. In this case one can use J
to represent the charge distribution of the nucleus, which produces its
effect through A . The basic problem is then the multiple scattering of an
FIGURE 2.13
.J (dp1) (dp2) 0
z ( n)3 ( n)3 lf/1(-P1)Y (-eq)Aµ{p1 -pi)
2 2
lf/1(-P1) = f (dx1)1f11(X1)eipixi,
f
lf/2(P2) = (dx2)e-ip2x21f1i{X2),
f
Aµ(p1 -p2) = (d()e- i(p1 - p2>~ Aµ((). (2.79)
°
The well-defined causal arrangement guarantees that x 1 > c; 0 , and c;0 > x 2 °.
On writing
(dp1) 2 (dp2) 2
(2n)3 = dmP1dM1 ' (2n)3 = dwP2dM2 ' (2.80)
in which A+(c;-x 2 , M 22) and A+(x 1 -c;, M 12) describe, respectively, the
transmission of mass M 2 from the extended source 17 2 to the nuclear potential
A(e), and transmission of mass M 1 from A(() to the extended source 1'/t •
After the space-time generalization, we return to the momentum-space
representation,
-z
·f (dp1) (dp 2 ) 0[
( n)4 ( n)41f11(-p1)Y K1yi(-eq)Aµ(p1 -pz)
2 2
+g2eq(1/2)aµvpµv(p1 -P2)}/f2{p2)
1
f
x dM 12 dM 2 2 [ •
P1 2 +M1 2 -ze
. 1
2+M 2 • f(<P1 -pz) 2 , Mi 2 , Mz 2 )] , (2.82)
P2 2 -ze
in whichp 1 2 andp/ are not restricted to -p 1 2 = -p 2 2 = m 2 • The additional
term/(<p 1 -p 2 ) 2 , M 12 , M/) appears here because the discussion by means
of the well-defined causal situation effectively studies the dependence upon
p 1 2 and p 2 2 , but leaves undetermined any contribution that involves only
the third invariant {p 1-p 2 )2. The/term can be removed by subtracting the
already known result for - p 1 2 = - p 2 2 = m 2 • Despite the way (2.82) has
been written, this is to be performed separately for the yA and the aFterms,
and turns out to be unnecessary for the latter.
3. STRONG AND WEAK INTERACTIONS
(3.1)
where x are the fields corresponding to the sources S, and .!l'(x) is the
Lagrange function of the system. The latter is not a local function of the
fields, but may be so represented under limited physical circumstances.
Consider for illustration the n + N system at low energy. The pion is an
isotopic triplet (T= 1), and the nucleon is an isotopic doublet (T=(1/2)).
One also has to distinguish a nucleon from an antinucleon. This distinction
is conveyed by the use of a complex source. We will use the particle symbols
to represent the corresponding fields. For the n+N system we have then
(3.2)
53
54 J. SCHWINGER
where
N = N*y 0 , (3.3)
and
(3.4)
f -
!t'11:N =-N~iy 5 r:N · oµn
m11:
2
/ 0 ) _
+ ( m11: N ~r: N · oµn x n. (3.7)
If we consider the exchange of a virtual 1t, the first term gives the long-range
behavior of nuclear forces. If the exchange of a virtual nucleon is considered,
it describes the low-energy limit of p-wave n-N scattering. The value off
obtained in this way is
f = 1.01±0.01, (3.8)
12
or ,...., 0.08. The second term is introduced phenomenologically to account
4n
for the low-energy s-wave n-N scattering. The experimental data are
m11:(a(1f2)-a(3f2)) = 0.292±0.02,
m11:(2a(3!2) + a(lf2)) = -0.035 ± 0.012, (3.9)
where a(l/l) and a< 312 > are, respectively, the scattering amplitudes for the
n+N system in the T= (1/2) and T= (3/2) states. The/0 2 term alone gives
2a(3f2>+ac 112 >= 0. But the/term also gives a smalls-wave contribution
(3.10)
m 1
where 2 ~ - . The result is consistent with experiment. The relation
M 6.9
(3.11)
PARTICLES AND SOURCES 55
gives
The two parameters I and lo are the information extracted from the low-
energy strong interaction phenomenology of the n + N system.
56 J. SCHWINGER
<5n = - <5w x n,
<5N = (i/2)r · <5wN, (3.13)
Mn=l=Mp.
Isotopic spin invariance is only a partial symmetry, abstracted by dis-
regarding certain terms in the Lagrange function, specificall:Y the mass
differences in the present case. This is a very general feature, empirically, of
strong interaction dynamics. Another significant example of partial sym-
metry is obtained upon the recognition that the pion mass is the smallest of
all meson masses. If it were zero, then the Lagrange function
n-+-n+q.1.
Does this symmetry have any significance for the nN interaction? Observe
that it is also true for the f interaction term of (3. 7), since this term involves
only derivatives of the n· field. However, it is not valid for the / 0 term.
Considering an infinitesimal displacement, we obtain
2
/ 0 ) _
<5.!l'nN = (
mn N'/r:N. oµ(n x <5q.1). (3.14)
i( ~:)'
JN= r · n xc5tpN,
- (/o)
<5N = - i mn
2
-
N r · n x <5q.1. (3.15)
(3.17)
or
(3.18)
where
(3.19)
if the parameters <5w 12 ± <5w 34 , etc. are introduced. Thus, we have now
two independent isotopic rotation groups with parameters
<5w ± = <5w ± M<p . (3.24)
Under a spatial reflection, '5w~ '5w, but '5<p~ - <5<p, therefore '5w ++-+'5w _ .
The combinations '5w+ and '5w_ thus correspond to opposite handness, or
chirality, which are interchanged under a spatial reflection. We shall call the
0 4 transformations in general and '5<p transformations in particular, chiral
transformations.
Thus far, we have abstracted the chiral group from the phenomenology of
the low-energy n + N system. Let's see if we can apply the result to deduce
new relations. The actual response of the Lagrange function to a constant
'5w and <5<p transformation is
(3.25)
(3.29)
where
(3.30)
PARTICLES AND SOURCES 59
(3.31)
~n = -&v(x) x n,
~Pµ = -&o(x) xpµ+(1/g)oµ~w(x) = (1/g)Dµ~w(x). (3.36)
The first term in the p transformation expresses the fact that the p meson is
an isotopic-spin-1 particle, while the second term is characteristic of a
gauge field. The general response of ft' is
(3.37)
(3.38)
We can find the connection with the previous approach through the p field
equation
=gjµT· (3.41)
Under the previous low-energy circumstances, the p meson cannot be
produced as a real particle, i.e., iJ2«mp2 , and (3.41) reduces to
m2
p •
--pµ'::::!.)µT' (3.42)
g
so (3.37) gives the previous result. It indicates that the introduction of the
p meson is a consistent generalization.
We can also make contact between the p and chiral transformations. Under
the displacement
we have
f5jµT = 0µ11: X f5rp+ .. .
= <Vn x <>rp)+ ... . (3.44)
. (3.45)
(3.49)
Ll+(x-x) =
I I (dp)
- -4 -
eip(x-x')
---
P (2n) p +m/-ie
2
1 I
-+- - 2
c5(x-x), (3.53)
mp
and (3.52) becomes
g2
f
- 2 (dx)Nyµ7:N · oµnxn
2mP
in which the connection (3.49) has been used. Notice that (3.54) is just the
s-wave coupling term in (3. 7).
Since the coupling constant g is known, it would seem that the interaction
term Pµ · oµn x n could be used to describe the instability of the p meson,
the decay p-+-2n, i.e., the p width could be calculated. However, the p decay
is a high-energy process, and a more elaborate high-energy theory is needed.
PARTICLES AND SOURCES 63
(3.58)
Since the combinations Ow±(~) Otp appear everywhere except the last
terms, the requirement of a group structure obviously demands
(3.59)
And we have
(3.60)
with
2g
<5w ± = <5w ± - <5rp. (3.61)
mA
64 J. SCHWINGER
2/o
<5w ± = <5w ± - <5rp;
m1t
therefore
g Jo g
-=-=-=--,
mA m1t .J2mP
and
(3.62)
Only the combinations (p + .91) and (p- .91) can occur in !l'P~ in order
2 2
to satisfy the chiral invariance, yet these two structures must occur with the
same weight since parity is conserved in strong interactions. The partial
symmetry response of !£ is
or
m2
<)ft'= - _P_pµ • Oµf5W-mA.9/µ • oµ<5rp. (3.66)
g
Comparison with the low-energy behavior (3.26) indicates the necessary
correspondences for low-energy behavior,
(3.67)
PARTICLES AND SOURCES 65
2
If we now include the n term [ (c\+gpµ x )n-(1/2)mAdµ] , it is
immediately seen that n and dµ are mixed objects. We must carry out a
diagonization, which will also change the normalization of the n term. In
anticipation of this, let us write
1
dµ = A 1 µ+-oµn, (3.69)
mA
or, more generally,
1
dµ = A 1 µ+-Dµn, (3.70)
mA
gives
a= 1.
And we get
!l' = -(1/2)[ (oµn) 2 +m/n 2 ]-(1/2)[ (1/2)(oµA 1 v-ovA 1 µ) 2
+m/A 1 /J,
i.e. m A is the mass of the A 1 particle. One compares the theoretical prediction
(3.72)
But
mAdµ = oµn+mAA 1 µ,
which shows that the n term is correct. We are left with the necessary
correspondence
f-
mAA1µ"' - -Nyµiy 5 rN, (3.73)
mn
which anticipates a NA 1 coupling of the form
(3.74)
oN = i2 ( -
g )21- r · n x JqiN, (3.77)
mA 2
which is a nonlinear gauge transformation depending on the n field. This
suggests that one should replace pµ in the Np coupling term by a new
field Pµ':
I 2g
Pµ = Pµ + -2 n x (mAdµ-Dµn). (3.78)
mA
Under a oqi(x) transformation, we have
I 2g
<5pµ = - 2 Dµ(n x <5qi), (3.79)
mA
PARTICLES AND SOURCES 67
which, in conjunction with (3. 77), correctly gives chiral invariance. In
arriving at (3.79), we have neglected the variation of Pµ in Dµ which is outside
the limited framework being considered. So far, the f term in nN coupling is
missing. However, we always have the possibility of adding a phenomeno-
logical coupling which is a chiral invariant by itself. The same situation
occurs also in electrodynamics at a phenomenological level. A magnetic
coupling, which is gauge invariant, has to be introduced to describe the
anomalous magnetic moment. In order to account for the low-energy
p-wave nN scattering, one must introduce the coupling
(3.80)
and
(3.86)
(3.88)
PARTICLES AND SOURCES 69
These expressions can be simplified,
(3.89)
~~100MeV (3.94)
[without the factor (3/4), the result would be rp"' 180 MeV] and
rA1~200 MeV. (3.95)
Experimentally, the average width of p is 125 MeV (116 MeV for p 0 and
132 MeV for p±), and the width of A 1 is 130±40 MeV. The theoretical
results are certainly qualitatively right, but indicate that the predicted I'P is
too small and I'A 1 is too big. However, the qualitative consistency lends
support to the underlying theoretical ideas.
Of course, we have used only the simplest possible theory in which no
adjustable parameter appears. As in the discussion of nN coupling, it is
always possible to add individually invariant terms with arbitrary coeffi-
cients. A relevant example here is
(3.96)
70 J. SCHWINGER
where
(3.97)
and
(3.98)
are chiral invariant combinations as far as our problem is concerned. The
contributions to pnn and A 1pn couplings are
~ ,,_!_pµv.
1 2
aµ nx an-
v
yg //,..µv. Ai µ XO v
n. (3.99)
2 mA mA
{3.103)
and therefore
v p·p
PnvP ~-mA- (3.104)
mP
Also
or
-ppPn~(1/2)mp 2 , (3.105)
neglecting the pion mass. Thus
PARTICLES AND SOURCES 71
or
(3.106)
(JN= ;(:J 2
t · 11: x &pN, (3.111)
n
respond universally. Thus,
in which we anticipate that a scale dilation of orp is also required for the
complete group property. The group property for the transformation law
(3.113) can be checked, as before, by comparing successive transformations
in one order
(:J (11:'
11:" = 11:' + (j2(P [ ! + (:A 11:')']- 2 X(j2(P) X11:',
-2 (:J 2
(;J, (P x (j2(P) x 1!' (3.115)
PARTICLES AND SOURCES 73
or
+2 (:J 2
(3.117)
with
(3.118)
since all higher order terms cancel completely. The nucleon transformation
law (3.111) is now also exact. The n transformation law (3.113) can also be
exhibited in the following form
(3.120)
(3.121)
(3.122)
74 1. SCHWINGER
with
+m/(1-~p )(n 2 2
) l (3.125)
is partially chiral invariant, broken only by the pion mass term in accordance
with (3.124). For the application to n-n scattering, the n field satisfies the
equation
(3.126)
and
(oµn 2 ) 2 + 2(oµn) 2 n 2 = 20µn. oµ(nn 2 )
-+ - 2o 2 n · nn 2
(3.127)
Consequently,
All sensitivity to the model parameter pis in the pion mass term.
The physical background of these considerations should be kept in mind.
Chiral invariance is recognized from the strong-interaction phenomenology
PARTICLES AND SOURCES 75
of the low-energy re+ N syste~ The conclusions are therefore mainly
about low-energy .processes. The possibility of generalizing to high-energy
processes is of course a tempting one, and in fact we have done so in the
discussion of A 1prc dynamics. But as a local interaction, (3.128) is expected
to be only applicable in the low-energy region. Thus we apply it to the
n-n scattering at the threshold. The implied scattering amplitudes for
T = 0, 2 states are
(3.129)
A model-independent combination is
(3.130)
17*(960)-+11+2n, (3.131)
where the bw on the right-hand side is a vector, while the ow on the left-
hand side is a matrix. Now (3.132) simply reduces to the ordinary isotopic
rotation of the n field.
The general chiral transformation on </> is
(3.136)
The exact group property
(3.137)
with
(3.138)
can be readily verified. For SU2 , &p = (1/2)r · &p, and (3.136) gives the
previous chiral transformation of the n field, since
(3.143)
(3.144)
(
coupling
(3.147)
'f]-n7+a11*n2'
rJ*-+rJ* +b'f/7t2'
will make z,, + 2,,. chiral invariant as far as the present applicatio,n is
concerned. The additional terms so obtained which contribute to the decay
'f/*-+'f/+ 2n are
since 1J and n* are being used as fields of real particles. Therefore, there is
no contribution from 2 11 + 2 11 •• This is a consequence of the stationary
action principle.
The structureless coupling n 2 r/'f/* simply introduces an invariant phase-
space factor. For the decay
where T,, is the kinetic energy of 11. It is very useful to note that
(3.151)
within experimental error. Experimental analysis is expressed in terms of
the Dalitz parameter
m,,+2mn T,, 8
y= -1, -1<y< - ' (3.152)
mn m,,.-m,,-2mn 7
and (3.150) becomes
7
1--y (3.153)
17 '
apart from a multiplicative factor. Thus, the spectrum will deviate from the
pure phase-space prediction by an asymmetry factor
( 1- 177 y y. (3.154)
151(1-2.
3
17
y )' 27, 21, 15, 11, 7,
(3.160)
PARTICLES AND SOURCES 81
Thus, the nucleon coupling, for example, becomes
gNyµ(1/2)rN · Pµ~gNyµ(1/2)rN · Pµ
+ N yµe( 1/2)r3 AµN, (3.161)
2 e
mp -plAw (3.162)
g
Through the identification of a photon source, (3.162) implies a coupling in
the action, through photon exchange,
(3.163)
(3.164)
(3.165)
where
(3.166)
82 J. SCHWINGER
(3.167)
OqiWe=
m
_P_
( g
2
e) --f
2
2g
(dx)(dx')
mA
(3.168)
and
m
o/we = ( ~
2e) 2(mA
2g)2 f (dx)(dx')Dµv(x-x')
x {(dµ x ocp)3(x). (dv x ocp)3(x')
+ [ (pµ x ocp) x bcp ]3(x)p/(x')}. (3.169)
Let us emphasize that we are now considering processes that are not included
in the strong-interaction phenomenology. A typical term in (3.169) contains
two fields. dP(x)dv(x'), or pµ(x)pv(x'), which operate at two different
FIGURE 3.1
or
FIGURE 3.2
- Gµv(P)a]. (3.177)
This is identified as the chiral response of the additional charged pion mass
term
(3.178)
and gives
(3.179)
84 J. SCHWINGER
Now
(3.180)
1
dµ=A 1µ+-oµn, (3.181)
mA
we also have
PµPv 1
gµv+--2 --2 PµPv
mA mA
Gµv(P)a =
p
2
+ mA 2 - ie
+ p 2 + mn 2 - is
• (3.182)
(3.183)
But
(3.184)
so that
om
n
2 =6e2m
P
2 f!__ (dp) _1_r ___1 __
i (2n) 4 p 2 -it:l_P 2 +m/-ie
om n
2
=[ 3
1X log 2Jm
2n P
2
'
(3.187)
FIGURE 3.3
2 3a 2
(omn ),mp= -mP , (omn)n1tp = 3.6 MeV. (3.188)
4n
(3.189)
But in this approach we can also consider the mn 2 correction to these pro-
cesses. For the most important one (nnp) we get
2
(omn )nnp = -3a mp [ 1 + -
2 2 m/ mP + -1 )]
- ( log-- (3.190)
4n mp 2 mn 2 2
in which the second term gives a 13°/0 correction, and (om1t)1C1tP becomes
(3.191)
(3.192)
and
2
fi'P = -(1/2)pµv · (aµPv-avPµ)+(1/4)(Pµv)
-( 1/2)mp 2 (pµ) 2 • (3.193)
(3.194)
(3.195)
and fi'n and fi'P restate the previous structures. But now the pion is des-
cribed by five fields (n and nµ), and the p meson by ten fields (Pµ and Pµv).
Thus, on supplying one more scalar, we have a one to one correspondence
between the multispinor and the tensor field description, each having
16 components.
Let's now consider a situation in which the meson fields carry arbitrary small
spatial momenta,
vx=O, (3.196)
The correspondence between the multispinor notation and the tensor field
description is established by introducing the 4 x 4 matrices
M< ±) = (TC± Hp) . O"'l" + (112 ± H (1)) • (J
-(±S1 +mPp 0 ) • r-(±S0 +mrow0 ). (3.199)
These constructions are such that the even-parity structures in the Lagrange
function are reproduced,
(1/16) Tr [(M<+>) 2 +(M<->) 2 ]
=(1/2)[(rc) 2 +H/+m/(p 0 ) 2 +Hro 2 +m 0 /(w 02 )+ ... ] . (3.200)
where each Ma<+> acts only on the corresponding indices. The scale factor is
fixed by the requirement that p0 couples universally to the isotopic-spin
3
density 'ff'* L (1/2)ra 'P giving the form
IX=l
,L(1/2)ra~C1/2)r,
IX
L(1f2)aa~(1/2)u,
IX
(3.204)
PARTICLES AND SOURCES 89
where the first two are simply the assertion that ( 1/2)r and ( 1/2)a are the
isotopic spin and ordinary spin of the nucleon, respectively. The third one
is obtained by a simple calculation. The nN coupling
(3.205)
(3.206)
I (5/3)
lo= .J2
1.01 +0.01
Exp, strong interaction: - = 1.20±0.03
= 1.18 0.84±0.03
(
Exp, weak interaction: 1.18±0.03, (3.207)
since
g 1 lo
(3.208)
2mP = .J2 mn ·
One also notices that the substitutions
(3.209)
reproduce the correct electrical charges of the baryons. Thus (3.203) pre-
dicts the total magnetic moments for the nucleon
5 e 1 e
µ=---7:3+---. (3.210)
3 2mP 3 2mw
If the mass difference between w and p0 is ignored, we recover the famous
ratio
_ µP =~
µn 2
An assumed mass difference of 25 MeV gives the ratio 1.48. The experi-
mental value is 1.46. The absolute moment predictions are about fifteen
percent too small.
90 J. SCHWINGER
When the two particles are close to each other, the n contribution alone
involves a term --(1/r 3 ), which is physically unacceptable. The particular
combination 1t +HP removes the overly singular potential. The nucleon not
only interacts with n but also with A 1 in the same manner. The much shorter
ranged A 1 coupling tends to increase the tensor force of the n coupling. This
suggests qualitively that the HP coupling could be stronger than indicated.
If so, the predictions about the absolute magnetic moments for the nucleon
will then be improved.
PARTICLES AND SOURCES 91
3.9 Concluding Remarks
These notes record the state of the art of Sourcery in July, 1967. For
more details and later developments, go to the Source papers published in
the Physical Review. They are:
See also:
"Chiral Transformations": 167, 1432 (1968)
and these Physical Review Letters:
"Partial Symmetry": 18, 923 (1967);
"Photons, Mesons and Form Factors": 19, 1154 (1967);
"Radiative Corrections in fJ Decay": 19, 1501 (1967).