Beruflich Dokumente
Kultur Dokumente
Contents
1. Topics 7
2. Participants 9
2.1. ICRANet participants . . . . . . . . . . . . . . . . . . . . . . . . 9
2.2. Past collaborators . . . . . . . . . . . . . . . . . . . . . . . . . . 9
2.3. Ongoing collaborations . . . . . . . . . . . . . . . . . . . . . . . 10
2.4. Students . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
3. Brief description 13
3.1. Electron-positron plasma . . . . . . . . . . . . . . . . . . . . . . 13
3.1.1. Bose enhancement and Pauli blocking in the pair plasma 13
3.1.2. Dynamics and emission from mildly relativistic plasma 14
3.1.3. Evolution of the pair plasma generated by a strong elec-
tric eld . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
3.1.4. Electron-positron plasma in GRBs and in cosmology . 15
3.2. Photospheric emission from ultrarelativistic outows . . . . . 16
3.3. Correlation dynamics in cosmology . . . . . . . . . . . . . . . 16
3.4. Neutrinos in cosmology . . . . . . . . . . . . . . . . . . . . . . 17
3.4.1. Massive neutrino and structure formation . . . . . . . . 18
3.4.2. Cellular structure of the Universe . . . . . . . . . . . . 18
3.4.3. Lepton asymmetry of the Universe . . . . . . . . . . . . 19
3.5. Semidegenerate self-gravitating system of fermions as a model
for dark matter halos and universality laws . . . . . . . . . . . 20
3.6. Constraining cosmological models with CMB observations . . 21
4. Publications 23
4.1. Publications before 2005 . . . . . . . . . . . . . . . . . . . . . . 23
4.2. Publications (2005 2010) . . . . . . . . . . . . . . . . . . . . . 26
4.3. Publications (2011) . . . . . . . . . . . . . . . . . . . . . . . . . 41
4.4. Invited talks at international conferences . . . . . . . . . . . . 45
4.5. Posters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
4.6. Lecture courses . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
5. APPENDICES 55
A. Bose enhancement and Pauli blocking in the pair plasma 57
A.1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
3
Contents
A.2. Boltzmann equation with QED collisional integrals . . . . . . 59
A.2.1. Two-particle interactions . . . . . . . . . . . . . . . . . . 59
A.2.2. Three-particle interactions . . . . . . . . . . . . . . . . . 61
A.3. The numerical scheme . . . . . . . . . . . . . . . . . . . . . . . 62
A.4. Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65
A.5. Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66
B. Dynamics and emission from mildly relativistic plasma 69
B.1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
B.2. Formulation of the problem . . . . . . . . . . . . . . . . . . . . 70
B.3. Computational method . . . . . . . . . . . . . . . . . . . . . . . 71
B.4. Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72
B.5. Conclusions and future perspectives . . . . . . . . . . . . . . . 76
C. Evolution of the pair plasma generated by a strong electric eld 77
C.1. Cylindrical coordinates in the momentum space . . . . . . . . 78
C.2. The Distribution Function . . . . . . . . . . . . . . . . . . . . . 79
C.3. Two temperature DF . . . . . . . . . . . . . . . . . . . . . . . . 80
C.4. Boltzmann equations . . . . . . . . . . . . . . . . . . . . . . . . 81
C.5. Computational scheme . . . . . . . . . . . . . . . . . . . . . . . 82
C.5.1. Acceleration and electric eld evolution . . . . . . . . . 82
C.5.2. Emission and absorption coefcients . . . . . . . . . . . 82
C.5.3. Two particle kinematics . . . . . . . . . . . . . . . . . . 84
C.6. Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86
C.6.1. Non interacting systems . . . . . . . . . . . . . . . . . . 86
C.6.2. Interacting systems . . . . . . . . . . . . . . . . . . . . . 90
C.7. Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92
D. Electron-positron plasma in GRBs and in cosmology 97
D.1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97
D.2. General equations . . . . . . . . . . . . . . . . . . . . . . . . . . 97
D.2.1. Early Universe . . . . . . . . . . . . . . . . . . . . . . . 99
D.2.2. GRBs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100
D.3. Heavy elements . . . . . . . . . . . . . . . . . . . . . . . . . . . 103
D.4. Recombination . . . . . . . . . . . . . . . . . . . . . . . . . . . . 104
D.5. Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107
E. Photospheric emission from relativistic outows 109
E.1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109
E.2. Optically thick relativistic outows: wind vs. explosion . . . . 110
E.3. Optical depth along the line of sight . . . . . . . . . . . . . . . 112
E.3.1. Pure electron-positron plasma . . . . . . . . . . . . . . 113
E.3.2. Acceleration phase . . . . . . . . . . . . . . . . . . . . . 114
E.3.3. Coasting phase . . . . . . . . . . . . . . . . . . . . . . . 116
4
Contents
E.3.4. Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . 117
E.4. Geometry and dynamics of the photosphere . . . . . . . . . . 119
E.4.1. Coasting photon thin case . . . . . . . . . . . . . . . . . 120
E.4.2. Accelerating and coasting photon thick cases . . . . . . 120
E.5. Observed ux . . . . . . . . . . . . . . . . . . . . . . . . . . . . 122
E.5.1. Adiabatic approximation for evaluation of observed ux 122
E.5.2. Acceleration phase . . . . . . . . . . . . . . . . . . . . . 124
E.5.3. Coasting phase . . . . . . . . . . . . . . . . . . . . . . . 124
E.5.4. Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . 130
E.6. Instantaneous and time-integrated spectra from the photosphere133
E.6.1. Photon thin outow . . . . . . . . . . . . . . . . . . . . 134
E.6.2. Photon thick outow . . . . . . . . . . . . . . . . . . . . 136
E.6.3. Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . 136
E.7. Application to GRBs . . . . . . . . . . . . . . . . . . . . . . . . 137
E.7.1. Shell model . . . . . . . . . . . . . . . . . . . . . . . . . 138
E.7.2. Wind model . . . . . . . . . . . . . . . . . . . . . . . . . 140
E.8. Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 146
F. Correlation dynamics in cosmology 149
G. Semidegenerate self-gravitating system of fermions as a model for
dark matter halos and universality laws 157
G.1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 157
G.2. Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 159
G.2.1. Properties of the equilibrium congurations . . . . . . 162
G.3. Comparison with other DM proles . . . . . . . . . . . . . . . 163
G.4. Scaling Laws . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 166
G.4.1. Solving the scaling laws system of equations . . . . . . 167
G.4.2. Application to spiral, elliptical and group of galaxies . 167
H. Features in the primordial spectrum: new constraints from
WMAP7+ACT data and prospects for Planck 171
H.1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 171
H.2. Inationary perturbations in models with interrupted slow roll 172
H.3. Analysis Method . . . . . . . . . . . . . . . . . . . . . . . . . . 174
H.4. Results and Discussion . . . . . . . . . . . . . . . . . . . . . . 178
H.5. Concluding Remarks . . . . . . . . . . . . . . . . . . . . . . . . 181
Bibliography 187
5
Contents
6
1. Topics
Electron-positron plasma
Bose enhancement and Pauli blocking in the pair plasma
Dynamics and emission from mildly relativistic plasma
Evolution of the pair plasma generated by a strong electric eld
Electron-positron plasma in GRBs and in cosmology
Photospheric emission from ultrarelativistic outows
Correlation dynamics in cosmology
Neutrinos in cosmology
Semidegenerate self-gravitating system of fermions as a model for dark
matter halos and universality laws
Constraining cosmological models with CMB observations
Constraints on the power spectrum of primordial density uctua-
tions
Constraints on the reionization history of the Universe
Constraints on the variation of fundamental constants
7
1. Topics
8
2. Participants
2.1. ICRANet participants
Carlo Luciano Bianco
Massimiliano Lattanzi
Remo Rufni
Gregory Vereshchagin
She-Sheng Xue
2.2. Past collaborators
Marco Valerio Arbolino (DUNE s.r.l., Italy)
Andrea Bianconi (INFN Pavia, Italy)
Neta A. Bahcall (Princeton University, USA)
Daniella Calzetti (University of Massachusets, USA)
Jaan Einasto (Tartu Observatory, Estonia)
Roberto Fabbri (University of Firenze, Italy)
Long-Long Feng (University of Science and Technology of China,
China)
Jiang Gong Gao (Xinjiang Institute of Technology, China)
Mauro Giavalisco (University of Massachusets, USA)
Gabriele Ingrosso (INFN, University of Lecce, Italy)
Yi-peng Jing (Shanghai Astronomical Observatory, China)
Hyung-Won Lee (Inje University, South Korea)
Marco Merana (University of Rome Sapienza, Italy)
9
2. Participants
Houjun Mo (University of Massachusetts, USA)
Enn Saar (Tartu Observatory, Estonia)
Jay D. Salmonson (Livermore Lab, USA)
Luis Alberto Sanchez (National University Medellin, Colombia)
Costantino Sigismondi (ICRA and University of Rome La Sapienza,
Italy)
Doo Jong Song (Korea Astronomy Observatory, South Korea)
Luigi Stella (Astronomical Observatory of Rome, Italy)
William Stoeger (Vatican Observatory, University of Arizona USA)
Sergio Taraglio (ENEA, Italy)
Gerda Wiedenmann (MPE Garching, Germany)
Jim Wilson (Livermore Lab, USA)
Urbano Franca (Instituto de Fsica Corpuscular, Valencia, Spain)
Julien Lesgourgues (CERN, Theory Division, Geneva, Switzerland)
Lidia Pieri (Institute dAstrophysique, Paris, France)
Sergio Pastor (Instituto de Fsica Corpuscolar, Valencia, Spain)
Joseph Silk (Oxford University, UK)
Gustavo de Barros (former IRAP PhD, Brazil)
Wien Biao Han (former IRAP PhD, now at Shanghai Astronomical Ob-
servatory, Chinese Academy of Science, China)
2.3. Ongoing collaborations
Alexey Aksenov (ICAD, RAS, Russia)
Roustam Zalaletdinov (Tashkent University, Uzbekistan)
Valeri Chechetkin (Keldysh Institute, Russia)
Alessandro Melchiorri (Univ. Sapienza di Roma, Italy)
10
2.4. Students
2.4. Students
Alberto Benedetti (Erasmus Mundus IRAP PhD, Italy)
Ivan Siutsou (IRAP PhD, Belarus)
Damien Begue (Erasmus Mundus IRAP PhD, France)
Micol Benetti (IRAP PhD, Italy)
Eloisa Menegoni (IRAP PhD, Italy)
Stefania Pandol (IRAP PhD, Italy)
11
2. Participants
12
3. Brief description
Astroparticle physics is a new eld of research emerging at the intersection
of particle physics, astrophysics and cosmology. Theoretical development in
these elds is mainly triggered by the growing amount of experimental data
of unprecedented accuracy, coming both from the ground based laboratories
and from the dedicated space missions.
3.1. Electron-positron plasma
Electron-positron plasma is of interest in many elds of astrophysics, e.g.
in the early universe, active galactic nuclei, the center of our Galaxy, com-
pact astrophysical objects such as hypothetical quark stars, neutron stars and
gamma-ray bursts sources. It is also relevant for the physics of ultraintense
lasers and thermonuclear reactions. We study physical properties of dense
and hot electron-positron plasmas. In particular, we are interested in the is-
sues of its creation and relaxation, its kinetic properties and hydrodynamic
description, baryon loading and radiation from such plasmas.
Two completely different states exist for electron-positron plasma: opti-
cally thin and optically thick. Optically thin pair plasma may exist in active
galactic nuclei and in X-ray binaries. The theory of relativistic optically thin
nonmagnetic plasma and especially its equilibrium congurations was es-
tablished in the 80s by Svensson, Lightman, Gould and others. It was shown
that relaxation of the plasma to some equilibrium state is determined by a
dominant reaction, e.g. Compton scattering or bremsstrahlung.
Developments in the theory of gamma ray bursts from one side, and ob-
servational data from the other side, unambiguously point out on existence
of optically thick pair dominated non-steady phase in the beginning of for-
mation of GRBs. The spectrum of radiation from optically thick plasma is
usually assumed to be thermal.
3.1.1. Bose enhancement and Pauli blocking in the pair
plasma
We consider relaxation of nonequilibrium optically thick pair plasma to com-
plete thermal equilibrium by integrating numerically relativistic Boltzmann
equations with exact QED two-particle and three-particle collisional inte-
13
3. Brief description
grals. Quantum nature of particle statistics is accounted for in collisional
integrals by the corresponding Bose enhancement and Pauli blocking factors.
We point out that unlike classical Boltzmann equation for binary interac-
tions such as scattering, more general interactions are typically described by
four collisional integrals for each particle that appears both among incoming
and outgoing particles.
Our numerical results indicate that the rates of three-particle interactions
become comparable to those of two-particle ones for temperatures exceeding
the electron rest-mass energy. Thus three particle interactions such as rela-
tivistic bremsstrahlung, double Compton scattering and radiative pair cre-
ation become essential not only for establishment of thermal equilibrium, but
also for correct estimation of interaction rates, energy losses etc. Our results
on this topic are reported in Appendix A.
3.1.2. Dynamics and emission from mildly relativistic plasma
Interactions and emission in a spherical region with optically thick relativistic
plasma is studied using kinetic Boltzmann equations, see Appendix B. Being
limited by the computational requirements, we selected initial optical depth
of the order of = 10
7
, and initial temperature of the order of electron rest
mass energy. Such initial conditions allow as rst of all to study optically
thick pair plasma. Secondly, we may follow the process of self acceleration
and formation of the shell which reaches mildly relativistic bulk velocity of
expansion before it becomes transparent for radiation, similarly to electron-
positron plasma in GRB sources. Such initial value of the optical depth is too
small for GRBs, as we show in Appendix E, and consequently large Lorentz
factors are not attained, but important relativistic effect may be studied. At
the same time such initial optical depth is too large for existing laboratory
experiments. However we believe that this simulation, exploring intermedi-
ate region between laboratory and astrophysical conditions, allows to obtain
some important physical insights into kinetic evolution of electron-positron
plasma. We follow dynamical evolution of particle number density, optical
depth, hydrodynamic velocity, luminosity and spectra. Most important we
nd unexpectedly that the spectrum of emission near its peak is different
from pure thermal one.
3.1.3. Evolution of the pair plasma generated by a strong
electric eld
We investigate the behavior of the electron-positron pairs created by a strong
electric eld, see Appendix C. This problem has been studied analytically in
our previous work using simple formalism based on continuity and energy-
momentum conservation equations. Now we extend that work using the
14
3.1. Electron-positron plasma
more general kinetic approach.
We consider a system which is uniform and homogeneous in the physical
space and axially symmetric in the momentum space. The axis of symme-
try is given by the direction of the initial external electric eld. With these
hypotheses, the relativistic Boltzmann equation is solved numerically for dif-
ferent starting values of the eld. In this framework we can describe the
Distribution Function for each kind of particle in a two dimensional momen-
tum space. Our numerical code allows us to take into account the interactions
between particles as well. We study the non-interacting and the interacting
cases separately, then we compare the two runs and the role of the interac-
tions can be analyzed; in particular we are interested in the dynamical ap-
proach to the thermal equilibrium conguration.
There are many analogies between the results we obtain using the two
methods methods, namely the time dependence during the rst half oscil-
lation of all the quantities involved. However, after this short period, the two
methods give substantially different results.
We analyzed in details the case when interactions are not taken into ac-
count. For all the considered parameter sets we nd that after several oscilla-
tions the magnitude of the electric eld becomes much less than the starting
value; as a consequence, acceleration and pair production are strongly sup-
pressed. The bulk momentum parallel to the external eld becomes very
small, while the number density of the pairs saturates to a small fraction of
the maximum achievable one, well below 10 percent for all cases considered.
This result comes out only when kinetic treatment is adopted, and conse-
quently distribution of particles in momentum space is accounted for. We
nd that production of pairs and their acceleration in the same electric eld
simultaneously produce a peculiar distribution function of particles which
quickly settles down in a sort of equilibrium. This equilibrium is well de-
scribed using a relativistic two-temperature distribution function, with the
temperature along the electric eld being much larger than the one in or-
thogonal direction. Therefore we nd that substantial part of total energy,
initially stored in the electric eld is converted after few oscillations into in-
ternal energy. This effect could not be obtained within the simple treatment
we used before, since all particles were assumed to have single momentum
(delta-function distribution in momentum space).
When particle interactions are taken into account photon interactions are
expected to bring this two-temperature system into thermal equilibrium.
3.1.4. Electron-positron plasma in GRBs and in cosmology
Analogy and difference between electron-positron plasma in the early Uni-
verse and in sources of GRBs are discussed in Appendix D. We focus on a)
dynamical differences, namely thermal acceleration of the outow in GRB
15
3. Brief description
sources vs. cosmological deceleration; b) nuclear composition differences as
synthesis of light elements in the early Universe and possible destruction of
heavy elements in GRB plasma; c) different physical conditions during last
scattering of photons by electrons in both cases leading to nearly perfect black
body spectrum of the microwave background radiation vs. non thermal spec-
trum of the photospheric emission in GRBs.
3.2. Photospheric emission from ultrarelativistic
outows
We study the photospheric emission from ultrarelativistic outows in Ap-
pendix E.
Two popular models of optically thick relativistic outows exist: the wind
and the shell. In this work the optical depth during the acceleration and
coasting phases is computed analytically within both models and its various
asymptotic limits are derived. In particular we show that quite counterin-
tuitively a geometrically thin shell may appear as a thick wind for photons
propagating inside it. For this reason we introduce notions of photon thick
and photon thin outows, which appear more general and better physically
motivated with respect to winds and shells.
We study the geometry of dynamic photospheres emerging from relativis-
tic outows. The photosphere of photon thin outow has always a convex
shape. In the photon thick one it is initially convex since there is always a
photon thin layer in any outow, and then it becomes concave. Assuming
that photons are emitted with comoving thermal spectrum at the moment
when the optical depth decreases to unity we compute the observed ux and
instantaneous spectra for both photon thick and photon thin outows. It is
our main nding that the photospheric emission from the photon thin out-
ow produces non thermal time integrated spectra, which may be described
by the Band function well known in the GRB literature.
We nd that energetic GRBs should produce photon thin outows and only
time integrated spectra may be observed fromsuch GRBs. In other words, ob-
served Band spectrum is obtained quite naturally from the comoving thermal
one by the integration over the photosphere.
3.3. Correlation dynamics in cosmology
Two fundamental processes are known to occur in a self-gravitating system
of collisionless massive particles: gravitational instability and violent relax-
ation. A new analytic approach is proposed in Appendix F aimed in describ-
ing these two apparently distinct phenomena as different manistations of
essentially the same physical process: gravitational structure formation. This
16
3.4. Neutrinos in cosmology
approach is based on application of two averaging schemes: spatial averag-
ing and coarse-graining. A master equation for spatially averaged coarse-
grained distribution function of dark matter is constructed and its limiting
cases are analyzed. Discussion of the related works, such as the recent work
of J. Einasto et al., (2011) discussing phase synchronization in the large scale
structure is presented.
3.4. Neutrinos in cosmology
Many observational facts make it clear that luminous matter alone cannot
account for the whole matter content of the Universe. Among them there
is the cosmic background radiation anisotropy spectrum, that is well tted
by a cosmological model in which just a small fraction of the total density is
supported by baryons.
In particular, the best t to the observed spectrum is given by a at CDM
model, namely a model in which the main contribution to the energy density
of the Universe comes from vacuum energy and cold dark matter. This result
is conrmed by other observational data, like the power spectrum of large
scale structures.
Another strong evidence for the presence of dark matter is given by the
rotation curves of galaxies. In fact, if we assume a spherical or ellipsoidal
mass distribution inside the galaxy, the orbital velocity at a radius r is given
by Newtons equation of motion. The peculiar velocity of stars beyond the
visible edge of the galaxy should then decrease as 1/r. What is instead ob-
served is that the velocity stays nearly constant with r. This requires a halo
of invisible, dark, matter to be present outside the edge. Galactic size should
then be extended beyond the visible edge. From observations is follows that
the halo radius is at least 10 times larger than the radius of visible part of the
galaxy. Then it follows that a halo is at least 10 times more massive than all
stars in a galaxy.
Neutrinos were considered as the best candidate for dark matter about
twenty years ago. Indeed, it was shown that if these particles have a small
mass m
>>
2 GeV. Such particles could also provide a large contribution into the energy
density of the Universe, in spite of much smaller value of number density.
Recent experimental results from laboratory (see Dolgov (2002) for a re-
view) rule out massive neutrinos with m
3
_
1/
R
S
, (3.4.1)
18
3.4. Neutrinos in cosmology
Figure 3.1.: Cellular structure of the Universe.
where R
S
is the sample size, = 3 D, and D is the Hausdorff dimension
of the fractal. Most challenging was the merging of the concepts of fractal,
Jeans mass of dark matter and the cellular structure in the Universe, ad-
vanced by Rufni et al. (1988). The cellular structure emerging from this
study is represented in Figure 3.1. There the upper cutoff in the fractal
structure R
cutoff
100 Mpc, was associated to the Jeans mass of the ino
M
cell
=
_
m
pl
m
ino
_
2
m
pl
.
3.4.3. Lepton asymmetry of the Universe
Lattanzi et al. (2005), Lattanzi et al. (2006) studied how the cosmological con-
straints on neutrino mass are affected by the presence of a lepton asymmetry.
The main conclusion is that while constraints on neutrino mass do not change
by the inclusion into the cosmological model the dimensional chemical po-
tential of neutrino, as an additional parameter, the value of lepton asymmetry
allowed by the present cosmological data is surprisingly large, being
L =
n
n
0.9. (3.4.2)
Therefore, large lepton asymmetry is not ruled out by the current cosmologi-
cal data.
19
3. Brief description
3.5. Semidegenerate self-gravitating system of
fermions as a model for dark matter halos and
universality laws
The problem of dark matter distribution in galactic halos has traditionally
been treated in the realm of newtonian physics in view of the low velocities of
the stars in the galaxies, like the simulations from Navarro, Frenk and White
(1997). In the meantime, phenomenological proles of dark matter have been
advanced by Einasto (1965); Burkert (1995), and universal properties of the
dark matter distribution have been inferred from dwarf galaxies and proba-
bly globular clusters all the way to very massive galaxies (Gentile et al., 2009;
Donato et al., 2009; Walker et al., 2010; Boyarsky et al., 2009). However, a
problem arises: while simulations like those from NFW point to a cusped
halo, observations from various types of galaxies seem to show cored halos
(Salucci et al., 2011). This discrepancy between theory and observations is not
yet fully understood, but could show a problem with the simulations done
so far.
Different approach to the study of properties of dark matter is coming from
cosmology and specially from theories of galaxy formation and evolution.
While studies of CMB allow to constrain number and masses of light neu-
trinos representing hot dark matter (Giusarma et al., 2011), properties and
masses of warm and cold dark matter are constrained by the total mass den-
sity of the Universe (Lee and Weinberg, 1977) and galactic halos structure
(Tremaine and Gunn, 1979) and substructure (Polisensky and Ricotti, 2011).
The Lee-Weinberg bound (Lee and Weinberg, 1977) limits the mass of dark
matter particles with given coupling constant from above implying that par-
ticle was in thermal equilibrium in early Universe. For typical weak inter-
action coupling G
F
mass is constrained to be more than 2 GeV/c
2
and
such particles is referred to as Weakly Interacting Massive Particles, WIMPs.
Bound by Tremaine and Gunn (1979) and its improvement by many authors
(Hogan and Dalcanton, 2000; de Vega and Sanchez, 2010) are based on as-
sumption of fermionic nature of dark matter and nondegeneracy of galactic
haloes of dwarf galaxies, that leads to lower limit on mass 0.41 keV/c
2
.
In a completely unrelated eld (as of yet), the physics of Active Galactic
Nuclei (AGN) and quasars has been recognized for more than 50 years as
dominated by relativistic gravitational effects of a black hole. The formation
of these black holes is not yet fully understood, although different black holes
mass estimates for AGNs and quasars show masses up to 10
10
M
= 0. Benetti et al. (2011) have instead the considered the possibility that
slow-roll is briey violated during ination; this naturally happens in theo-
ries with many interacting scalar elds, as it is the case, for example, in a class
of multield, supergravity-inspired models, where supersymmetry-breaking
phase transitions occur during ination. These phase transitions correspond
to sudden changes in the inaton effective mass and can be modeled as steps
in the inationary potential, leading in turn to the appearance of characterisc-
tic oscillatory features in the primordial perturbation spectrum. Thus CMB
observations, like those of the WMAP and ACT experiments, and in future
of the Planck satellite, can be used to constrain such models (Benetti et al.,
2011).
Models with a dynamical dark energy have been considered by Serra et al.
(2009). Models in which the fundamental constants are allowed to vary with
time were instead studied, also in the presence of dark energy, by Menegoni
et al. (2009); Martins et al. (2010); Menegoni et al. (2010); Menegoni (2010);
Calabrese et al. (2011). For details see Appendix H.
22
4. Publications
4.1. Publications before 2005
1. R. Rufni, D. J. Song, and L. Stella, On the statistical distribution of
massive fermions and bosons in a Friedmann universe Astronomy and
Astrophysics, Vol. 125, (1983) pp. 265-270.
The distribution function of massive Fermi and Bose particles in an expanding
universe is considered as well as some associated thermodynamic quantities,
pressure and energy density. These considerations are then applied to cosmo-
logical neutrinos. A new limit is derived for the degeneracy of a cosmological
gas of massive neutrinos.
2. R. Rufni and D. J. Song, On the Jeans mass of weakly interacting neu-
tral massive leptons, in Gamow cosmology, eds. F. Melchiorri and R.
Rufni, (1986) pp. 370385.
The cosmological limits on the abundances and masses of weakly interacting
neutral particles are strongly affected by the nonzero chemical potentials of
these leptons. For heavy leptons (m
x
> GeV), the value of the chemical po-
tential must be much smaller than unity in order not to give very high values
of the cosmological density parameter and the mass of heavy leptons, or they
will be unstable. The Jeans mass of weakly interacting neutral particles could
give the scale of cosmological structure and the masses of astrophysical ob-
jects. For a mass of the order 10 eV, the Jeans mass could give the scenario
of galaxy formation, the supercluster forming rst and then the smaller scales,
such as clusters and galaxies, could form inside the large supercluster.
3. D. Calzetti, M. Giavalisco, R. Rufni, J. Einasto, and E. Saar, The corre-
lation function of galaxies in the direction of the Coma cluster, Astro-
physics and Space Science, Vol. 137 (1987) pp. 101-106.
Data obtained by Einasto et al. (1986) on the amplitude of the correlation func-
tion of galaxies in the direction of the Coma cluster are compared with theo-
retical predictions of a model derived for a self-similar observer-homogeneous
structure. The observational samples can be approximated by cones of angu-
lar width alpha of about 77 deg. Eliminating sources of large observational
error, and by making a specied correction, the observational data are found
to agree very well with the theoretical predictions of Calzetti et al. (1987).
23
4. Publications
4. R. Rufni, D. J. Song, and S. Taraglio, The ino mass and the cellu-
lar large-scale structure of the universe, Astronomy and Astrophysics,
Vol. 190, (1988) pp. 1-9.
Within the theoretical framework of a Gamowcosmology with massive inos,
the authors show how the observed correlation functions between galaxies
and between clusters of galaxies naturally lead to a cellular structure for the
Universe. From the size of the elementary cells they derive constraints on
the value of the masses and chemical potentials of the cosmological inos.
They outline a procedure to estimate the effective average mass density of
the Universe. They also predict the angular size of the inhomogeneities to be
expected in the cosmological black body radiation as remnants of this cellular
structure. A possible relationship between the model and a fractal structure is
indicated.
5. D. Calzetti, M. Giavalisco, and R. Rufni, The normalization of the
correlation functions for extragalactic structures, Astronomy and As-
trophysics, Vol. 198 (1988), pp. 1-15.
It is shown that the spatial two-point correlation functions for galaxies, clus-
ters and superclusters depend explicitly on the spatial volume of the statistical
sample considered. Rules for the normalization of the correlation functions are
given and the traditional classication of galaxies into eld galaxies, clusters
and superclusters is replaced by the introduction of a single fractal structure,
with a lower cut-off at galactic scales. The roles played by randomand stochas-
tic fractal components in the galaxy distribution are discussed in detail.
6. M. V. Arbolino and R. Rufni, The ratio between the mass of the halo
and visible matter in spiral galaxies and limits on the neutrino mass,
Astronomy and Astrophysics, Vol. 192, (1988) pp. 107-116.
Observed rotation curves for galaxies with values of the visible mass ranging
over three orders of magnitude together with considerations involving equi-
librium congurations of massive neutrinos, impose constraints on the ratio
between the masses of visible and dark halo comporents in spiral galaxies.
Upper and lower limits are derived for the mass of the particles making up the
dark matter.
7. A. Bianconi, H. W. Lee, and R. Rufni, Limits from cosmological nu-
cleosynthesis on the leptonic numbers of the universe, Astronomy and
Astrophysics, Vol. 241 (1991) pp. 343-357.
Constraints on chemical potentials and masses of inos are calculated using
cosmological standard nucleosynthesis processes. It is shown that the elec-
tron neutrino chemical potential (ENCP) should not be greater than a value of
the order of 1, and that the possible effective chemical potential of the other
neutrino species should be about 10 times the ENCP in order not to conict
24
4.1. Publications before 2005
with observational data. The allowed region (consistent with the He-4 abun-
dance observations) is insensitive to the baryon to proton ratio , while those
imposed by other light elements strongly depend on .
8. R. Rufni, J. D. Salmonson, J. R. Wilson, and S.-S. Xue, On the pair
electromagnetic pulse of a black hole with electromagnetic structure,
Astronomy and Astrophysics, Vol. 350 (1999) pp. 334-343.
We study the relativistically expanding electron-positron pair plasma formed
by the process of vacuum polarization around an electromagnetic black hole
(EMBH). Such processes can occur for EMBHs with mass all the way up to
6 10
5
M
pairs,
is here considered in the special case of a Reissner-Nordstrom geometry. The
interaction of the PEM pulse with the baryonic matter is described using a sim-
plied model of a slab of constant thickness in the laboratory frame (constant-
thickness approximation) as well as performing the integration of the general
relativistic hydrodynamical equations. Te validation of the constant-thickness
approximation, already presented in a previous paper Rufni et al. (1999) for a
25
4. Publications
PEM pulse in vacuum, is here generalized to the presence of baryonic matter.
It is found that for a baryonic shell of mass-energy less than 1% of the total
energy of the dyadosphere, the constant-thickness approximation is in excel-
lent agreement with full general relativistic computations. The approximation
breaks down for larger values of the baryonic shell mass, however such cases
are of less interest for observed Gamma Ray Bursts (GRBs). On the basis of
numerical computations of the slab model for PEM pulses, we describe (i) the
properties of relativistic evolution of a PEM pulse colliding with a baryonic
shell; (ii) the details of the expected emission energy and observed tempera-
ture of the associated GRBs for a given value of the EMBH mass; 10
3
M
, and
for baryonic mass-energies in the range 10
8
to 10
2
the total energy of the
dyadosphere.
10. M. Lattanzi, R. Rufni, and G. Vereshchagin, On the possible role of
massive neutrinos in cosmological structure formation, in Cosmology
and Gravitation, eds. M. Novello and S. E. Perez Bergliaffa, Vol. 668 of
AIP Conference Series, (2003) pp. 263287.
In addition to the problem of galaxy formation, one of the greatest open ques-
tions of cosmology is represented by the existence of an asymmetry between
matter and antimatter in the baryonic component of the Universe. We believe
that a net lepton number for the three neutrino species can be used to under-
stand this asymmetry. This also implies an asymmetry in the matter-antimatter
component of the leptons. The existence of a nonnull lepton number for the
neutrinos can easily explain a cosmological abundance of neutrinos consistent
with the one needed to explain both the rotation curves of galaxies and the
atness of the Universe. Some propedeutic results are presented in order to
attack this problem.
4.2. Publications (2005 2010)
1. A.G. Aksenov, R. Rufni and G.V. Vereshchagin, Pair plasma relax-
ation time scales, Physical Review E, Vol. 81 (2010) 046401.
By numerically solving the relativistic Boltzmann equations, we compute the
time scale for relaxation to thermal equilibrium for an optically thick electron-
positron plasma with baryon loading. We focus on the time scales of elec-
tromagnetic interactions. The collisional integrals are obtained directly from
the corresponding QED matrix elements. Thermalization time scales are com-
puted for a wide range of values of both the total energy density (over 10 or-
ders of magnitude) and of the baryonic loading parameter (over 6 orders of
magnitude). This also allows us to study such interesting limiting cases as the
almost purely electron-positron plasma or electron-proton plasma as well as
intermediate cases. These results appear to be important both for laboratory
experiments aimed at generating optically thick pair plasmas as well as for
26
4.2. Publications (2005 2010)
astrophysical models in which electron-positron pair plasmas play a relevant
role.
2. R. Rufni, G.V. Vereshchagin and S.-S. Xue, Electron-positron pairs in
physics and astrophysics: from heavy nuclei to black holes Physics
Reports, Vol. 487 (2010) No 1-4, pp. 1-140.
From the interaction of physics and astrophysics we are witnessing in these
years a splendid synthesis of theoretical, experimental and observational re-
sults originating from three fundametal physical processes. They were origi-
nally proposed by Dirac, by Breit and Wheeler and by Sauter, Heisenberg, Eu-
ler and Schwinger. For almost seventy years they have all three been followed
by a continued effort of experimental verication on Earth-based experiments.
The Dirac process, e
+
e
annihilation to hadronic
channels, in addition to the Dirac electromagnetic channel; B) ongoing Earth
based experiments to detect electron-positron production in strong elds by
focusing coherent laser beams and by electron beam-laser collisions; and C) the
multiyear attempts to detect electron-positron production in Coulomb elds
for a large atomic number Z > 137 in heavy ion collisions. These attempts
follow the classical theoretical work of Popov and Zeldovich, and Greiner and
their schools. We then turn to astrophysics. We rst review the basic work
on the energetics and electrodynamical properties of an electromagnetic black
hole and the application of the Schwinger formula around Kerr-Newman black
holes as pioneered by Damour and Rufni. We only focus on black hole masses
larger than the critical mass of neutron stars, for convenience assumed to coin-
cide with the Rhoades and Rufni upper limit of 3.2M
8
is also considerably lower providing a better consistency with a low ampli-
tude of the Sunyaev-Zeldovich signal.
9. S. Pandol, A.Cooray, E.Giusarma, E.W.Kolb, A.Melchiorri, O.Mena
and P.Serra, Harrison-Zeldovich primordial spectrum is consistent
with observations, Phys. Rev. D 81, 123509 (2010).
Abstract: Ination predicts primordial scalar perturbations with a nearly scale-
invariant spectrum and a spectral index approximately unity (the Harrison
Zeldovich (HZ) spectrum). The rst important step for inationary cosmol-
ogy is to check the consistency of the HZ primordial spectrum with current
observations. Recent analyses have claimed that a HZ primordial spectrum is
excluded at more than 99% c.l.. Here we show that the HZ spectrum is only
marginally disfavored if one considers a more general reionization scenario.
Data from the Planck mission will settle the issue.
10. P. Serra, A. Cooray, D. E. Holz, A. Melchiorri, S. Pandol, and D. Sarkar,
No evidence for dark energy dynamics from a global analysis of cos-
mological data, Phys. Rev. D 80, 121302 (2009).
Abstract: We use a variant of principal component analysis to investigate the
possible temporal evolution of the dark energy equation of state, w(z). We
constrain w(z) in multiple redshift bins, utilizing the most recent data from
Type Ia supernovae, the cosmic microwave background, baryon acoustic oscil-
lations, the integrated Sachs-Wolfe effect, galaxy clustering, and weak lensing
data. Unlike other recent analyses, we nd no signicant evidence for evolving
dark energy; the data remains completely consistent with a cosmological con-
stant. We also study the extent to which the time-evolution of the equation of
31
4. Publications
state would be constrained by a combination of current- and future-generation
surveys, such as Planck and the Joint Dark Energy Mission.
11. E. Menegoni, S. Pandol, S. Galli, M. Lattanzi, A. Melchiorri Con-
straints on the dark energy equation of state in presence of a varying
ne structure constant in Int. J. Mod. Phys D19, 507 (2010).
We discuss the cosmological constraints on the dark energy equation of state
in the pres- ence of primordial variations in the ne structure constant. We
nd that the constraints from CMB data alone on w and the Hubble constant
are much weaker when variations in the ne structure constant are permitted.
Vice versa, constraints on the ne struc- ture constant are relaxed by more
than 50% when dark energy models different from a cosmological constant are
considered.
12. C.J.A.P. Martins, E. Menegoni, S. Galli and A. Melchiorri, Varying cou-
plings in the early universe: correlated variations of and G, Physical
Review D 82 023532 (2010)
The cosmic microwave background anisotropies provide a unique opportu-
nity to constrain simultaneous variations of the ne-structure constant and
Newtons gravitational constant G. Those correlated variations are possible
in a wide class of theoretical models. In this brief paper we show that the
current data, assuming that particle masses are constant, give no clear indi-
cation for such variations, but already prefer that any relative variations in
should be of the same sign of those of G for variations of 1%. We also show
that a cosmic complementarity is present with big bang nucleosynthesis and
that a combination of current CMB and big bang nucleosynthesis data strongly
constraints simultaneous variations in and G. We nally discuss the future
bounds achievable by the Planck satellite mission.
13. E. Menegoni, New Constraints on Variations of Fine Structure Con-
stant from Cosmic Microwave Background Anisotropies, GRAVITA-
TIONAL PHYSICS: TESTING GRAVITY FROM SUBMILLIMETER TO
COSMIC: Proceedings of the VIII Mexican School on Gravitation and
Mathematical Physics. AIP Conference Proceedings, Volume 1256, pp.
288-292 (2010).
The recent measurements of Cosmic Microwave Background temperature and
polarization anisotropy made by the ACBAR, QUAD and BICEP experiments
substantially improve the cosmological constraints on possible variations of
the ne structure constant in the early universe. In this work I analyze this
recent data obtaining the constraint /0 = 0.987+/-0.012 at 68% c.l.. The in-
clusion of the new HST constraints on the Hubble constant further increases
the bound to /0 = 1.001+/-0.007 at 68% c.l., bringing possible deviations
from the current value below the 1% level.
32
4.2. Publications (2005 2010)
14. A. Melchiorri, F. De Bernardis, E. Menegoni, Limits on the neutrino
mass fromcosmology. GRAVITATIONAL PHYSICS: TESTINGGRAV-
ITY FROM SUBMILLIMETER TO COSMIC: Proceedings of the VIII
Mexican School on Gravitation and Mathematical Physics. AIP Con-
ference Proceedings, Volume 1256, pp. 96-106 (2010).
We use measurements of luminosity-dependent galaxy bias at several different
redshifts, SDSS at z = 0.05, DEEP2 at z = 1 and LBGs at z = 3.8, combined with
WMAP ve-year cosmic microwave background anisotropy data and SDSS
Red Luminous Galaxy survey three-dimensional clustering power spectrum
to put constraints on cosmological parameters.
15. A.G. Aksenov, R. Rufni and G.V. Vereshchagin, Thermalization of the
mildly relativistic plasma, Physical Review D, Vol. 79 (2009) 043008.
In the recent Letter Aksenov et al. (2007) we considered the approach of
nonequilibrium pair plasma towards thermal equilibrium state adopting a ki-
netic treatment and solving numerically the relativistic Boltzmann equations.
It was shown that plasma in the energy range 0.1-10 MeV rst reaches kinetic
equilibrium, on a timescale t
k
10
14
sec, with detailed balance between
binary interactions such as Compton, Bhabha and Mller scattering, and pair
production and annihilation. Later the electron-positron-photon plasma ap-
proaches thermal equilibrium on a timescale t
th
10
12
sec, with detailed
balance for all direct and inverse reactions. In the present paper we system-
atically present details of the computational scheme used in Aksenov et al.
(2007), as well as generalize our treatment, considering proton loading of
the pair plasma. When proton loading is large, protons thermalize rst by
proton-proton scattering, and then with the electron-positron-photon plasma
by proton-electron scattering. In the opposite case of small proton loading
proton-electron scattering dominates over proton-proton one. Thus in all cases
the plasma, even with proton admixture, reaches thermal equilibrium cong-
uration on a timescale t
th
10
11
sec. We show that it is crucial to account
for not only binary but also triple direct and inverse interactions between elec-
trons, positrons, photons and protons. Several explicit examples are given and
the corresponding timescales for reaching kinetic and thermal equilibria are
determined.
16. A. G. Aksenov, R. Rufni, and G. V. Vereshchagin, Thermalization of
pair plasma with proton loading in the Proceedings of PROBING
STELLAR POPULATIONS OUT TO THE DISTANT UNIVERSE meet-
ing, AIP Conference Proceedings 1111 (2009) 344-350.
We study kinetic evolution of nonequilibrium optically thick electron-positron
plasma towards thermal equilibrium solving numerically relativistic Boltz-
mann equations with energy per particle ranging from 0.1 to 10 MeV. We gen-
eralize our results presented in Aksenov et al. (2007), considering proton load-
ing of the pair plasma. Proton loading introduces new characteristic timescales
33
4. Publications
essentially due to proton-proton and proton-electron Coulomb collisions. Tak-
ing into account not only binary but also triple direct and inverse interactions
between electrons, positrons, photons and protons we show that thermal equi-
librium is reached on a timescale t
th
10
11
sec.
17. M. Lattanzi, J. Silk Can the WIMP annihilation boost factor be boosted
by the Sommerfeld enhancement? , in Phys. Rev. D79, 083523 (2009).
We demonstrate that the Sommerfeld correction to cold dark matter (CDM)
annihilations can be appreciable if even a small component of the dark matter
is extremely cold. Subhalo substructure provides such a possibility given that
the smallest clumps are relatively cold and contain even colder substructure
due to incomplete phase space mixing. Leptonic channels can be enhanced
for plausible models and the solar neighbourhood boost required to account
for PAMELA/ATIC data is plausibly obtained, especially in the case of a few
TeV mass neutralino for which the Sommerfeld-corrected boost is found to
be 10
4
10
5
. Saturation of the Sommerfeld effect is shown to occur below
10
4
, thereby making this result largely independent on the presence of
substructures below 10
5
M
(z =
0) < 0.43 (0.28) eV at 95% C.L. for the case in which the neutrino mass was
lighter (heavier) in the past, a result competitive with the ones imposed for
standard (i.e., constant mass) neutrinos. Moreover, for the ratio of the mass
variation of the neutrino mass m
[/m
0
] < 1.3 (2.7) at 95% C.L. for m
< 0 (m
F
E
F
m
e
c
2
,
where E
F
is Fermi energy, corresponding to Fermi momentum
p
F
= h
_
3n
8
_1
3
.
For relativistic gas of electrons and positrons E
2
= p
2
c
2
+ m
2
e
c
4
and
F
=
_
_
h
m
e
c
_
2
_
3n
8
_2
3
+ 1
_
1/2
1.
In this way the degeneracy parameter
D =
F
< 1 (A.1.2)
determines the temperature where relativistic degeneracy occurs. Thermal
electron-positron plasma becomes degenerate at kT 3m
e
c
2
. It should be
noted that the average occupation numbers for kinetic equilibrium state are
not high even in the ultrarelativistic limit with chemical potential = 0: 8.7 %
for electrons and positrons and 36.8 % for photons.
In this work we extend the previous results on thermalization timescale
of pair plasma, including exact QED treatment of three-particle interaction
and quantum corrections to collision integrals by the corresponding Pauli
blocking and Bose enhancement factors. In section A.2 relativistic Boltzmann
equation with quantum corrections is introduced for both two- and three-
particle interactions. In section 3 the details od adopted numerical scheme
are given. In section 4 we report our results. Discussion and conclusions
follow.
58
A.2. Boltzmann equation with QED collisional integrals
A.2. Boltzmann equation with QED collisional
integrals
In uniform isotropic pair plasma relativistic Boltzmann equation for distribu-
tion function f
(t) =
_
f
(p, t)d
3
p, (A.2.1)
where n
(p, t) =
q
_
(p, t)
_
, (A.2.2)
where the sum is taken over all two- and three-particle reactions q,
q
and
q
a
+
b
=
i
+
j
, p
a
+p
b
= p
i
+p
j
. (A.2.3)
The differential emission and absorption coefcients for this reaction are
(p
a
, t)d
3
p
a
=
(p
b
, t)d
3
p
b
=
(p
i
, t)d
3
p
i
=
(p
j
, t)d
3
p
j
=
d
3
p
i
d
3
p
j
d
3
p
a
d
3
p
b
W
p
a
,p
b
;p
i
,p
j
f
(p
a
, t) f
(p
b
, t)
_
1 f
(p
i
, t)/h
3
_ _
1 f
(p
j
, t)/h
3
_
, (A.2.4)
where W is the transition function and
_
1 f (p, t)/h
3
_
are respectively Bose
enhancement (sign +) and Pauli blocking (sign ) factors (Uehling and
59
A. Bose enhancement and Pauli blocking in the pair plasma
Uhlenbeck, 1933; Uehling, 1934).
In a specic case when the particles and undergo scattering from initial
states a and b into new states i and j, respectively, the emission coefcient of
particle is
(p
i
, t) =
_
d
3
p
j
d
3
p
a
d
3
p
b
W
p
a
,p
b
;p
i
,p
j
f
(p
a
, t) f
(p
b
, t)
_
1 f
(p
i
, t)/h
3
_ _
1 f
(p
j
, t)/h
3
_
. (A.2.5)
Considering the inverse process, corresponding to exchange of indices a i,
b j, we obtain the absorption coefcient
(p
i
, t) =
_
d
3
p
j
d
3
p
a
d
3
p
b
W
p
i
,p
j
;p
a
,p
b
f
(p
i
, t) f
(p
j
, t)
_
1 f
(p
a
, t)/h
3
_ _
1 f
(p
b
, t)/h
3
_
. (A.2.6)
Then Boltzmann equation for particle reads
1
c
d
dt
f
(p
i
, t) =
_
d
3
p
j
d
3
p
a
d
3
p
b
_
W
p
a
,p
b
;p
i
,p
j
f
(p
a
, t) f
(p
b
, t)
_
1 f
(p
i
, t)/h
3
_ _
1 f
(p
j
, t)/h
3
_
W
p
i
,p
j
;p
a
,p
b
f
(p
i
, t) f
(p
j
, t)
_
1 f
(p
a
, t)/h
3
_ _
1 f
(p
b
, t)/h
3
_
_
.
(A.2.7)
Taking into account the detailed balance condition W
p
a
,p
b
;p
i
,p
j
= W
p
i
,p
j
;p
a
,p
b
,
classical form of Boltzmann equation is recovered
1
c
d
dt
f
(p
i
, t) =
_
d
3
p
j
d
3
p
a
d
3
p
b
W
p
a
,p
b
;p
i
,p
j
_
f
(p
a
, t) f
(p
b
, t)
_
1 f
(p
i
, t)/h
3
_ _
1 f
(p
j
, t)/h
3
_
f
(p
i
, t) f
(p
j
, t)
_
1 f
(p
a
, t)/h
3
_ _
1 f
(p
b
, t)/h
3
_
_
. (A.2.8)
The Boltzmann equation for particle is obtained in complete analogy.
When the process is not a scattering, but, for instance, pair annihilation into
two photons, the Boltzmann equation is identical to (A.2.8). In the case of
Mller scattering of only one specie (electron or positron) there is additional
1/2 multiplier in front of the integral due to the presence of two indistin-
guishable particles before and after the interaction.
60
A.2. Boltzmann equation with QED collisional integrals
A.2.2. Three-particle interactions
Consider now three-particle interactions. Let two incoming particles of
species and have momenta in the phase space volumes d
3
p
a
around p
a
and d
3
p
b
around p
b
, respectively. Three outgoing particles of species , ,
and have momenta in the phase space volumes d
3
p
i
around p
i
, d
3
p
j
around
p
j
, and d
3
p
k
around p
k
, respectively.
The energy and momentum conservation read
a
+
b
=
i
+
j
+
k
, p
a
+p
b
= p
i
+p
j
+p
k
. (A.2.9)
The differential emission and absorption coefcients for this reaction are
(p
i
, t)d
3
p
i
=
(p
j
, t)d
3
p
j
=
(p
k
, t)d
3
p
k
=
(p
a
, t)d
3
p
a
=
(p
b
, t)d
3
p
b
=
d
3
p
a
d
3
p
b
d
3
p
i
d
3
p
j
d
3
p
k
W
p
a
,p
b
;p
i
,p
j
,p
k
f
(p
a
, t) f
(p
b
, t)
_
1 f
(p
i
, t)/h
3
_ _
1 f
(p
j
, t)/h
3
_ _
1 f
(p
k
, t)/h
3
_
, (A.2.10)
and for the inverse one
(p
i
, t)d
3
p
i
=
(p
j
, t)d
3
p
j
=
(p
k
, t)d
3
p
k
=
(p
a
, t)d
3
p
a
=
(p
b
, t)d
3
p
b
=
d
3
p
a
d
3
p
b
d
3
p
i
d
3
p
j
d
3
p
k
W
p
i
,p
j
,p
k
;p
a
,p
b
f
(p
i
, t) f
(p
j
, t) f
(p
k
, t)
_
1 f
(p
a
, t)/h
3
_ _
1 f
(p
b
, t)/h
3
_
. (A.2.11)
Consequently Boltzmann equation for particle specie reads
1
c
d
dt
f
(p
a
, t) =
_
d
3
p
i
d
3
p
j
d
3
p
k
d
3
p
b
_
W
p
a
,p
b
;p
i
,p
j
,p
k
f
(p
a
, t) f
(p
b
, t)
_
1 f
(p
i
, t)/h
3
_ _
1 f
(p
j
, t)/h
3
_ _
1 f
(p
k
, t)/h
3
_
+W
p
i
,p
j
,p
k
;p
a
,p
b
f
(p
i
, t) f
(p
j
, t) f
(p
k
, t)
_
1 f
(p
a
, t)/h
3
_ _
1 f
(p
b
, t)/h
3
_
_
. (A.2.12)
It may happen that some particle specie is present among both incoming
and outgoing particles, for instance coincides with . There are the fol-
lowing emission and absorption terms for this particle specie: absorption of
61
A. Bose enhancement and Pauli blocking in the pair plasma
incoming particle and emission of outgoing particle , corresponding to
the direct reaction; and in addition emission of outgoing particle and ab-
sorption of incoming particle , corresponding to the inverse reaction. Since
particles and are of the same specie, the collisional integral for that parti-
cle consists of four terms instead of two. Formally two additional terms may
be obtained by interchanging particle states a i. Boltzmann equation in
this case reads
1
c
d
dt
f
(p
a
, t) =
_
d
3
p
i
d
3
p
j
d
3
p
k
d
3
p
b
_
W
p
a
,p
b
;p
i
,p
j
,p
k
f
(p
a
, t) f
(p
b
, t)
_
1 f
(p
i
, t)/h
3
_ _
1 f
(p
j
, t)/h
3
_ _
1 f
(p
k
, t)/h
3
_
+W
p
i
,p
j
,p
k
;p
a
,p
b
f
(p
i
, t) f
(p
j
, t) f
(p
k
, t)
_
1 f
(p
a
, t)/h
3
_ _
1 f
(p
b
, t)/h
3
_
+W
p
i
,p
b
;p
a
,p
j
,p
k
f
(p
i
, t) f
(p
b
, t)
_
1 f
(p
a
, t)/h
3
_ _
1 f
(p
j
, t)/h
3
_ _
1 f
(p
k
, t)/h
3
_
W
p
a
,p
j
,p
k
;p
i
,p
b
f
(p
a
, t) f
(p
j
, t) f
(p
k
, t)
_
1 f
(p
i
, t)/h
3
_ _
1 f
(p
b
, t)/h
3
_
_
. (A.2.13)
Generally speaking, such four terms should be present in collisional inte-
gral for a particle specie which is present both among incoming and outgoing
particles, unless the process is a scattering. This state is valid for arbitrary
number of incoming and outgoing particles. It is not limited to the case of
quantum electrodynamics but applies to kinetic theory in general.
As for QEDprocesses considered in this work the corresponding collisional
integrals in many three-particle interactions contain four terms instead of
two, as can be seen from table A.1. It should be noted that the equilibrium
conditions may be obtained by only two terms in collisional integrals with-
out interchanging the states a and i. However, due to different structure of all
four emission and absorption coefcients their presence is essential in Boltz-
mann equation (A.2.13).
A.3. The numerical scheme
The main difculty arising from computing collisional integrals is that the
rate of particle emission/absorbtion now depends not only on the distribu-
tion functions of incoming particles, but also on the ones of outgoing par-
62
A.3. The numerical scheme
Two-particle interactions Three-particle interactions
Compton scattering Double Compton
e
e
/
/
e
e
/
//
Coulomb, Mller and Bhabha scattering Bremmstrahlung
e
1
e
2
e
/
1
e
/
2
e
1
e
2
e
/
1
e
/
2
e
+
e
e
+/
e
/
e
+
e
e
+/
e
/
2
e
+
e
3
Pair creation/annihilation
2
e
+
e
/
e
e
/
e
+
e
i
d
i
, dn
i
= f
i
dp
i
, (A.3.2)
while the corresponding energy density is
i
=
_
i
f
i
dp
i
=
_
E
i
d
i
.
In these variables the Boltzmann equations (A.2.2) read
1
c
dE
i
dt
=
q
(
q
i
q
i
E
i
), (A.3.3)
where
q
i
= (4
3
i
i
/c
3
)
q
i
.
To obtain emission and absorbtion coefcients the computational grids are
introduced in the phase space
i
,
i
,
i
, where
i
= cos
i
,
i
and
i
are
angles in spherical coordinates of particle momentum space p
i
. The zone
boundaries are
i,1/2
,
k1/2
,
l1/2
for 1
max
, 1 k k
max
,
1 l l
max
. The length of the i-th interval is
i,
i,+1/2
i,1/2
. On
63
A. Bose enhancement and Pauli blocking in the pair plasma
the nite grid the functions (A.3.1) become
E
a
= E
i,
1
i,
_
i,
d E
i
(), (A.3.4)
where for simplication of formulae we use collective indices a = i, .
The collisional integrals in (A.3.3) are replaced by the corresponding sums.
When particles are treated classically we have for time derivative of each vari-
able the following expression
E
a
=
b,c
A
(b,c[a,d)
E
b
E
c
b,c
B
(a,b[c,d)
E
a
E
b
, (A.3.5)
where rst sum on the right side is for emission in reaction b +c a +d and
second is for absorbtion in reaction a + b c + d. There is no third summa-
tion (by index d) because of delta-function in the initial integrals originating
from the energy conservation. This can be effectively rewritten as just one
sum
E
a
=
b,c
A
a
b,c
E
b
E
c
, (A.3.6)
A
a
b,c
= A
(b,c[a,d)
c
a
e
B
(a,b[e,d)
, (A.3.7)
and this sum can be found by direct computation.
When the quantum statistics effects are included we have instead
E
a
=
b,c
(1 E
a
/E
a
)(1 E
d
/E
d
)A
(b,c[a,d)
E
b
E
c
b,c
(1 E
d
/E
d
)(1 E
c
/E
c
)B
(a,b[c,d)
E
a
E
b
, (A.3.8)
which E
a
is the critical spectral energy density for which quantum statistics
correction becomes relevant (corresponding to occupation numbers equal to
unity)
E
a
=
_
a
g
d
3
p
h
3
, (A.3.9)
where g is the number of helicity states for the particle. It turns out that while
the sums on the right-hand side of (A.3.8) can be reduced to one sum only,
but due to different structure of Bose enhancement and Pauli blocking factors
the numerical scheme based on the resulting expression will not be optimal.
Instead noticing that the phase space blocking/enhancement factors (1
E
d
/E
d
)(1 E
e
/E
e
) are the same for all four particles involved in the process
b + c d + e (a is one of the b, c, d, e), the corresponding parts of collisional
integrals arising in the above-mentioned sums can be computed only once
64
A.4. Results
instead of four times. As a result it is convenient not to x a and sum over all
possible b, c, as in (A.3.8), which we refer to as particle-oriented approach,
but instead sum over all possible reactions, which we refer to as reaction-
oriented approach. It means that at each step of calculations we x b, c and
for all possible reaction results the emission rates of outcomes d and e and the
absorbtion rates of incomes b, c are added to array of derivatives
E
a
. This ap-
proach considerably reduces the computational time and memory consump-
tion.
In our method exact energy and number of particles conservation laws are
satised, as we adopt interpolation of grid functions E
a
inside the energy
intervals. The number of energy intervals is 15, while internal grid of angles
has 32 points in
i
and 64 in
i
. Most time-consuming part of the numerical
solution is initial integration over particle angles.
A.4. Results
We solved numerically Boltzmann equation (A.2.2) in two cases. Initially
only photons are present with constant spectral energy density and total en-
ergy density = 10
23
erg/cm
3
and = 10
29
erg/cm
3
. Such energy densities
corresponds to the temperature in thermal equilibrium of 0.3 and 8, respec-
tively.
In Figs. A.1 and A.2 we present number density, energy density, tempera-
ture and chemical potential of photons and pairs in both cases. We also show
the difference between quantum and Boltzmann statistics by including and
omitting the Pauli blocking and Bose enhancement factors in evolution equa-
tions (A.3.8). Time is expressed in units of Compton time
C
=
1
T
n
c
, (A.4.1)
where n
3
3 2 1 0 1 2 3
0.0
0.2
0.4
0.6
0.8
1.0
log
10
t
C
,
1
0
2
3
e
r
g
c
m
3
3 2 1 0 1 2 3
0.0
0.2
0.4
0.6
0.8
1.0
1.2
log
10
t
C
3 2 1 0 1 2 3
14
12
10
8
6
4
2
0
log
10
t
C
C
. Thus the ratio
of the timescales of kinetic and thermal equilibrium is no longer but larger.
This indicates the necessity of exact treatment of three-particle interactions,
especially for high energy densities.
In Figs. A.3 and A.4 we show spectral evolution for both our initial con-
ditions. The nal spectra shown for t = 10
3
C
are in good agreement with
Planck/Fermi-Dirac distribution functions, correspondingly, obtained for the
given energy density, typically within 5 % accuracy. Notice that at the Comp-
ton time both electron/positron and photon spectra are far from equilibrium
shape, with the only exception of leptonic spectrum for = 10
23
erg/cm
3
.
This quick relaxation of leptonic component is due to large Coulomb loga-
rithm for non-relativistic temperatures.
A.5. Conclusions
In this work we consider relaxation of nonequilibrium optically thick pair
plasma to complete thermal equilibrium by integrating numerically relativis-
66
A.5. Conclusions
3 2 1 0 1 2 3
0
1
2
3
4
5
6
log
10
t
C
n
,
1
0
3
3
c
m
3
3 2 1 0 1 2 3
0.0
0.2
0.4
0.6
0.8
1.0
log
10
t
C
,
1
0
2
9
e
r
g
c
m
3
3 2 1 0 1 2 3
0
5
10
15
20
log
10
t
C
3 2 1 0 1 2 3
6
4
2
0
log
10
t
C
Figure A.2.: The same as in Fig. A.1, but for total energy density =
10
29
erg/cm
3
.
0.1 0.2 0.5 1.0 2.0
10
26
10
27
10
28
10
26
10
27
10
28
10
29
10
26
10
27
10
28
10
26
10
27
10
28
10
29
,
c
m
3
d
,
c
m
3
0.1 0.2 0.5 1.0 2.0
10
26
10
27
10
28
10
26
10
27
10
28
10
29
10
26
10
27
10
28
10
26
10
27
10
28
10
29
,
c
m
3
d
,
c
m
3
Figure A.3.: Numerical spectral energy densities of photons (black line) and
pairs (red line) at t =
C
(left) and at t = 10
3
C
(right) for = 10
23
erg/cm
3
.
Thick curves show the corresponding Bose-Einstein and Fermi-Dirac distri-
butions with the same number and energy densities, respectively. Dashed
thin line shows initial photon spectrum.
67
A. Bose enhancement and Pauli blocking in the pair plasma
2 5 10 20 50 100
10
30
10
31
10
32
10
30
10
31
10
32
10
33
10
30
10
31
10
32
10
30
10
31
10
32
10
33
,
c
m
3
d
,
c
m
3
2 5 10 20 50 100
10
30
10
31
10
32
10
30
10
31
10
32
10
33
10
30
10
31
10
32
10
30
10
31
10
32
10
33
,
c
m
3
d
,
c
m
3
Figure A.4.: The same as in Fig. A.3 for = 10
29
erg/cm
3
.
tic Boltzmann equations with exact QED two-particle and three-particle colli-
sional integrals. Quantum nature of particle statistics is accounted for in col-
lisional integrals by the corresponding Bose enhancement and Pauli blocking
factors.
We point out that unlike classical Boltzmann equation for binary interac-
tions such as scattering, more general interactions are typically described by
four collisional integrals for each particle that appears both among incoming
and outgoing particles.
Our numerical results indicate that the rates of three-particle interactions
become comparable to those of two-particle ones for temperatures exceeding
the electron rest-mass energy. Thus three particle interactions such as rela-
tivistic bremsstrahlung, double Compton scattering and radiative pair cre-
ation become essential not only for establishment of thermal equilibrium, but
also for correct estimation of interaction rates, energy losses etc.
68
B. Dynamics and emission from
mildly relativistic plasma
B.1. Introduction
Electron-positron plasma are expected to be present in compact astrophysi-
cal objects (Damour and Rufni, 1975), (Alcock et al., 1986), (Usov, 1998) and
plays crucial role in the Gamma-Ray Burst (GRB) phenomenon (Rufni et al.,
2009b), (Rufni et al., 2010). Electron-positron pairs are detected in the center
of our Galaxy (Churazov et al., 2005), leaving their characteristic imprint in
the observed radiation spectra. The observed annihilation line 511 keV near
the Earth indicates that pair plasma can be created in the upper atmosphere
at the thunderstorm (Briggs et al., 2011). Also interest in electron-positron
plasmas is due to the exciting possibility of generating such plasmas in lab-
oratory facilities using ultraintense focused short laser pulses, (Rufni et al.,
2010; Blaschke et al., 2006; Myatt et al., 2009; Thoma, 2009; Chen et al., 2009;
Wilks et al., 1992).
Electrons-positron plasmas in the laboratory experiment or in thunder-
storms have low density and are consequenly optically thin (Mustafa and
K ampfer, 2009; Katz, 2000). In contrast, in GRB source pair plasma is dense
and optically thick (Rufni et al., 2009b; Goodman, 1986; Piran, 1999). Such
optically thick plasma is expected to expand, cool down and eventually be-
come transparent for radiation, producing the characteristic ash of quasi-
thermal radiation, the so called photospheric emission. Many recent works
are devoted to study of this phenomenon, see e.g. (Beloborodov, 2011), (Peer
and Ryde, 2011), (Toma et al., 2011).
For the simulation of the photospheric emission of expanding plasma one
can use suitable Monte Carlo techniques based on well known reaction rates
(Bisnovatyi-Kogan et al., 1971; Lightman, 1982; Svensson, 1982; Guilbert and
Stepney, 1985). While in the optically thick case usually hydrodynamic ap-
proach is postulated (Piran, 1999). Such approximation is justied for large
optical depth, or in the beginning of expansion. Due to complexity of the cal-
culations only very few work adopt the kinetic approach for the description
of the plasma and try to calculate the spectra when the optical depth is not
very far from one (Beloborodov, 2010).
In this paper we report on our study of mildly relativistic plasma, which
is initially optically thick. The description of plasma is based on relativistic
69
B. Dynamics and emission from mildly relativistic plasma
Boltzmann equations. By means of this instrument we verify the applica-
tion of hydrodynamic description of the plasma. We focus on the difference
between the hydrodynamic description and more detailed kinetic one. In
particular, we consider the optical depth, the number density and average
particle energy evolution with time. We also report photon spectra from this
expanding electron-positron plasma, which allows us to describe the photo-
spheric emission in great details.
B.2. Formulation of the problem
Table B.1.: Physical Processes Included in Simulations.
Binary interactions Radiative and pair producing variants
Mller and Bhabha scattering Bremsstrahlung
e
1
e
2
e
/
1
e
/
2
e
1
e
2
e
/
1
e
/
2
e
e
/
e
/
e
e
/
e
/
/
e
e
/
//
Pair production Radiative pair production
and annihilation and three photon annihilation
/
e
/
e
//
e
//
e
e
/
e
e
//
We use the relativistic Boltzmann equations for the e
, ),
satises (Mihalas and Mihalas, 1984)
1
c
f
i
t
+
i
_
f
i
r
+
1
2
r
f
i
_
=
q
(
q
i
q
i
f
i
), (B.2.1)
where = (rp)/([r[[p[),
i
= v
i
/c =
_
1 (m
i
c
2
/
i
)
2
,
i
= c
_
p
2
+ (m
i
c
2
).
q
i
is the emission coefcient for the production of a particle of type i via
the physical process by q, and
q
i
is the corresponding absorption coefcient.
The summation runs over all considered physical processes that involve a
particle of type i. We take into account all two-body interactions, see e.g.
Berestetskii et al. (1982), Greiner and Reinhardt (2003), and corresponding
triple interactions(Svensson, 1984) written in Table B.1.
The particle concentrations and energy densities are
n
i
(r, t) =
_
f
i
(p, r, t)dp,
i
(r, t) =
_
f
i
(p, r, t)dp. (B.2.2)
70
B.3. Computational method
For the GRB source plasma, for example, the corresponding temperatures
in thermal equilibrium are (Aksenov et al., 2007)
0.1 T
th
10 MeV. (B.2.3)
At such conditions the number of particles in Debye sphere is large. We can
use one-particle distributions functions. Another important property is that
such pair plasma can not generate additional particles like neutrinos.
B.3. Computational method
To make the scheme conservative we use, instead of f
i
, the spectral energy
densities
E
i
(
i
, , r, t) =
2
3
i
i
f
i
c
3
, (B.3.1)
because the energy density in our phase space (, , r, t) is
i
f
i
drdp =
2
3
i
f
i
c
3
rd
i
d = E
i
rd
i
d. (B.3.2)
For E
i
the Boltzmann equation has the conservative form
1
c
E
i
t
+
r
2
r
(r
2
i
E
i
) +
1
r
[(1
2
)
i
E
i
)] =
q
(
q
i
q
i
f
i
). (B.3.3)
We use the lines method to solve above task (Aksenov et al., 2004). We in-
troduced grid in phase space
+1/2
,
k+1/2
, r
j+1/2
. After the replacing of all
derivatives except the derivative over time in Eq. (B.3.3) by nite differences
(Aksenov et al., 2004) and collisional integrals by sums (Aksenov et al., 2009)
we have the set of ordinary differential equations for grid values
E
i,,k,j
_
k
(r
3
j
/3)
ddd(r
3
/3)E
i
(, , r)
k
(r
3
j
/3)
. (B.3.4)
To carry out numerical evolution in the optically thick regions we use im-
plicit Gears method to solve the stiff system of ordinary differential equa-
tions (Hall and Watt, 1976).
We evaluate numerically exact expressions for collisional integrals with
matrix elements from QED (Aksenov et al., 2009). Our investigations shows
the timescale of triple interaction is
1
137 times larger, compared to
the timescale of binary interactions. Consequently plasma reaches so called
kinetic equilibrium with the common temperature but nonnull chemical po-
tentials before triple interactions become important. This fact allows us to
71
B. Dynamics and emission from mildly relativistic plasma
simplify calculations of triple interactions in the nondegenerate cases as de-
scribed in Aksenov et al. (2009).
For very large optical depths reactions rates cannot be resolved with a
rough angle grid. For this reason we adopt the following method for comput-
ing reaction rates in a such region 1. First we compute hydrodynamic
velocity from the known distribution functions, and transform our quanti-
ties in the comoving frame by using the following Lorentz transformations,
see e.g. Mihalas and Mihalas (1984), p. 414
(, , , r, t) =
_
_
_
_
/
+
V
c
cp
/
/
_
,
p
/
/
+
V
c
/
c
_
_
/
c
+
V
c
p
/
/
_
2
m
2
c
2
2
,
/
, r
/
, t
/
_
_
_
_
.
Then we calculate average values n
/
i
in the comoving frame, which do not
depend on angles. Then in comoving frame we evaluate average absorption
coefcients
/
= const. The emission coefcients
/
in the comoving frame
are taken to be proportional to equilibrium intensities. Finally we transform
the emission and absorption coefcients back into the laboratory frame by
I(, ) =
2
p
/2
p
/
I
/
(
/
,
/
), E(, , r, t)
2
3
f
c
3
I,
(, ) =
p
/
p
/
/
(
/
,
/
), (, ) =
/
/
(
/
,
/
),
also preserving exact energy and momentum conservation on the nite grid.
B.4. Results
We consider kinetic evolution of nonequilibrium optically thick plasma con-
sisting at the moment t = 0 of electron-positron pairs with number density
n = 10
30
cm
3
in the small region with radius R
0
= 200 cm. Although
such parameters are far from both laboratory conditions and the real GRB
sources, we consider this choice of parameters important since it provides
some new insights with respect to the traditional hydrodynamic description
(Goodman, 1986), (Piran et al., 1993), (M esz aros et al., 1993), (Mustafa and
K ampfer, 2009), (Yaresko et al., 2010). Figs. B.1B.5 show the plasma evolu-
tion with time, and Fig. B.6 shows spectra of photons near the maximum of
emission at t = 7 10
7
sec, crossing the sphere with radius 2.2 10
4
cm.
Initially nonequilibrium plasma relaxes to the thermal state on the
timescale 5 10
11
sec, and it starts to expand, on the dynamical timecale
R
0
/c 6 10
9
sec, see Fig. B.3. The concentrations (Fig. B.2) and the optical
depths of both electrons/positrons and photons (Fig. B.1) decrease with time.
72
B.4. Results
10
2
10
3
10
4
10
-1
10
0
10
1
10
2
10
3
10
4
10
5
10
6
10
7
1.7 10
-7
10
-7
3 10
-8
7 10
-9
2 10
-15
, e
r, cm
Figure B.1.: Mean optical depth for photons (solid) and pairs (dashed) as a
function of the radius at different time moments, as marked on the curves.
10
2
10
3
10
4
10
20
10
21
10
22
10
23
10
24
10
25
10
26
10
27
10
28
10
29
10
30
, e
n
g
,
n
e
r, cm
Figure B.2.: Photon number density (solid) and pair number density (dashed)
as a function of the radius at different time moments, as in Fig. B.1.
73
B. Dynamics and emission from mildly relativistic plasma
10
2
10
3
10
4
1
2
3
4
, e
r, cm
Figure B.3.: Photon (solid) and pair (dashed) radial velocity as a function of
the radius at different time moments, as in Fig. B.1.
10
2
10
3
10
4
10
34
10
35
10
36
10
37
10
38
10
39
10
40
, e
L
,
e
r
g
/
s
r, cm
Figure B.4.: Rates of energy ow in photons (solid), in pairs (dashed) through
the surface at radius r as a function of the radius at different time moments,
as in Fig. B.1.
74
B.4. Results
10
2
10
3
10
4
10
1
10
2
10
3
, e
<
>
,
k
e
V
r, cm
Figure B.5.: Average energy of photons (solid) and of pairs (dashed) as a func-
tion of the radius at different time moments, as in Fig. B.1.
100 1000
10
33
10
34
10
35
10
36
1.3 10
-6
8.7 10
-7
7 10
-7
4 10
-7
d
L
/
d
,
e
r
g
/
s
*
k
e
V
, keV
Figure B.6.: Energy spectrum of emerging photons in the laboratory reference
frame, at the radius 2.2 10
4
cm at selected time moments near the maximum
luminosity.
75
B. Dynamics and emission from mildly relativistic plasma
As the temperature becomes smaller than m
e
c
2
the energy density in pairs
starts to decrease exponentially, since they remain in thermal equilibrium.
When the optical depth of photons is large, we nd hydrodynamic descrip-
tion accurate enough to calculate the photons spectra. At t = 1.5 10
7
sec the
optical depth for photons decreases below unity, see Fig. B.1. After this mo-
ment the radial distribution of photons in the expanding shell bocomes xed,
see Fig. B.4. The average energy shown in Fig. B.5 indicates that the peak
of the photon spectrum is in the range 10 1000 keV. Photon spectra around
the maximal luminosity are nonthermal, see Fig. B.6. They possess an expo-
nential cut-off at high energy, and the spectral energy density is almost at in
its low energy part. This result may be an explanation of additional power in
low energy spectra observed in GRBs (Patricelli et al., 2011).
B.5. Conclusions and future perspectives
In this work we obtained for the rst time the detailed kinetic information in-
cluding photon spectra from initially optically thick outow, composed of
electron-positron pairs and photons. Initial conditions are selected in the
intermediate region between conditions which are expected to be reached
in laboratory experiments, and those thought to occur in astrophysical set-
tings. This allows us to study the radial self-acceleration of electron-positron
plasma up to mildly relativistic velocities. We found that photon spectra near
the peak of the luminosity are non-thermal, they possess an exponential cut-
off at high energy, and are almost at in the low energy part. This result
may shed some light on the issue why GRBs have nonthermal spectra, and
in particular why the low energy part of the spectrum contains more power
with respect to Planck spectrum. We hope to extend our analysis for ultra-
relativistic case which is necessary to be explored for realistic description of
Gamma-Ray Bursts phenomenon.
76
C. Evolution of the pair plasma
generated by a strong electric
eld
Introduction
Electron-positron pairs can be produced by vacuum polarization in strong
electric eld E if it satises the following relation
E
m
2
e
c
3
e h
= E
c
, (C.0.1)
where m
e
is the electron mass, c is the speed of light, h is the Planck constant
and E
c
is the so called critical electric eld (Rufni et al., 2010).
In (Benedetti et al., 2011) we addressed the problem of pair creation
in a strong electric by simple treatment based on continuity and energy-
momentum conservation equations. A second order ordinary differential
equation has been worked out and all the other physical quantities of inter-
est can be obtained from its solution. According to previous works (Kluger
et al., 1991),(Rufni et al., 2003b),(Benedetti et al., 2011), the pairs move back
and electric eld oscillates as well with the same frequency but with a dif-
ferent phase. This approach allowed us to study the behavior of the system
beyond the asymptotic time
a
, when the pairs oscillation frequency is close
to the plasma frequency; at this stage almost all the energy density of the ini-
tial electric eld is converted into rest energy density of the pairs. The optical
depth has been computed in order to estimate the time scale
after which
the interactions can not be neglected. We found that
a
for all the con-
sidered initial conditions, concluding that the role of the interactions should
be taken into account in a long run. Besides, we assumed that pairs are orig-
inally produced at rest even though the most general rate of pair production
already gives a specic distribution in the momentum space. In particular
they are in a momentum state such that there is motion only in orthogonal
direction to the initial electric eld.
We now extend and generalize the results of the previous work (Benedetti
et al., 2011) with the intention to investigate the effects of pairs interactions
and the dynamical role of a most general rate of pair production. Collisions
can be naturally described within the kinetic approach. There is invariance
77
C. Evolution of the pair plasma generated by a strong electric eld
under rotations around the direction of the electric eld. In this perspective
we solve numerically the relativistic Boltzmann-Vlasov equation in a uni-
form and homogeneous physical space but with an axially symmetric mo-
mentum space. The pairs interaction is accounted for by collision integrals
computed from the exact QED cross sections for the two particle interactions,
namely electron-positron annihilation into two photons and its inverse pro-
cess, Bhabha, M oller and Compton scatterings. As a consequence, also pho-
tons are described by the relativistic Boltzmann equation. The three particle
interactions are not taken into account because their cross section is roughly
times smaller than the two particle ones.
In the adopted kinetic scheme the Distribution Function (DF) is the basic
object we are dealing with because all the fundamental physical quantities
can be extracted from it; among them the number density, the rest and ki-
netic energy densities, the bulk momentum. Since it is well known that a
pair plasma thermalizes in a very short time scale (Aksenov et al., 2009), we
are also interested in this characteristic time for different initial conditions.
The temperature in kinetic equilibrium may be found from the total energy
and number densities of the plasma (Aksenov et al., 2007). Moreover, we
already know that the electric eld accelerates electrons and positrons up to
high Lorentz factors of the order of hundreds. Along the direction orthogonal
to that, pairs as they are produced have Lorentz factors at most of the order of
the initial electric eld. It means that distribution of electrons and positrons
in the momentum space is strongly anisotropic being elongated in the direc-
tion of initial electric eld. So we can think about system that is characterized
by different temperatures T
|
and T
is established.
C.1. Cylindrical coordinates in the momentum
space
Based on the symmetry of the problem we consider axially symmetric mo-
mentum space. Hence, the momentum of the particle is described by two
components, one parallel (p
|
) and one orthogonal (p
[0, +) , p
|
(, +) . (C.1.1)
78
C.2. The Distribution Function
Within the chosen phase space conguration, the prescription for the integral
over the entire momentum space is
_
d
3
p =
_
2
0
d
_
+
dp
|
_
+
0
dp
, (C.1.2)
and the relativistic energy is given by the following equation
=
_
p
2
|
+ p
2
+ m
2
, (C.1.3)
where m is the mass of the considered particle.
C.2. The Distribution Function
Instead of using the usual DF f such that the number density is given by the
integral over the momentum space
n =
_
d
3
p f = 2
_
+
dp
|
_
+
0
dp
f , (C.2.1)
we introduce a new DF F such that the energy density is given by the follow-
ing integral role
=
_
+
dp
|
_
+
0
dp
F . (C.2.2)
In isotropic momentum space this DF is reduced to the spectral energy den-
sity d/d. Hence we can recover f from F using the following prescription
f =
F
2 p
. (C.2.3)
Besides, because of the axial symmetry, F does not depend on , it depends
only on the two components of the momentum, that is
F = F(p
|
, p
) . (C.2.4)
79
C. Evolution of the pair plasma generated by a strong electric eld
C.3. Two temperature DF
Neglecting the relativistic degeneracy an ensamble of particles in thermal
equilibrium can be described by the Relativistic Maxwellian DF
f (p
|
, p
, T) = (C.3.1)
=
m
e
c
2
/k
B
T
4K
2
(m
e
c
2
/k
B
T)
exp
_
_
m
e
c
2
k
B
T
1 +
_
p
|
m
e
c
_
2
+
_
p
m
e
c
_
2
_
_
,
where T is the temperature and K
2
is the modied Bessel function of the sec-
ond kind. Therefore we expect that after a characteristic time scale, depen-
dent on the initial conditions, the distribution function of each sort of particle
should achieve the limit given by Eq. (C.3.1).
On the other hand, we are dealing with two independent components of
the momentum and the rate Eq. (C.4.2) indicates that initially the DF of elec-
trons and positrons is strongly anisotropic. For that reason we describe the
different particles populations with a Two Temperature DF before the ther-
mal equilibrium is reached. In this lines we adopt the following ducial DF
f (p
|
, p
, T
|
, T
, p
0
|
, k) = (C.3.2)
= k exp
_
_
_
m
e
c
2
k
B
T
_
1 +
_
T
T
|
_
2 _
p
|
p
0
|
m
e
c
_
2
+
_
T
T
_
2
_
p
m
e
c
_
2
_
_
_
,
where T = (T
|
+ T
,
|
,
, p
0
|
, k
/
) = (C.3.3)
= k
/
exp
_
m
e
c
2
k
B
_
1
|
_
p
|
p
0
|
m
e
c
_
2
+
1
_
p
m
e
c
_
2
__
,
80
C.4. Boltzmann equations
where the new temperatures are dened as follows
|
=
2 T
2
|
T
|
+ T
=
2 T
2
T
|
+ T
. (C.3.4)
C.4. Boltzmann equations
The relativistic Boltzmann equation for electrons and positrons can be written
in our framework as follows
1
2 p
_
F
(p
|
, p
)
t
E
F
(p
|
, p
)
p
|
_
=
=
(p
|
, p
(p
|
, p
) F
(p
|
, p
) + S(p
|
, p
,
E) , (C.4.1)
where
E = E/E
c
is the dimensionless electric eld,
,
E) =
_
m
e
h
_
4
[
E[ log
_
1 exp
_
(1 + (p
/(m
e
c))
2
)
[
E[
__
(p
|
) .
(C.4.2)
From the previous equation it is clear that if
E < 1 this rate is exponentially
suppressed. Going further, we see that for a given overcritical eld
E the
range of orthogonal momentum on which pairs will be distributed is approx-
imatively
_
0,
E m
e
c
(p
|
, p
)
t
=
(p
|
, p
(p
|
, p
) F
(p
|
, p
) , (C.4.3)
because their DF changes accordingly to the collisional term only. In more
detail, photons must be produced rst from the pairs annihilation, then they
affect the electron-positron DF through Compton and nally they can anni-
hilate producing one electron-positron pair.
81
C. Evolution of the pair plasma generated by a strong electric eld
C.5. Computational scheme
C.5.1. Acceleration and electric eld evolution
Once electrons and positrons are produced, they are accelerated by the elec-
tric eld toward opposite directions. The time derivative of the electron or
positron parallel momentum dp
|
in the presence of an electric eld E is given
by the equation of motion
dp
|
dt
= e E, (C.5.1)
where the sign + () refer to the positron (electron) and e is the electron
charge. Numerically, we move particles from one cell to another one such
that the number of particles is conserved and Eq. (C.5.1) is satised.
Also the electric eld evolves according to the Maxwell equations. Once
the currents of the moving pairs are computed, the time derivative of the
electric eld is known. Consequently a new ordinary differential equation
must be added to the system of Eqs. (C.4.1) and (C.4.3). However, due to
the uniformity and homogeneity of the physical space, we can describe the
electric eld simply using the energy conservation law.
C.5.2. Emission and absorption coecients
While the numerical description of the acceleration is quite easy, it is not so
for the collisional term: in fact the emission and absorption coefcients
contain several integrals over the momentum space of the distribution func-
tions. Since we describe a given interaction denoting the involved particles
as follows
1 + 2 3 + 4 , (C.5.2)
it means that the particle 1 and 2 are absorbed while particle 3 and 4 are
produced, then we have
1
,
2
and
3
,
4
. In order to obtain these emission
and absorption coefcients for each particle we make use of the following
quantity
C =
_
d
3
p
1
_
d
3
p
2
_
d
3
p
3
_
d
3
p
4
w
1,2;3,4
F
1
(p
1
|
, p
1
) F
2
(p
2
|
, p
2
)
(2)
2
1
p
1
2
p
2
, (C.5.3)
where
w
1,2;3,4
=
h
2
c
6
(2)
2
()
(3)
(p)
M
f i
2
16
1
4
(C.5.4)
contains the matrix element for the considered interaction.
The interaction cross section is invariant by rotations around an arbitrary
axis, therefore the angle
1
can be set to be constant. Then we have 4 con-
82
C.5. Computational scheme
servation laws, energy and momentum, given by the Dirac deltas in w. Our
choice is to eliminate the three integrals over the particle 4 by
(3)
(p) and
eliminate the integration over
2
using the ().
Rewriting eq. (C.5.3) in our computational scheme we get
C =
_
dp
1
|
_
dp
1
_
dp
2
|
_
dp
2
_
dp
3
|
_
dp
3
(C.5.5)
R(p
1
|
, p
1
, p
2
|
, p
2
, p
3
|
, p
3
) F
1
(p
1
|
, p
1
) F
2
(p
2
|
, p
2
) ,
where
R(p
1
|
, p
1
, p
2
|
, p
2
, p
3
|
, p
3
) =
h
2
c
6
2(4)
3
_
d
3
p
3
2
1
2
2
3
4
J
M
f i
2
, (C.5.6)
and the Jacobian is given by
J =
4
/p
2
sin
2
(p
3
cos
3
p
1
cos
1
) + cos
2
(p
1
sin
1
p
3
sin
3
)
.
(C.5.7)
All the information about each reaction are stored in R and the integral over
3
within its expression can be computed. For each particle we have different
coefcients bU
1
, bU
2
, aU
3
, aU
4
and each of them depends on p
1
|
, p
1
, p
2
|
, p
2
,
p
3
|
, p
3
1
(p
1
|
, p
1
) =
_
dp
2
|
_
dp
2
_
dp
3
|
_
dp
3
(C.5.8)
bU
1
(p
1
|
, p
1
, p
2
|
, p
2
, p
3
|
, p
3
) F
2
(p
2
|
, p
2
2
(p
2
|
, p
2
) =
_
dp
1
|
_
dp
1
_
dp
3
|
_
dp
3
(C.5.9)
bU
2
(p
2
|
, p
2
, p
1
|
, p
1
, p
3
|
, p
3
) F
1
(p
1
|
, p
1
3
(p
3
|
, p
3
) =
_
dp
1
|
_
dp
1
_
dp
2
|
_
dp
2
(C.5.10)
aU
3
(p
3
|
, p
3
, p
1
|
, p
1
, p
2
|
, p
2
) F
1
(p
1
|
, p
1
) F
2
(p
2
|
, p
2
4
(p
m
|
, p
n
) =
_
dp
1
|
_
dp
1
_
dp
2
|
_
dp
2
(C.5.11)
aU
4
(p
m
|
, p
n
, p
1
|
, p
1
, p
2
|
, p
2
) F
1
(p
1
|
, p
1
) F
2
(p
2
|
, p
2
) .
By denition, the particles 1, 2 and 3 exactly on the momentum points of
the grid, as a consequence it is not guaranteed that the particle 4 will be on the
grid point as well. Nevertheless, we need to conserve the number of particles,
the energy and the orthogonal and parallel components of the momentum.
83
C. Evolution of the pair plasma generated by a strong electric eld
Therefore we have to distribute it also on in the nearby cells as follows
F
4
(p
4
|
, p
4
)
t
=
n
x
mn
F
4
(p
m
|
, p
n
)
t
, (C.5.12)
where x
mn
are given by the conservation laws.
C.5.3. Two particle kinematics
The interaction between two particles, 1 and 2, that gives the particle 3 and 4
as a product can be represented schematically as follows
1 + 2 3 + 4 . (C.5.13)
The i-th particle has 3 degrees of freedom (p
i
|
, p
i
,
i
) in the momentum space
and summing all of them we get 12. Since the interaction process conserves
momentum and energy, they reduce to 8 independent degrees of freedom.
That means that 4 quantities can be determined uniquely once the others are
specied. For us, the 8 independent variables are p
1
|
, p
1
,
1
, p
2
|
, p
2
, p
3
|
,
p
3
,
3
; then p
4
|
, p
4
,
4
,
2
are functions of the previous ones.
The conservation of the parallel momentum gives use easily the corre-
sponding component of the 4-th particle
p
4
|
= p
1
|
+ p
2
|
p
3
|
. (C.5.14)
The orthogonal momentum for the same particle can be worked out using
the energy conservation law
1
+
2
=
3
+
4
(C.5.15)
and using the denition of the energy given by Eq. (C.1.3) we get
p
4
=
_
(
1
+
2
3
)
2
p
2
4
|
m
2
4
, (C.5.16)
where p
4
|
has been obtained in Eq. (C.5.15).
Fromthe conservation of the orthogonal momentumwe have the following
relations
p
1
cos(
1
) + p
2
cos(
2
) = p
3
cos(
3
) + p
4
cos(
4
) (C.5.17)
p
1
sin(
1
) + p
2
sin(
2
) = p
3
sin(
3
) + p
4
sin(
4
) (C.5.18)
from which we can write down analytical expressions for
4
and
2
. Unfor-
tunately for the previous system of equations we have two valid solutions.
84
C.5. Computational scheme
For
2
we have the following equation
a cos(
2
) + b sin(
2
) + c = 0 , (C.5.19)
where the coefcients are given by
a = p
1
cos(
1
) p
3
cos(
3
) ; (C.5.20)
b = p
1
sin(
1
) p
3
sin(
3
) ; (C.5.21)
c =
p
2
1
+ p
2
2
+ p
2
3
p
1
p
3
cos(
1
3
)
2p
2
. (C.5.22)
The solution of eq. (C.5.19) are given by the following conditions
if b ,= 0, a
2
+ b
2
,= 0, c = a
2
= 2 arctan
_
a
b
_
+ 2n ;
if b = 0, c = a
2
= + 2n ;
if a ,= c, a
2
+ b
2
ac b
a
2
+ b
2
c
2
,= 0
2
= 2 arctan
_
b
a
2
+ b
2
c
2
a c
_
;
if a ,= c, a
2
+ b
2
ac + b
a
2
+ b
2
c
2
,= 0
2
= 2 arctan
_
b +
a
2
+ b
2
c
2
a c
_
.
Once
2
has been chosen,
4
can be easily obtained from the eqs. (C.5.17)
and (C.5.18).
The Jacobian of Eq. (C.5.7) has been computed using using the following
identity for the Dirac Delta
( f (x)) =
i
(x x
i
)
[(d f /dx)
x
i
[
(C.5.23)
where f is a function such that f (x
i
) = 0. In our framework the function
inside the Dirac delta is given by the energy conservation
f (
2
) =
1
+
2
4
(
2
) (C.5.24)
where
2
is now the independent variable. From the previous equation we
compute its derivative with respect to
2
and the value
2
such that f (
2
) =
0. Rewriting explicitly Eq. (C.5.23) we have that
( f (
2
)) =
(
2
2
)
[(d f /d
2
)
2
[
= J (
2
2
) . (C.5.25)
85
C. Evolution of the pair plasma generated by a strong electric eld
C.6. Results
C.6.1. Non interacting systems
1 10 100
Time (tc)
0.001
0.01
0.1
1
n
/
n
_
m
a
x
Boltzmann equation
Previous result
0.028
Figure C.1.. The evolutions of the number density with time
found in (Benedetti et al., 2011) (dashed) and in our treatment
without interactions (solid) are compared for
E
i
= 10. A dif-
ferent behavior occurs after the rst oscillation: in the former
it continues to grow almost linearly but in the latter it satu-
rates after a few oscillations.
In this section we present the results obtained from the relativistic Boltz-
mann equation (C.4.1) for electrons and positrons with
= 0. In this
sense we neglect the interactions of electron positrons pairs. We expect this
result to be closely related to those reported in (Benedetti et al., 2011).
We solved numerically the Boltzmann equation for different initial electric
elds (
E
i
= 1, 3, 10, 30, 100) and with no pairs at the beginning; consequently
the electrons and positrons are produced exclusively by the vacuum polar-
ization process.
86
C.6. Results
0 50 100 150 200
Time (tc)
-40
-20
0
20
40
60
80
Electric Field (Ec)
Electron bulk parallel momentum (m_e*c)
Figure C.2.. Electric eld and bulk parallel momentum ob-
tained for the non-interacting system with initial condition
E
i
= 10. It is evident how the electric eld magnitude drops
within several oscillations in contrast with the result obtained
in (Benedetti et al., 2011). The bulk parallel momentum is de-
creasing rapidly with time as well, but we stress that its mag-
nitude in the rst two peaks is very similar to those attained
in (Benedetti et al., 2011)
For all the explored initial conditions, there are important analogies be-
tween the approach adopted in (Benedetti et al., 2011) and the one presented
in this work. We remind that the maximum achievable pairs number density
is
n
max
=
E
2
16m
e
c
2
, (C.6.1)
which corresponds to the case of maximum conversion of the initial energy
density into electron-positron rest energy density. In Fig. C.1 we see the ratio
of the pairs number density of pairs n
max
obtained in these two treatments.
The two curves are very close up to the rst zero of the electric eld identied
by the rst plateau since particles can not be produced.
From the electrons and positrons DFs we can extrapolate the bulk parallel
87
C. Evolution of the pair plasma generated by a strong electric eld
-200 -100 0 100 200
parallel momentum (m_e*c)
0
1x10
46
2x10
46
3x10
46
4x10
46
n
u
m
b
e
r
o
f
p
a
r
t
i
c
l
e
s
(
a
r
b
i
t
r
a
r
y
u
n
i
t
s
)
1st e-
2nd e-
3rd e-
4th e-
5th e-
6th e-
7th e-
8th e-
9th e-
9th e+
Figure C.3.. In this plot the integral along the orthogonal mo-
mentum of the distribution function is shown. Each curve
corresponds to a different zero of the electric eld; because
of that they are snapshots of the DF when a maxima the
bulk parallel momentum is achieved. This gure put in evi-
dence how the distribution function evolves toward a station-
ary conguration characterized by a bulk parallel momentum
very close to zero.
momentum of electrons and positrons as follows
p
|
=
_
dp
|
_
dp
(p
|
, p
) p
|
/
_
dp
|
_
dp
(p
|
, p
)/
, (C.6.2)
From the symmetry of our problem it must be that p
|
= p
|+
identically.
This quantity is plotted in Fig. C.2 and it is very close to that obtained in
(Benedetti et al., 2011) during this initial time interval.
Nevertheless, after this short period, the results of the two treatments di-
verge. As it was found in (Benedetti et al., 2011), the subsequent oscillations
occur with a quickly increasing frequency. In our case, the frequency of the
oscillations increases as well but slowly.
Besides, in (Benedetti et al., 2011) we found that the absolute peak value
88
C.6. Results
of the electric eld decreases slowly with time. Now we nd that this peak
electric eld decreases much faster and it does not exceed a few percent of
the starting value. Subsequently, as it follows from Eq. (C.4.2) the pair pro-
duction saturates as illustrated in Fig. C.1.
There is a third consequence of the strong damping of the electric eld.
Let us consider two different populations of electrons (or positrons): the rst
with very small orthogonal momentum and the second one with very large
orthogonal momentum. If we compute their bulk parallel momentum p
small
|
and p
large
|
separately at the end of the run, we nd that the former is approxi-
matively zero while the latter is not. This fact can be grasped from Eq. (C.4.2).
In fact electrons and positrons are originally generated with null parallel mo-
mentum, and the smaller the electric eld the shorter the range of orthogonal
momentum in which pairs are produced. The new born pairs gives more
weight to the particles with p
|
= 0 and its average p
|
is consequently low-
ered. Again, this effect manifests for particles with orthogonal momentum in
the range given by Eq. (C.4.2).
In Fig. C.2 we see how p
|
decreases rapidly and it attains a non relativistic
value at the end of the run. Each maximum of [ p
|
[ has a corresponding curve
in Fig. C.3 where we show the following marginalDF
dN(p
|
)
dp
|
=
_
+
0
dp
F(p
|
, p
. (C.6.3)
In Fig C.4 we show various normalized energy densities. The total energy
density of the pairs which is related to the actual and initial electric elds E
and E
i
by the energy conservation law
=
E
2
E
2
i
8
. (C.6.4)
The rest energy of the pairs, namely the sum of the electrons and positrons
rest energy densities, is
rest
=
+
+
= (n
+ n
+
) m
e
c
2
, (C.6.5)
where n
and n
+
are the electrons and positrons number densities. By de-
nition the plot of this quantity reproduces exactly the behavior of the number
density of pairs in Fig. C.1. Then we dene the kinetic energy density
kin
= (n
+ n
+
) m
e
c
2
_
_
_
p
|
m
e
c
_
2
+ 1 1
_
_
, (C.6.6)
which is the energy given by particles moving with the same magnitude for
89
C. Evolution of the pair plasma generated by a strong electric eld
the parallel momentum p
|
and with p
in
=
rest
kin
. (C.6.7)
We call this internal since it is related to the dispersion of the DF around a
given point with coordinates ( p
|
, p
,
normalized parallel momentum p
|1
and number density n
1
of pairs at the
rst zero of the electric eld obtained in (Benedetti et al., 2011), normalized
saturation number density n
s
for the non-interacting system and for different
initial electric eld.
E
i
k
B
T
/m
e
c
2
k
B
T
|
/m
e
c
2
p
|1
/m
e
c n
1
/n
max
n
s
/n
max
1 0.2 70 146 0.0068 0.01 - 0.02
3 0.3 120 73 0.0137 0.01 - 0.02
10 0.5 45 83 0.0120 0.03
30 0.8 65 125 0.0080 0.015
100 1.5 110 217 0.0046 0.01
eration with highest electric eld where the interactions are expected to play
soon an important role.
Then, we solved numerically Eq. (C.4.1) and (C.4.3) for
E
i
= 100 with all
the two particle QED interactions. The rst striking outcome is that after
t
= 1.27 10
2
t
C
from the beginning, the energy density of the photons is
roughly 10% of the electrons or positrons ones. This points out that at this
stage, also the presence of the photons can not be neglected anymore.
Again, looking at Fig. C.5 we realize that the photons DF is perfectly sym-
metric with respect to the plane identied by the condition p
|
= 0. The situa-
tion changes slightly for electrons and positrons because the electric eld that
accelerate them toward opposite directions, namely the regions characterized
by p
|
> 0 and p
|
< 0 respectively. However the differences with a perfectly
symmetric DF are very small and this fact can be understood easily having in
mind the time laps during which the electric eld acted on the pairs.
Another consequence due to the relative scatterings between particles is
that the bulk parallel momentum of the pairs is smaller that the one achieved
after the same evolution time in the non-interacting system. The latter can
be readily estimated since the electric eld is nearly constant during the time
interval t
m
e
c
p
m
e
c
Log
10
f
6.8
22.5
Figure C.5.. Density plot of the photons DF f
in the mo-
mentum space in the case with
E
i
= 100 after 1.27 10
2
t
C
.
At this instant the energy densities of photons and electrons
(or positrons) are
= 1.13 10
24
erg/cm
3
,
=
+
=
1.12 10
25
erg/cm
3
and the electric eld is
E = 99.99848.
95
C. Evolution of the pair plasma generated by a strong electric eld
96
D. Electron-positron plasma in
GRBs and in cosmology
D.1. Introduction
Electron-positron plasmas are discussed in connection with astrophysi-
cal phenomena such as Galactic Center, microquasars, Gamma-Ray Bursts
(GRBs), as well as laboratory experiments with high power lasers, for details
see Rufni et al. (2010). According to the standard cosmological model, such
plasma existed also in the early Universe. It is naturally characterized by the
energy scale given by the electron rest mass energy, 511 keV. It is interest-
ing that at the epoch when Universe had this temperature, several important
phenomena took place almost contemporarily: electron-positron pair annihi-
lation, the Big Bang Nucleosynthesis (BBN) and neutrino decoupling.
Electron-positron plasma also is thought to play an essential role in GRB
sources, where simple estimates for the initial temperature give values in
MeV region. Such plasma is energy dominated and optically thick due to
both Compton scattering and electron-positron pair creation, and relaxes to
thermal equilibrium on a time scale less than 10
11
sec, see Aksenov et al.
(2007). The latter condition results in self-accelerated expansion of the plasma
until it becomes either transparent or matter dominated.
In the literature there have been several qualitative arguments mentioning
possible similarities between electron-positron plasmas in the early Universe
and in GRB sources. However, until now there is no dedicated study which
draws analogies and differences between these two cases. This paper aims in
confronting dynamics and physical conditions in both cases.
D.2. General equations
The framework which describes electron-positron plasma both in cosmology
and in GRB sources is General Relativity. Both dynamics of expansion of
the Universe, and the process of energy release in the source of GRB should
be considered within that framework. Hydrodynamic expansion of GRB
sources may, however, be studied within much simplier formalism of Spe-
cial Relativity.
97
D. Electron-positron plasma in GRBs and in cosmology
We start with Einstein equations
R
1
2
g
R =
8G
c
4
T
, (D.2.1)
where R
, g
and T
_
;
=
(
g T
)
x
+
_
g
= 0, (D.2.2)
where
= p g
+ U
, (D.2.3)
where U
_
T
_
;
= U
;
+ U
;
= 0. (D.2.4)
Using the second law of thermodynamics
d
_
n
_
= Td
_
n
_
+
1
n
dp, (D.2.5)
where = /T is proper entropy density, T is temperature, one may rewrite
(D.2.4) as
(U
)
;
= U
;
+ U
;
= 0. (D.2.6)
Baryon number conservation equation has exactly the same form
(nU
)
;
= U
n
;
+ nU
;
= 0. (D.2.7)
Now recalling that U
x
=
d
d
and U
;
= d lnV/d, where V is comov-
ing volume, is the proper time, from (D.2.4) and (D.2.7) we get
d + d lnV = 0, d ln n + d lnV = 0, (D.2.8)
Finally, introducing the thermal index = 1 +
p
T
, and (D.2.6) gives
V
T
= const. (D.2.10)
For nmc
2
, which is the energy dominance condition, internal energy
plays dynamical role by inuencing the laws of expansion. For nmc
2
,
which is the matter dominance condition, internal energy does not play any
dynamical role, but determines the scaling law of the temperature. In order
to understand the dynamics of thermodynamic quantities in both early Uni-
verse and in GRBs, one should write down the corresponding equations of
motion.
D.2.1. Early Universe
For the description of the early Universe we take the Robertson-Walker met-
ric with the interval
ds
2
= c
2
dt
2
+ a
2
(t)
_
dr
2
1 kr
2
+ r
2
d
2
+ r
2
sin
2
d
2
_
, (D.2.11)
where a (t) is the scale factor and k = 0, 1 stands for the spatial curvature.
In homogeneous and isotropic space described by (D.2.11), Einstein equa-
tions (D.2.1) are reduced to Firedmann equations together with the continuity
equation
_
da
dt
_
2
+ c
2
k =
8G
3c
2
a
2
, (D.2.12)
2a
d
2
a
dt
2
+
_
da
dt
_
2
+ c
2
k =
8G
c
2
pa
2
, (D.2.13)
d
dt
+
3
a
da
dt
( + p) = 0, (D.2.14)
where a is the scale factor. Notice, that only two equations in the system
above are independent. The continuity equation (D.2.14) follows from the
1
Nonrelativistic component is represented by baryons. For simplicity we assume only one
sort of baryons, say protons, having mass m. Ultrarelativistic component is represented
by photons and electron-positron pairs.
99
D. Electron-positron plasma in GRBs and in cosmology
Einstein equations (D.2.12) and (D.2.13) as the energy conservation. In fact,
(D.2.14) may be also obtained from the entropy conservation (D.2.4). The
comoving volume in Friedmanns Universe is V = a
3
, so (D.2.14) and the
rst equality in (D.2.9) are equivalent.
On the radiation dominated stage of Universe expansion one has
V
4/3
a
4
, n V
1
a
3
, (D.2.15)
while on the matter dominated stage
n V
1
a
3
. (D.2.16)
Entropy conservation (D.2.10) leads to the unique temperature dependence
on the scale factor
T V
1/3
a
1
. (D.2.17)
The corresponding time dependence of thermodynamic quantities may be
obtained from solutions of Friedmann equation (D.2.12) and continuity equa-
tion (D.2.14).
D.2.2. GRBs
Different situation takes place for the sources of GRBs. Assuming spherical
symmetry for the case of GRB the interval
2
is
ds
2
= c
2
dt
2
+ dr
2
+ r
2
d
2
+ r
2
sin
2
d
2
. (D.2.18)
Optically thick to Compton scattering and pair production electron-positron
plasma in GRB sources is radiation dominated. Its equations of motion follow
from the energy-momentum conservation law (D.2.2) and baryon number
conservation law (D.2.7). Initially plasma expands with acceleration driven
by the radiative pressure.
In spherically symmetric case the number conservation equation (D.2.7) is
(n)
t
+
1
r
2
r
_
r
2
n
_
2
1
_
= 0, (D.2.19)
Integrating this equation over the volume from certain r
i
(t) to r
e
(t) which we
assume to be comoving with the uid
dr
i
(t)
dt
= (r
i
, t),
dr
e
(t)
dt
= (r
e
, t), and
ignoring a factor 4 we have
2
General Relativity effects may be included by taking Schwarzschild or Kerr-Newman met-
ric. However, we are interested in optically thick plasma which expands with acceleration
and propagates far from its source, where the spatial curvature effects may be neglected.
For this reason we simplify the treatment and adopt a spatially at metric.
100
D.2. General equations
r
e
_
r
i
(n)
t
r
2
dr +
r
e
_
r
i
r
_
r
2
n
_
2
1
_
dr = (D.2.20)
t
r
e
_
r
i
(n) r
2
dr
dr
e
dt
n(r
e
, t)(r
e
, t)r
2
e
+
dr
i
dt
n(r
i
, t)(r
i
, t)r
2
i
+
+r
2
e
n(r
e
, t)
_
2
(r
e
, t) 1 r
2
i
n(r
i
, t)
_
2
(r
i
, t) 1 =
=
d
dt
r
e
_
r
i
(n) r
2
dr = 0,
Since we deal with arbitrary comoving boundaries, this means that the total
number of particles integrated over all differential shells is conserved
N = 4
R(t)
_
0
nr
2
dr = const, (D.2.21)
where R(t) is the external radius of the shell.
Following Piran et al. (1993) one can transform (D.2.19) from the variables
(t, r) to the new variables (s = t r, r) and then show that
1
r
2
r
_
r
2
n
_
2
1
_
=
s
_
n
+
2
1
_
. (D.2.22)
For ultrarelativistic expansion velocity 1, the RHS in (D.2.22) tends to
zero, and then the number of particles in each differential shell between the
boundaries r
i
(t) and r
e
(t) is also conserved with a good approximation, i.e.
dN = 4nr
2
dr const. (D.2.23)
Relations (D.2.21) and (D.2.23) then imply
4
r
e
_
r
i
_
nr
2
_
dr = 4
_
n(r, t)(r, t)r
2
_
r
e
_
r
i
dr (D.2.24)
= 4
_
nr
2
_
const,
where the rst argument of functions n(r, t) and (r, t) is restricted to the in-
terval r
i
< r < r
e
, and consequently r
e
r
i
const. Taking into account
that r
i
(t) and r
e
(t) are arbitrary, this means that ultrarelativistically expand-
101
D. Electron-positron plasma in GRBs and in cosmology
ing shell preserves its width measured in the laboratory reference frame. This
fact has been used by Rufni et al. (2000) and referred there as the constant
thickness approximation.
The volume element measured in the laboratory reference frame is dV =
4r
2
dr, while the volume element measured in the reference frame comov-
ing with the shell is dV = 4r
2
dr. Comoving volume of the expanding
ultrarelativistic shell with const will be
V = 4
_
r
r
r
2
dr 4r
2
. (D.2.25)
Then we rewrite the conservation equations (D.2.9) as
r
2
= const, nr
2
= const, (D.2.26)
Unlike the early Universe, where both energy and entropy conservations re-
duce to (D.2.14), in the case of GRBs the energy conservation is a separate
equation coming from the zeroth component of (D.2.2) as
(T
0
)
;
= U
0
U
;
+ U
(U
0
)
;
= 0. (D.2.27)
which gives
2
r
2
= const. (D.2.28)
From (D.2.26) and (D.2.28) we then nd
r
2(1)
2
, n r
2
2
, r
2
2
. (D.2.29)
For the ultrarelativistic equation of state with = 4/3 we immediately obtain
r, n r
3
, r
4
. (D.2.30)
Taking into account that the relation between the comoving and the physi-
cal coordinates in cosmology is given by the scale factor a, it follows from
(D.2.30) that both energy density and baryonic number density behave as
in the radiation dominated Universe, see (D.2.15). This analogy between
the GRB source and the Friedmann Universe is noticed by Shemi and Piran
(1990), Piran et al. (1993).
In the presence of baryons as the pressure decreases, plasma becomes mat-
ter dominated and expansion velocity saturates. Hence for the nonrelativistic
equation of state with = 1 different scaling laws come out
= const, n r
2
, r
2
. (D.2.31)
Transition between the two regimes (D.2.30) and (D.2.31) occurs at the radius
R
c
= B
1
R
0
, where R
0
is initial size of plasma.
102
D.3. Heavy elements
Therefore, one may reach the conclusion that for comoving observer
the radiation-dominated plasma looks indistinguishable from a portion of
radiation-dominated Universe. However, this is true only in the absence of
pressure gradients. Strong gradients are likely present in GRB sources, and
they should produce local acceleration in the radiation-dominated electron-
positron plasma, making it distinct from the early Universe, where matter
inhomogeneities are known to be weak.
It is easy to get from (D.2.26) and (D.2.28) for internal energy density and
temperature
r
4
, T r
1
, R
0
< r < R
c
, (D.2.32)
and
r
8/3
, T r
2/3
, R
c
< r < R
tr
, (D.2.33)
where R
tr
is the radius at which the outow becomes transparent. The out-
ow may become transparent for photons also at the acceleration phase, pro-
vided that R
tr
< R
c
. For instance, a pure electron-positron plasma gets trans-
parent at the acceleration phase.
D.3. Heavy elements
Cosmological nucleosynthesis is a well established branch of cosmology.
Classical computations made in the middle of the XXth century revealed that
heavy elements cannot be built in the early Universe. Hydrogen and helium
contribute approximately 3/4 and 1/4, leaving some room, much less than 1
per cent for deuterium, tritium and lithium. All the heavier elements must
have been produced in stars.
Some of these stars, as indicated by observations, end their life as progeni-
tors of GRBs. For this reason it is likely that initially in the source of GRBs el-
ements heavier than hydrogen are present. In this section we consider chem-
ical evolution of plasma in the sources of GRBs.
Assume that in the source of a GRB the amount of energy E
0
is released
in the volume with linear size R
0
during the time t, making this region
optically thick to Compton scattering and pair production. The amount of
baryons which may be present as well is parametrized by
B
_
_
Mc
2
E
0
, t R
0
/c,
Mc
2
L
, t R
0
/c,
(D.3.1)
where L = dE/dt is the luminosity,
M = dM/dt is the mass ejection rate
and M is total baryonic mass. Ultrarelativistic outow is generated through
103
D. Electron-positron plasma in GRBs and in cosmology
thermal acceleration of baryons by the radiative pressure if plasma is initially
energy dominated, i.e.
B 1. (D.3.2)
In the case of instant energy release with t R
0
/c initial temperature in
the source of GRB may be estimated neglecting the baryonic contribution,
provided (D.3.2) is satised as
T
0
_
3E
0
4aR
3
0
_
1/4
6.5E
1/4
54
R
3/4
8
MeV, (D.3.3)
where a = 4
SB
/c,
SB
is the Stefan-Boltzmann constant and the last value
is obtained by substituting numerical values for E
0
= 10
54
E
54
erg and R
0
=
10
8
R
8
cm.
As it has been shown in Kafexhiu (2010) for temperatures above 1 MeV
even low density plasma with density n = 10
18
cm
3
quite quickly destroys
all heavier nuclei, and the nal state contains just protons and neutrons and
some small traces of Deuterium and
4
He. The timescale of this process (
10
2
sec for T
0
= 1 MeV) strongly depends on temperature, but the rates
of almost all reactions increase with temperature, and correspondingly the
abundances of nuclei evolve much faster. Therefore, nuclei disintegration is
fast enough to occur before plasma starts to expand and cool on the timescale
R
0
/c.
During early stages of plasma expansion its temperature decreases in the
same way as it happened in the early Universe. Therefore similar synthesis
of light elements to BBN occurs also in sources of GRBs. Most important is,
however, another similarity with the early Universe: it is well known that
practically all free neutrons have been captured into elements heavier than
hydrogen. So we do not expect dynamically important free neutrons present
in GRB plasma after it started to expand and cool down unless they are en-
gulfed by the expanding plasma later. The role of such free neutrons have
been considered in the literature, see e.g. Derishev et al. (2000).
D.4. Recombination
On the radiation dominated phase both in the early Universe and in GRB
plasma entropy conservation (D.2.4) results in decrease of temperature
(D.2.17). In the early Universe, after the BBN epoch and electron-positron
annihilation, cosmological plasma consists of fully ionized hydrogen, helium
and small admixture of other light elements. The temperature continues to
decrease until it gets sufciently low to allow formation of neutral atoms:
that is the moment in the cosmic history where the formation of the Cosmic
Microwave Background Radiation (CMB) happens.
104
D.4. Recombination
The theory of cosmological recombination of hydrogen, based on three
level approximation, has been developed by Zeldovich, Kurt and Sunyaev
Zeldovich et al. (1968) and Peebles Peebles (1968) in the late 60s. The only
modication that such theory undergone in the later years is the account for
dark matter and addition of more levels to the model, currently about 300.
There is a basic difference with respect to the equilibrium recombination es-
sentially by the process e + p H + , described by the Saha equation
n
e
n
p
n
H
=
g
e
g
p
g
H
(2m
e
kT)
3/2
h
3
exp
_
E
i
kT
_
, (D.4.1)
where g
i
are statistical weights, h is Plancks constant, E
i
= 13.6 eV is hydro-
gen ionization energy. This difference is due to the presence of the 2p quan-
tum level, which produces Ly- photons. The absorption of such photons is
very strong. However, ionization from the 2p level requires only E
i
/4. There-
fore the formation of neutral hydrogen proceed through the 2s 1s transition
in the presence of abundant Ly- photons.
In fact, the early Universe would become transparent for radiation even if
formation of hydrogen would have been forbidden, see e.g. Naselsky et al.
(2011). The optical depth to Thomson scattering is
=
_
t
0
t
T
n
b
cdt
4 10
2
m
h
_
_
+
m
(1 + z)
3
_
1/2
1
_
, (D.4.2)
where
T
is the Thomson cross section,
i
=
i
/
c
,
c
= 3H
2
c
2
/8G,
H = 100h km s/Mpc and b, m, stand for, respectively baryons, dark mat-
ter and cosmological constant contributions to the total energy density of the
Universe. For large z we have
(z
) = 1, z
8.4
2/3
b
1/3
m
h
2/3
.
For typical values
b
h
2
0.02,
m
0.3, and h 0.7 we have z
60.
At such redshift the Universe would be expected to become transparent to
Thomson scattering. That is exactly what happens in plasma in GRB sources.
Below we show that, unlike radiation-dominated cosmological expansion
where comoving quantities also fulll relations (D.2.32), the comoving tem-
perature in GRB outows remains always high enough to prevent recombi-
nation of hydrogen.
From (D.2.32) and (D.2.33) one can see that during both acceleration and
coasting phases the comoving temperature decreases. However, when the
outow reaches the radius R
s
= B
2
R
0
, the comoving temperature saturates.
It can be shown, see Rufni et al. (2011) for details, using the expression for
105
D. Electron-positron plasma in GRBs and in cosmology
the photospheric radius
R
tr
=
_
E
0
B
4m
p
c
2
_
1/2
, (D.4.3)
where m
p
and are proton mass and Thompson cross section, respectively.
Indeed, using (D.3.3), (D.4.3), (D.2.32) and (D.2.33) we have
T
min
= BT
0
_
R
c
R
tr
_
2/3
= (D.4.4)
=
_
3
4a
_
1/4
_
4m
p
c
2
_
1/3
(E
0
R
0
)
1/12
. (D.4.5)
Notice how extremely insensitive this value is with respect to both E
0
and R
0
!
Expressed in units of typical energy and size
T
(s)
min
42 (E
54
R
8
)
1/12
eV. (D.4.6)
In the case of gradual energy release with with t R
0
/c and constant
luminosity and mass ejection rate the initial temperature is
T
0
_
3L
16
SB
R
2
0
_
1/4
, (D.4.7)
and similar expression to (D.4.4) may be derived
T
min
=
_
3
16
SB
_
1/4
_
4m
p
c
2
_
1/3
L
1/12
R
1/6
0
t
1/3
, (D.4.8)
which may be rewritten, introducing L = 10
50
L
50
erg/s and t = 1t
1
s, as
T
(w)
min
22L
1/12
50
R
1/6
8
t
1/3
1
eV. (D.4.9)
Even if (D.4.9) appears to be less stringent that (D.4.6), they are both quite
insensitive to initial parameters. As a result, even if the comoving tempera-
ture decreases very much compared to its initial value, typically of the order
or MeV, it saturates well above the ionization energy of hydrogen E
i
= 13.6
eV, thus preventing formation of neutral hydrogen. In fact, such hydrogen
formed would not allow formation of the afterglow of GRBs. Indeed, even
having ultrarelativistic velocities of expansion neutral hydrogen would not
interact with the interstellar medium!
A simplied way to look at this lower bound on the comoving temperature
is to say that if a fraction of solar mass is released in the volume having
106
D.5. Conclusions
radius solar Schwarzschild radii, then its minimum comoving temperature
before transparency is
T
(s)
min
66 ()
1/12
eV, (D.4.10)
in the case of instant energy release and
T
(w)
min
3.7
1/12
1/6
t
1/4
1
eV, (D.4.11)
in the case of gradual energy release during time t
1
. Clearly in both cases
> 1, and likely < 1. Notice, that while in the case of instant energy re-
lease the lower bound on temperature decreases with increasing , it instead
increases in the case of gradual energy release.
Therefore, we have reached the conclusion that hydrogen recombination
which is responsible for transparency of cosmological plasma does not hap-
pen in GRB plasma. This difference in physical conditions may result in
deviations from black body spectrum, as observed in GRBs. Recently we
presented a theory of photospheric emission from relativistic outows, see
Rufni et al. (2011). Assuming that the spectrum of radiation in the comov-
ing reference frame is the perfect black body one, we have shown that the
spectrum seen by a distant observer may be essentially nonthermal due to
both geometric and dynamical special relativistic effects. The possibility that
the spectrum of photospheric emission is nonthermal also in the comoving
frame should be investigated.
D.5. Conclusions
Regarding the dynamical aspects, there is an apparent similarity between the
electron-positron plasma in the early Universe and the one in GRB sources.
For an observer comoving with the radiation-dominated plasma in GRB
source it may look indistinguishable from a portion of radiation-dominated
Universe. However, this is true only in the absence of pressure gradients.
Strong gradients are likely present in GRB sources, and they should pro-
duce local acceleration in the radiation-dominated electron-positron plasma,
making it distinct from the early Universe, where matter inhomogeneities are
known to be weak.
There is also an apparent similarity with respect to the nucleosynthesis
phenomenon. Given that the temperature reached in GRB sources, see Eq.
(D.3.3), may be as high as several MeV, nuclear reactions are expected to op-
erate on timescales of 10
2
sec or shorter. That is on the order of magnitude
of dynamical timescale of the GRB sources. It means that reprocessing of nu-
clear abundances may likely take place in GRB sources. Since observations
imply that GRBs may originate from compact stellar objects elements heavier
107
D. Electron-positron plasma in GRBs and in cosmology
than helium are likely to be present in GRB sources. Such heavy elements are
then destroyed, resulting mainly in protons with small admixture of helium.
Thus, similarly to the early Universe, we do not expect dynamically impor-
tant free neutrons present in GRB plasma after it started to expand and cool
down unless they are engulfed by the expanding plasma later.
Finally, there is an important difference between the electron-positron
plasma in the early Universe and the one in GRB sources. We show in this
paper that unlike the primordial plasma which recombines to form neutral
hydrogen, and emits the Cosmic Microwave Background Radiation, GRB
plasma does not cool down enough to recombine. Therefore GRB plasma be-
comes transparent due to Compton scattering. The kinetics of this process
is different from the cosmic recombination. In particular, cosmic recombina-
tion is a photon non-conserving process, which results in an almost perfect
black body spectrum of CMB. At the same time, Compton scattering, which
is a dominant process at GRB transparency is a photon conserving process.
This difference may result in deviations from the perfect thermal spectrum of
emission to be observed when GRB plasma becomes transparent.
108
E. Photospheric emission from
relativistic outows
E.1. Introduction
High Lorentz factors of the bulk motion of various outows are common in
relativistic astrophysics. The best known examples are Active Galactic Nuclei
(Maraschi, 2003), microquasars and Gamma-Ray Bursts (GRBs) (Piran, 2004).
In the latter case outows indeed reach ultrarelativistic velocities. For this
reason GRB emission a) originates far from the source of energy release and
b) is observed as a transient event with a typical duration on the order of
seconds.
Various models are suggested to explain the acceleration of outows to ul-
trarelativistic velocities. The electromagnetic model (Lyutikov, 2006) assumes
that the energy in the source of GRB is converted into electromagnetic energy
which is transported in the form of a Poynting ux. We adopt here another
popular idea that the energy release leads to creation of an optically thick
source which expands due to thermal acceleration. This idea is the basis of
both the reball (Piran, 1999) and the reshell (Rufni et al., 2009a) models.
In pioneer works by Goodman (1986) who considered an instant explosion,
and by Paczynski (1986) who discussed a gradual energy release, a conclu-
sion was reached that the electron-positron plasma is created in the source
of GRB. Assuming further that the plasma reaches thermal equilibrium they
focused on hydrodynamic expansion in such models and gave photometric
and spectroscopic predictions for GRBs. Later, baryonic loading of reballs
was considered for explosions by Shemi and Piran (1990) and for winds by
Paczynski (1990). Abramowicz et al. (1991) considered the appearance of the
photosphere of the relativistic wind to a distant observer showing that it is
concave.
The interest to photospheric emission from relativistic winds has been re-
vived recently in papers by Daigne and Mochkovitch (2002), Peer et al.
(2007), Beloborodov (2011), Ryde et al. (2011), Peer and Ryde (2011), and
others.
In the reshell model, which assumes an explosive energy release, the rst
potentially visible component of any GRB, the Proper GRB (Bianco et al.,
2001), comes from the thermal ash of photons emitted when the outow
becomes transparent for photons.
109
E. Photospheric emission from relativistic outows
In this paper we readdress the issue of the photospheres of relativistic out-
ows using a simple model valid for both instant and gradual energy release.
We rederive analytic expression for the optical depth, give asymptotic solu-
tions for the photospheric radius providing clear physical interpretation of
our results. Then we present the analytic treatment and tting formulae for
observed ux of photospheric emission assuming isotropic thermal distribu-
tion of photons in the comoving frame. Finally, we compute both instanta-
neous and time-integrated spectra of photospheric emission. These results
are then applied within both the shell and the wind models of GRBs.
The structure of the paper is as follows. In Section 2 we discuss, compare
and contrast an impulsive explosion and gradual energy release, giving rise
respectively to an ultrarelativistic shell and wind. In Section 3 we compute
the optical depth and the radius of photosphere of a portion of relativistic
wind. Section 4 discusses the geometry and dynamics of the photosphere
in the relativistic outow. In Section 5 we treat the observed ux emerging
from the photosphere. Observed spectra are computed in Section 6. Then we
apply these results to GRBs in Section 7. Conclusions follow.
E.2. Optically thick relativistic outows: wind vs.
explosion
Relativistic outows are generally classied as winds or shells depending on
how fast the energy in their sources is released. Consider energy release in
a source of linear dimension R
0
. If the time scale of energy release is short
t R
0
/c, an explosion occurs, which may be characterized by the size
R
0
, total energy released E
0
and total baryonic mass M. When the energy is
released gradually, on a time scale t R
0
/c, but the source luminosity L
exceeds the Eddington limit, a wind is formed, which is characterized by its
activity time t, luminosity L and mass ejection rate
M.
It is possible that the temperature in the source of the relativistic outow
is large enough for electron-positron pair creation. In that case e
+
e
pairs
make an additional contribution to the optical depth. Whether the outow
becomes relativistic or not depends on the entropy in the region where the
energy is released. Both the wind and explosion cases can be parametrized
(Rufni et al., 2000) by a dimensionless baryonic loading parameter
B
_
_
Mc
2
E
0
, explosion,
Mc
2
L
, wind.
(E.2.1)
When the baryonic loading is sufciently small the baryons will be acceler-
110
E.2. Optically thick relativistic outows: wind vs. explosion
ated to a relativistic velocity v of bulk motion, attaining large Lorentz factors
=
_
1 (v/c)
2
_
1/2
, while in the opposite case of large baryonic loading
the outow remains nonrelativistic:
=
m
B
1
, B 1 (E.2.2)
v c
B/2, B 1. (E.2.3)
In what follows we consider only ultrarelativistic spherically symmetric
outows with 1. In the simplest cases of wind or explosion in vacuum,
the dynamics of the outow is divided into an acceleration phase and a coast-
ing phase (Piran, 1999) with respectively
R
R
0
, n n
0
_
R
R
0
_
3
, R
0
< R <
R
0
B
, (E.2.4)
B
1
= const, n n
0
_
R
R
0
_
2
,
R
0
B
< R < R
tr
, (E.2.5)
where R
tr
is the radius at which the outow becomes transparent to Thomson
scattering and n is the comoving number density of baryons in the outow.
The outow may also become transparent for photons at the acceleration
phase, provided that R
tr
< B
1
R
0
. Notice that in the case of an impulsive
explosion for r R
0
the matter and energy appear to a distant observer to
be concentrated in a geometrically thin shell having width l R
0
due to the
relativistic contraction.
It is important to stress that both an innitely long wind with a time-
independent mass ejection rate and luminosity on the one hand, and an in-
nitely thin shell originating from an instantaneous explosion in innitely thin
region represent two limiting cases for the energy release.
During both acceleration and coasting phases the continuity equation for
the laboratory number density reduces to n r
2
. We parametrize generic
density prole as
n =
_
_
n
0
_
R
0
r
_
2
f (), R
o
(t) l < r < R
o
(t),
0, otherwise,
(E.2.6)
where R
o
(t) is the equation of motion of the outer boundary of the outow,
f () is an arbitrary function of the depth of the outow measured from the
outer boundary. Analogously one may introduce the corresponding func-
tions for temperature and Lorentz factor dependences within expanding out-
ows.
In this paper we focus on a simplest case with constant number density,
111
E. Photospheric emission from relativistic outows
Figure E.1.: The laboratory number density is shown for a shell with initially
constant density and a wind with nite duration at two different times t = t
1
and t = t
2
. While the wind preserves the r
2
shape, the shape of the shell
changes making its density prole steeper with time.
temperature and Lorentz factor measured by an observer with xed labora-
tory radius. This prole corresponds to a portion of relativistic wind with
f () = 1. Such an outow may be produced by a gradual energy release
with constant luminosity and mass ejection rate on a nite time t and we
will refer to it as the wind.
Below we show that both the shell and the wind dened above may appear
for photons emitted inside it as long wind or as thin shell, depending on the
initial conditions that specify respectively their width l and activity duration
t. It is crucial to keep in mind that photons emitted inside the expanding
outow propagate in a medium whose laboratory number density depends
both on the radial coordinate and on time (r, t) n (r, t). For photons prop-
agating in the wind the spatial dependence of the number density plays the
key role, while for photons propagating in the shell its time dependence is
crucial.
In the next section we compute the optical depth of the portion of wind.
E.3. Optical depth along the line of sight
In this section we compute the optical depth of relativistic outows and de-
termine the corresponding photospheric radii along the line of sight. The
optical depth along the photon world line L is
=
_
L
j
dx
, (E.3.1)
112
E.3. Optical depth along the line of sight
where is the cross section, j
n
2
dr. (E.3.3)
In the remaining part of this section we will consider the optical depth under
different conditions characterizing the outow and derive expressions for the
photospheric radius along the line of sight.
E.3.1. Pure electron-positron plasma
A pure electron-positron plasma reaches thermal equilibrium before expan-
sion and remains thermal and accelerating until it becomes transparent to
radiation (Aksenov et al., 2007, 2009, 2010). Comoving number density of
electrons and positrons is a function of their comoving temperature, which
is decreasing during accelerating adiabatic expansion (e.g. Shemi and Piran
(1990)) as
T(R) = T
0
R
0
R
, (E.3.4)
where T
0
is the temperature of plasma in the source. Since the transparency
temperature is nonrelativistic, T
tr
m
e
c
2
/k, we can treat electron-positron
pairs as nondegenerate and nonrelativistic. Their number density is then
n(T) =
1
2
_
kTm
e
h
2
_
3/2
exp
_
m
e
c
2
kT
_
. (E.3.5)
113
E. Photospheric emission from relativistic outows
From (E.3.3), (E.3.4), and (E.3.5) we obtain the optical depth along the line of
sight
=
_
R+R
R
2
_
kT
0
m
e
h
2
_
3/2
_
R
0
r
_
5/2
exp
_
m
e
c
2
r
kT
0
R
0
_
dr. (E.3.6)
Initial temperature T
0
can be expressed in terms of initial size and energy.
Then by equating (E.3.6) to unity we obtain the radius at which the outow
becomes completely transparent along the line of sight, the transparency radius
R
tr
.
Due to exponential dependence on the radial coordinate in (E.3.6), trans-
parency is reached at
kT
0.040m
e
c
2
(E.3.7)
independent of the initial conditions. Note that the optical depth for an ex-
panding electron-positron-photon shell computed by Shemi and Piran (1990)
is incorrect since it uses photon thin asymptotics, see (E.3.23) below, which
never applies to the pure e
+
e
_
3E
0
l
4a
_
1/4
, (E.3.8)
where a = 4
SB
/c and
SB
is the Stefan-Boltzmann constant. Analogously in
the case of the wind
R
tr
=
1
T
_
E
0
R
2
0
4la
_
1/4
=
1
T
_
LR
2
0
16
SB
_
1/4
, (E.3.9)
where we used a relation
E
0
= Lt Ll/c. (E.3.10)
E.3.2. Acceleration phase
Now we focus on the acceleration phase described by (E.2.4), assuming that
the optical depth (E.3.2) along the line of sight corresponds to electrons with
density equal to the baryon density n. Here and below we take for the labo-
ratory density a prole (E.2.6) with f () = 1 which gives
n =
_
_
_
n
0
_
R
0
r
_
2
, R
o
(t) l < r < R
o
(t),
0, otherwise.
(E.3.11)
114
E.3. Optical depth along the line of sight
Actually positrons give a contribution to the optical depth even as comov-
ing temperature decreases below (E.3.7) due to their freeze out. Their contri-
bution evaluates to
n
n
=
4lr
2
N
, (E.3.12)
where quantities with subscript refer to the temperature (E.3.7), and total
number of baryons is obtained from (E.3.11) as
N = 4n
0
R
2
0
l. (E.3.13)
In what follows we neglect the electron-positron pair contribution, assuming
n
n.
Assuming that the cross section is constant, and using (E.2.4), (E.3.3) and
(E.3.11) one has
=
_
R+R
R
n
0
_
R
0
r
_
2
_
R
2
0
2r
2
_
dr =
1
6
n
0
R
4
0
_
1
R
3
1
(R +R)
3
_
. (E.3.14)
In the ultrarelativistic approximation from equations of motion of the photon
and of the outow one can see that such photon cannot leave the outow
during acceleration phase. Therefore (E.3.14) reduces to
=
n
0
R
4
0
6R
3
=
NR
2
0
24R
3
l
. (E.3.15)
We then obtain the following expression for the transparency radius of the
accelerating outow:
R
tr
=
_
NR
2
0
24l
_
1/3
. (E.3.16)
Finally we can express the number of baryons in the outow for either the
shell or wind, assuming baryons are represented by protons with mass m
p
.
For the wind N =
Mt/m
p
Ml/(m
p
c), where the outow width is given
by l ct, while for the shell with l R
0
the number of baryons is N =
M/m
p
. Then using (E.2.1) we have
N =
_
_
E
0
B
m
p
c
2
, shell,
LBl
m
p
c
3
, wind.
(E.3.17)
115
E. Photospheric emission from relativistic outows
As result we obtain for the shell
R
tr
=
_
E
0
BR
2
0
24m
p
c
2
l
_
1/3
, (E.3.18)
and for the wind
R
tr
=
_
LBR
2
0
24m
p
c
3
_
1/3
. (E.3.19)
E.3.3. Coasting phase
At the coasting phase described by (E.2.5), the equation of motion for the
external boundary of the outow is
R
o
(t) = R + l + ct, (E.3.20)
so the photon sees the expanding outow 2
2
times thicker than its width
measured in the laboratory frame
R =
l
1
2
2
l. (E.3.21)
The optical depth along the line of sight computed along the path of the pho-
ton is again given by (E.3.3)
=
_
R+R
R
n
0
2
2
_
R
0
r
_
2
dr =
n
0
R
2
0
2
2
_
1
R
1
R + 2
2
l
_
. (E.3.22)
This expression has two asymptotics, which using (E.3.13) can be written as
follows
=
_
_
n
0
R
2
0
2
2
R
=
N
8
2
lR
, 2
2
l R,
n
0
R
2
0
l
R
2
=
N
4R
2
, 2
2
l R.
(E.3.23)
Finally the transparency radius for the shell is
R
tr
=
_
_
E
0
B
3
8m
p
c
2
l
, 2
2
l R
tr
,
_
E
0
B
4m
p
c
2
_
1/2
, 2
2
l R
tr
,
(E.3.24)
116
E.3. Optical depth along the line of sight
while for the wind it is
R
tr
=
_
_
LB
3
8m
p
c
3
, 2
2
ct R
tr
,
_
LBt
4m
p
c
2
_
1/2
, 2
2
ct R
tr
.
(E.3.25)
E.3.4. Discussion
Let us interpret the results obtained above. On the one hand, Eq. (E.3.15) and
the rst line of (E.3.23) with R
3
and R
1
correspond to the case
when a photon emitted inside the outow stays there for a signicant time
so that it feels its density decreasing with time (or radius). In this respect the
photon sees the outow as a long wind, even if the laboratory thickness of
the outow is small, l R. We refer to this case as a photon thick outow.
On the other hand, the second line in (E.3.23) with R
2
corresponds to
the case when the photon spends too little time inside the outow to feel
its density decreasing with radius (for the wind) or time (for the shell). The
photon sees the outow as a thin shell even if the duration of explosion
could be long and a wind was launched. We refer to this case as a photon thin
outow. In other words, a geometrically thin ultrarelativistically expanding
shell may be both thin or thick as viewed by the photon propagating inside
it.
Similar consideration may be applied to a photon emitted form any dis-
tance from the outer boundary of the outow. It is clear then, that even in
a photon thick outow there is always a photon thin layer located near the
outer boundary. During acceleration phase such a photon thin part accounts
for a fraction not larger than B of the entire width of the outow.
In the derivation of (E.3.23) we used the approximation (E.3.11) represent-
ing a portion of a relativistic wind. The results for both density proles are
equal, provided that R l.
The expressions for the radius of the photosphere of a relativistic wind
were obtained in e.g. M esz aros and Rees (2000). These formulas coincide up
to a numerical factor (which comes from the integration over the radial coor-
dinate) with our (E.3.19) and (E.3.25). It should be noted, however, that only
the photon thick asymptotic limit is applied in M esz aros and Rees (2000).
The photon thin asymptotic limit may also be valid for relativistic winds
in the coasting phase, provided that l R/(2
2
). This is an independent
condition from t R
0
/c and it is therefore possible to give the following
constraints for t under which the outow takes the form of a wind, but it is
photon thin:
R
0
c
t
LB
5
16m
p
c
4
. (E.3.26)
117
E. Photospheric emission from relativistic outows
Transparency radius for photon thick and photon thin asymptotics for a por-
tion of the coasting relativistic wind (E.3.25) was obtained by Daigne and
Mochkovitch (2002).
Similar considerations apply to an ultrarelativistic shell which is consid-
ered e.g. in Shemi and Piran (1990) and M esz aros et al. (1993) and by Rufni
et al. (2000) in the photon thin approximation. The corresponding condition
that the shell appears to be photon thin is
ct R
0
_
E
0
B
5
16m
p
c
2
_
1
2
.
It is possible, however, that initial conditions satisfy the opposite constraint
R
0
ct, R
0
_
E
0
B
5
16m
p
c
3
_
1
2
,
which results in a photon thick shell. This photon thick asymptotic limit has
been used by Nakar et al. (2005) following Grimsrud and Wasserman (1998)
without any discussion of its applicability. All cases discussed above have
been considered by M esz aros et al. (2002), except the transparency of pure
electron-positron outow. Finally, Toma et al. (2011) consider all asymptotic
solutions, applying them to a relativistic wind.
Numerical hydrodynamic simulations produce complex density, tempera-
ture and Lorentz factor proles of the outow. In particular, Piran et al. (1993)
and M esz aros et al. (1993) considered an explosion in a compact region with
radius R
0
and studied numerically the hydrodynamic evolution of an opti-
cally thick plasma with various initial conditions. They have shown that the
plasma forms a relativistically expanding shell with some density and ve-
locity proles. The characteristic width of the density prole appears to be
constant up to large radii, but later it increases linearly with radius due to the
fact that the Lorentz factor appears to be monotonically increasing within the
shell
l R
0
, R < B
2
R
0
, (E.3.27)
l R, R > B
2
R
0
. (E.3.28)
Such a spreading in density prole may result in a substantial increase in the
width l R
0
of the shell when it becomes transparent to radiation. Note that
condition (E.3.28) coincides with the condition in the second line of (E.3.23),
which corresponds to the case of a photon thin outow. Numerical integra-
tion of (E.3.3) shows nevertheless that the photon thin asymptotics (E.3.23) is
valid even for the shell undergoing such spreading.
118
E.4. Geometry and dynamics of the photosphere
E.4. Geometry and dynamics of the photosphere
Unlike traditional static sources usually dealt with in astrophysics, relativistic
outows have strongly time-varying photospheres. For a given laboratory
time t the photosphere geometry r = r() is obtained by equating (E.3.2) to
unity.
For the portion of wind (E.3.11) the optical depth can be calculated analyt-
ically both at acceleration and coasting phases for photon thin and photon
thick outows:
(r, , t) = n
0
R
2
0
_
1
r sin
_
tan
1
_
r sin
cT + r cos
__
_
1
1
2
2
m
_
_
1
r
1
_
(cT + r cos )
2
+ (r sin )
2
_
+
R
2
0
6
_
1
r
3
1
[(cT + r cos )
2
+ (r sin )
2
]
3/2
__
, (E.4.1)
where T is the time interval during which photon remains inside the outow,
determined by the equations of motion of the photon and of the outow, and
m
is dened by (E.2.2).
For the innitely long relativistic wind with = const (Abramowicz et al.,
1991) t, T , the photosphere is a static surface
r = n
0
R
2
0
_
sin
_
, (E.4.2)
which has a concave shape for > 2/3, see Fig. E.2. It appears for a distant
observer as a static spot with radius
R
obs
= n
0
R
2
0
(E.4.3)
and brightness decreasing from the center to the edge. For the accelerating
innite wind the photosphere is also a static surface and its curvature is larger
than that of the coasting wind (see Fig. E.2). These static photospheres repre-
sent asymptotic limits of the dynamic photospheres of coasting and acceler-
ating photon thick outows, respectively.
Inside the relativistic beaming cone = arccos the photosphere of an
innitely thin shell at a xed laboratory time is an innitely thin ring. The
collection of such rings for all laboratory times represents a surface
r =
_
1 cos
[ cos [
_
1/2
, (E.4.4)
119
E. Photospheric emission from relativistic outows
where = n
0
R
2
0
l is the shell surface density. The curvature of this surface is
even larger than that of accelerating wind, as can be seen from Fig. E.2 and
the small angle 1/ expansion of (E.4.1)
=
_
_
Nl
24R
3
_
1 +
R
2
2
l
2
_
, R R
0
,
N
8lR
2
_
1 +
1
3
2
_
, R
0
R 2
2
l,
N
4R
2
_
1 + 2
2
2
_
, R 2
2
l.
(E.4.5)
The surface (E.4.4) represents asymptotic limit of photospheres of photon thin
outows.
In ultrarelativistic case the surfaces (E.4.2) and (E.4.4) give the position of
the corresponding photon thick and photon thin outow photospheres with
very good accuracy.
The part of dynamic photosphere seen by a distant observer at a given in-
stant of arrival time t
a
represents equitemporal surface (EQTS) of the photo-
spheric emission. EQTS has been initially introduced for the GRBs by Bianco
et al. (2001). We will refer to that surfaces as Photospheric EQTS, (PhE).
E.4.1. Coasting photon thin case
Geometrical shape of dynamical PhE of the photon thin outow is similar to
EQTS of innitesimally thin constantly emitting relativistic shell considered
rstly by Couderc (1939) and then by Rees (1966, 1967). The EQTS of this
shell appears to a distant observer as an ellipsoid with axes ratio equal to
. However the PhE of photon thin outow is not the entire ellipsoid: it is
only a part of that surface, see Fig. E.6. The external boundary of the surface
for a given t
a
is dened by the condition ( = 0, , t
a
) = 1, which means
that the optical depth of photons emitted from the outermost layer of the
outow equals unity. In the photon thin asymptotics this surface coincides
with the relativistic beaming cone = . As soon as the outow reaches
the transparency radius given by the second lines of (E.3.24) for the shell or
(E.3.25) for the wind an internal boundary appears as well. The surface of
internal boundaries is described by (E.4.4), see also Fig. E.6.
E.4.2. Accelerating and coasting photon thick cases
As we already mentioned in Sec. E.3.4, in a photon thick outow there is
always an external photon thin layer. Since this layer gets transparent rst,
initial evolution of PhE in the photon thick case is similar to the photon thin
120
E.4. Geometry and dynamics of the photosphere
5 6 7 8 9
0.0
0.2
0.4
0.6
0.8
x
y
R
tr
Figure E.2.: The shape of photospheres of innitely long coasting (dashed
curve) and accelerating (dotted curve) winds as well as time-integrated pho-
tosphere of innitely thin shell (solid curve) in logarithmic coordinates x =
cos log
10
r
tr
, y = sin log
10
r
tr
for = 1 10
3
.
121
E. Photospheric emission from relativistic outows
one. When transition to the photon thick asymptotics occurs, initially convex
PhE of the photon thin layer of the outow transforms to a concave PhE of its
photon thick part, see Fig. E.4 and Fig. E.7 for accelerating and coasting cases,
respectively. This concave PhE in both cases approaches the photosphere of
innitely long wind. In the coasting case the approach to the surface (E.4.2)
is only asymptotic, while in the accelerating case the photosphere actually
reaches it at nite arrival time. Similarly to the photon thin case, there ex-
ists an external boundary of the PhE determined by the same condition, and
shown in Fig. E.7. Notice that this boundary is wider than the relativistic
beaming surface (these are tube and cone for accelerating and coasting out-
ows, respectively). As soon as the innermost part of the outow reaches the
transparency radius, i.e. observer sees the switching off of the wind, the in-
ner boundary of the PhE appears with increasing (t
a
). The surface of these
boundaries is given by (E.4.2) in the case of coasting photon thick outow.
E.5. Observed ux
In this section we consider the appearance of the outow photosphere to a
distant observer and describe time evolution of luminosity in adiabatic ap-
proximation.
E.5.1. Adiabatic approximation for evaluation of observed
ux
Take a volume element dV located at the laboratory radial distance from
the outer boundary of the outow at angular separation from the line of
sight, see Fig. E.3. It is convenient to introduce the notation = cos . The
basic assumption of our method is that photons are locked inside this volume
element for (, , t) > 1 and stream freely when the optical depth decreases
below unity. In this approximation we neglect diffusion of photons from the
optically thick outow. Besides, we assume that all photons at given (, )
are emitted simultaneously. It means that the last scattering surface dened
by the condition = 1 is innitely thin. The probability of photon emission
is function in (, ). Laboratory volume of this element is then
dV = 2r(t)
2
dd. (E.5.1)
Photons are emitted to the observer from the element at the laboratory time
t
e
(, ) dened by
(, , t
e
) = 1, (E.5.2)
when the radial coordinate of the element, determined from the equation of
motion at given t
e
, is r = R
e
(, ). Emitted energy is obtained fromthe energy
122
E.5. Observed ux
d
R
o
r
R
obs
Figure E.3.: Geometry and illustration of variables used for calculation of the
ux received from relativistically expanding outow. Observer is located to
the right at innity.
density per unit solid angle u() as
dE
obs
= u()dVd =
I
c
dVd, (E.5.3)
where dis the solid angle of the observers detector as seen from the photo-
sphere in the laboratory frame and I is bolometric intensity of radiation (see
e.g. Eq. (1.6) of Rybicki and Lightman (1979)). Assuming isotropic photon
distribution in the comoving frame I
c
= const we get in the laboratory frame
(see e.g. Eq. (4.97b) of Rybicki and Lightman (1979))
I() =
I
c
4
(1 )
4
=
4
I
c
, (E.5.4)
where = (1 ), and nally changing from I
c
to the laboratory energy
density we arrive to (see e.g. Eq. (382) of Pauli (1958))
dE
obs
=
d
4
2
(1 +
2
/3)
4
dV
3
16
2
4
dVd. (E.5.5)
Dene the arrival time of that radiation as
t
a
= t
e
R
e
/c, (E.5.6)
i.e. as time delay with respect to a trigger photon emitted at t
e
= 0, R
e
= 0
(see e.g. Peer and Ryde (2011)). The observed ux at a given t
a
is obtained
by integration of dF =
dE
obs
dt
a
over the PhE.
123
E. Photospheric emission from relativistic outows
E.5.2. Acceleration phase
When transparency is achieved at acceleration phase (E.2.4) the equation of
motion of the element dV is
r(, t) =
_
R
2
0
+ (ct )
2
, t > /c. (E.5.7)
Entropy conservation gives for laboratory energy density
=
0
(R
0
/r)
2
. (E.5.8)
Using as integration variable over the outow PhE, the element of observed
bolometric ux becomes
dF =
dE
obs
d
t
a
=
3
8
_
R
0
R
e
_
6
0
R
2
0
dd
[1 (R
e
)]
4
t
a
, (E.5.9)
where (R
e
) =
_
1 R
2
0
/R
2
e
.
The observed ux as function of arrival time is represented in Fig. E.4 by
thick red curve. The characteristic raising and decaying time is
t = R
2
0
/(R
tr
c). (E.5.10)
There is no simple analytic expression describing the light curve. For large
enough t t the light curve has almost rectangular shape due to the fact
that its increase and decay times are much shorter than t.
The outow appears to a distant observer s a spot with size R
obs
=
_
R
2
0
(t
a
/c)
2
, for R
0
/c t
a
0. As soon as the PhE reaches the cor-
responding accelerating innitely long wind photosphere at t
a
= 0 the spot
size starts to increase almost linearly R
obs
R
0
+ ct
a
, see Fig. E.4. Finally,
as the innermost part of the outow reaches the transparency radius the spot
transforms to a ring with rapidly decreasing width and brightness.
E.5.3. Coasting phase
At the coasting phase the element of the outow has the following equation
of motion
r(t) = ct , (t) = = const. (E.5.11)
Arrival time of the photospheric emission (E.5.6) from that element
t
a
=
R
e
c
(1 ) +
c
(E.5.12)
124
E.5. Observed ux
0 50 100 150 200
0
1
2
3
4.0 2.0 0.0 2.0 4.0
0.0
0.2
0.4
0.6
0.8
1.0
1.2
xR
0
0.0 0.5 1.0 1.5 2.0
TT
0
y
R
0
t
a
t
F
F
m
a
x
Figure E.4.: Evolution of PhE and the light curve of photospheric emis-
sion (red thick curve) from the photon thick accelerating outow. Thick
black curve represents the photosphere of innitely long accelerating wind.
PhEs shown by thin curves correspond from right to left to arrival times
t
a
= (2
10
t, 2
9
t, . . . ), see (E.5.10). The surface = is given by
R
obs
= R
0
and it is shown by thick black line. Dashed curves show the max-
imal visible R
obs
for t
a
= (2
10
t, 2
9
t, . . . ) from bottom to top. The PhE at
that arrival times is a part of the wind photosphere limited by the correspond-
ing curves. Observed temperature of photospheric emission is illustrated by
color, see legend. Here R
tr
= 100R
0
, and t = 2t.
125
E. Photospheric emission from relativistic outows
depends on time of emission and as
t
e
=
R
e
(, ) +
c
. (E.5.13)
Entropy conservation at the coasting phase gives (see e.g. Piran et al.
(1993))
=
0
(R
0
/r)
8/3
. (E.5.14)
Change of variables can be made from to t
a
and the observed bolometric
ux from the element dV is then
dF =
dE
obs
d
t
a
=
=
3
8
R
8/3
0
d
c
4
0
R
10/3
e
d
_
t
a
c
_
4
R
e
c
d ln R
e
d
_
t
a
c
_
. (E.5.15)
The rst parenthesis in denominator is the consequence of relativistic beam-
ing, while the second one describes geometry of the photosphere. The ob-
served ux from the outow is obtained by integration over from 0 to
max
which is obtained from the position of the PhE at the line of sight, see Fig.
E.5.
Now we compute the photospheric emission from the coasting photon thin
and photon thick outows. In the case of the portion of wind (E.3.11)
max
can
be found analytically:
max
=
2(1 )
_
[t
a
(2 ) + D]
_
[t
a
(2 ) + D]
2
4t
2
a
(1 )
_
, (E.5.16)
where
D = n
0
r
2
0
/(2
2
)
2
. (E.5.17)
The function R
e
(, ) dened by (E.5.2) cannot be found in closed analytic
form. Nevertheless the ux can be evaluated numerically and it can be well
tted by the following expression
dF
_
_
0, t
a
< t
0
,
F
0
_
1 +
_
D
_
5/6
_ _
t
0
c
t
a
c
_
3
d, t
a
> t
0
,
(E.5.18)
126
E.5. Observed ux
R Rl
0.0
0.5
1.0
1.5
r
n
max
Figure E.5.: Optical depth along the line of sight (black solid line) and labora-
tory density prole (red dashed line) at the coasting phase for density prole
(E.3.11) with = 20 at t = const.
127
E. Photospheric emission from relativistic outows
where
t
0
() =
c
_
1 +
2
_
1 +
4D
1
__
(E.5.19)
is the minimal arrival time of photons emitted from the outow at given
along the line of sight. This form of pulses is generally referenced in the lit-
erature as FRED-like, i.e. fast rise, exponential decay pulses (Fenimore et al.,
1996), although decay is not exponential but has a power-law shape. Integra-
tion of (E.5.18) over gives the form of the light curve of the photospheric
emission.
The photon thin outow initially appears to a distant observer as a bright
spot with increasing radius
R
obs
= ct
a
(E.5.20)
and bolometric ux
F t
1/3
a
, (E.5.21)
see the raising part of the light curve in Fig. E.6. The peak ux is reached at
t
p
R
tr
2
2
c
, (E.5.22)
when the observer sees photons emitted from the innermost part of the out-
ow. Then the spot transforms to a ring with steeply decreasing thickness
and brightness resulting in the observed ux
F t
3
a
, (E.5.23)
see the decaying part of the light curve in Fig. E.6.
The observed photospheric emission of the photon thick outowstarts sim-
ilarly to the previous case, namely a spot with ux increasing as (E.5.21), due
to the presence of the outer photon thin layer. Then transition to the photon
thick asymptotics results in the corresponding change of ux to
F = F
max
_
1 (t
p
/t
a
)
2
_
, (E.5.24)
i.e. increase up to the saturation value F
max
L, see the raising part of the
light curve in Fig. E.7. As soon as the arrival time exceeds t
p
+ t the ux
starts to decrease rapidly
F t
2
p
_
1
(t
a
t)
2
1
t
2
a
_
, (E.5.25)
and for t
a
t the ux again behaves like (E.5.23), see the decreasing part
of the light curve in Fig. E.7.
The photon thick outow is observed initially as a spot with radius increas-
128
E.5. Observed ux
0 500 1000 1500 2000 2500
0
20
40
60
80
100
120
0.0 0.5 1.0 1.5 2.0 2.5
0.0
0.2
0.4
0.6
0.8
1.0
xR
0
1.0 2.5 4.0 5.5 7.0
TT
0
y
R
0
t
a
t
p
F
F
m
a
x
Figure E.6.: Evolution of PhE and the light curve (thick red curve) of photo-
spheric emission from the photon thin coasting outow. PhEs shown by thin
curves correspond from left to right to arrival times t
a
= (t
p
/5, 2t
p
/5, . . . ),
see (E.5.22). Thick black curves bounding PhEs correspond to the position of
transparency of the innermost part of the outow = l (lower curve) and the
surface = (upper curve). Observed temperature of photospheric emis-
sion is illustrated by color, see legend. Here = 1 10
3
.
129
E. Photospheric emission from relativistic outows
ing linearly in t
a
as (E.5.20), and then reaching its maximal size (E.4.3). Again
as the innermost part of the outow approaches the wind photosphere (E.4.2)
along the line of sight, the spot transforms to a ring, see Fig. E.7.
Similarly to the accelerating photon thick outow the light curve for t
t
p
has almost rectangular shape due to the fact that its increase and decay
times are much shorter than t.
E.5.4. Discussion
It should be noted that expression (E.5.23) is known in the literature for the
bolometric ux of instantaneously ashing shell (e.g. Fenimore et al. (1996);
Granot et al. (1999a,b); Woods and Loeb (1999)). It may be obtained from the
expression for detected spectral ux (see Granot et al. (1999b,a))
F
(t
a
)
_
0
2r
2
dr
_
1
1
d
2
j
/
/
(n
/
, r, t
a
+ r/c), (E.5.26)
where j
/
is the emissivity of the matter in its comoving frame (the energy
per unit time per unit volume per unit frequency per unit solid angle),
/
=
is the corresponding frequency of emission, andn
/
is the direction to the
detector in that frame. Assuming isotropic emissivity in the comoving frame
and integrating over the spectrum one arrives to the bolometric ux
F(t
a
)
_
0
2r
2
dr
_
1
1
d
3
_
0
j
/
(
/
, r, t
a
+ r/c)d
/
, (E.5.27)
and for instantaneously ashing shell with
_
j
/
(
/
, r, t
a
+ r/c)d
/
(r R)
_
t
a
+
r
c
R
c
_
(E.5.28)
expression (E.5.23) is recovered.
Actually (E.5.26) is incorrect. All the above mentioned works did not ac-
count for the transformation in the computation of observed ux from the
laboratory time t to the arrival time t
a
for dynamic emitter (see, e.g. page
141 of Rybicki and Lightman (1979)). Accounting for that effect we have an
additional multiplier (1 )
1
appearing in the integral (E.5.26) and corre-
spondingly in Eq. (3) of Granot et al. (1999a), that gives for the bolometric
ux after the integration over
F(t
a
) = F
0
t
4
a
,
R
c
(1 ) < t
a
<
R
c
(1 + ) . (E.5.29)
The same result can be actually obtained from (E.5.18). Analogous mistake
was done in the work of Woods and Loeb (1999), and all the subsequent arti-
130
E.5. Observed ux
0.0 0.5 1.0 1.5 2.0
0.00
0.04
0.08
0.12
0.16
0.0 0.5 1.0 1.5 2.0
0.0
0.2
0.4
0.6
0.8
1.0
xR
tr
0.0 0.5 1.0 1.5 2.0
TT
0
y
R
t
r
t
a
tt
p
F
m
a
x
Figure E.7.: Evolution of PhE and the light curve (thick red curve) of
photospheric emission of a photon thick coasting outow. PhEs shown
by thin curves correspond from left to right to arrival times t
a
=
(10
2
t
p
, 10
1.75
t
p
, 10
1.5
t
p
, . . . ), see (E.5.22). Thick black curves bounding
PhEs correspond to the position of innite wind photosphere (lower curve)
and the maximal visible angles at given t
a
(upper curve). Notice that these
angles exceed the relativistic beaming angle = , shown by dashed red
line. Observed temperature of photospheric emission is illustrated by color,
see legend. Here = 1 10
3
and t = 5t
p
.
131
E. Photospheric emission from relativistic outows
cles are suffering from the same problem, see e.g. Genet and Granot (2009).
Correct expressions for spectral and bolometric uxes were obtained by
Chiang and Dermer (1999) and by Rufni et al. (2002). Using transformation
laws for intensity of radiation from comoving to laboratory frame (see Equa-
tion before (381b) of Pauli (1958)) the expression for spectral ux instead of
(E.5.26) becomes
F
(t
a
)
_
0
2r
2
dr
_
1
1
d
3
j
/
/
(n
/
, r, t
a
+ r/c). (E.5.30)
Finally, we discuss the applicability of our adiabatic approximation. Be-
loborodov (2011) considered the photospheric emission from innitely long
wind both at acceleration and coasting phases and solved the correspond-
ing radiative transfer equation. His main conclusion is that in addition to
usual relativistic beaming leading to anisotropy of radiation in laboratory
frame (E.5.4), in the coasting wind there develops another anisotropy in the
comoving frame of the outow. This comoving anisotropy results from the
fraction of photons which already underwent their last scattering in the bulk
photon eld of the outow. The anisotropy of such photons grows with in-
creasing radius for geometrical reasons. Since the amount of such photons
increases with radius the intensity of the entire photon eld becomes increas-
ingly anisotropic. Such phenomenon is referred to as fuzzy photosphere.
We expect this additional anisotropy in the case of photon thick coasting
outows. In other words, in photon thick outows the probability distribu-
tion of last scattering positions is wide in radius and angles, see also Peer
(2008). For photon thin outows instead such effect is not relevant. It is pos-
sible to show that practically all photons leave the outow before the above
mentioned anisotropy develops. It means that the corresponding probability
distribution is sharp in radius.
In order to describe the late-time photospheric emission of switching off
relativistic wind Peer (2008) considered emission from a single differential
layer. He proposed a model for calculation of observed ux and spectrum
based on the probability density function describing the last scattering of
photons. Actually he computes not the traditional energy ux understood
as energy crossing unit area in unit time, but photon ux as number of pho-
tons crossing unit area in unit time. For this reason his decay law for photon
ux at late times is
F
ob
(t
a
) t
2
a
. (E.5.31)
Lorentz transformation of the photon energy from the comoving frame to the
laboratory one results in additional multiplier (1 )
1
in the energy ux
that leads to the observed ux F t
3
a
, which agrees with our (E.5.18).
We conclude that adiabatic approximation used in our computation is
clearly applicable for photon thin outows due to the presence of well de-
ned, sharp last scattering surface. As for photon thick outows our sim-
132
E.6. Instantaneous and time-integrated spectra from the photosphere
ple adiabatic approximation gives the same results for the bolometric ux
as more accurate treatments of radiation transfer (Beloborodov, 2011) and of
probability density function (Peer, 2008).
E.6. Instantaneous and time-integrated spectra
from the photosphere
In the adiabatic approximation the observed temperature T
obs
of the PhE de-
pends on angle and arrival time t
a
:
T
obs
(t
a
, R
e
()) =
T
c
(R
e
())
(1 )
, (E.6.1)
where T
c
(R
e
) is the comoving temperature on the PhE. When temperature
dependence on radius is given, the distribution of observed temperatures
across the PhE may be obtained from (E.6.1).
Luminosity of photospheric emission is proportional to the volume of the
outow that becomes transparent to radiation at angle between t
a
and t
a
+
dt
a
, i.e. to
v
b
=
d
dt
a
=const
, (E.6.2)
which we will refer to as the blooming velocity. Finally, assuming thermal
spectrum in the comoving frame the differential spectral ux is
dF
=
dE
obs
ddt
a
=
2h
c
3
3
d
exp
_
h
kT
obs
_
1
2R
2
e
v
b
d. (E.6.3)
Blooming velocity v
b
can be obtained from equations of motion of PhE and
that of the outow. In the case of photon thick outow it is ultrarelativistic
and is approximately given by the velocity of the outow itself since its PhE
quickly becomes almost static. From (E.5.12) and (E.5.14) for transparency at
accelerating phase we have
T
obs
(t
a
, ) =
_
R
0
R
e
_
2
T
0
1
_
1
_
R
0
R
e
_
2
, (E.6.4)
while at coasting phase
T
obs
(t
a
, ) =
_
R
0
R
e
_
2/3
T
0
(1 )
=
T
0
R
2/3
0
R
1/3
e
(t
a
c )
. (E.6.5)
133
E. Photospheric emission from relativistic outows
E.6.1. Photon thin outow
Instantaneous spectrum of a photon thin outow integrated over PhE is very
close to the thermal one for both accelerating and coasting phases. At ac-
celerating phase observed temperature does not vary with arrival time. At
coasting phase observed temperature decreases as
T t
2/3
a
, (E.6.6)
while emitting area increases as
A t
3
a
, (E.6.7)
leading to the increase of ux (E.5.21). For t
a
> t
p
the emitting area starts to
decrease
A t
1/3
a
, (E.6.8)
leading to diminishing ux as described by (E.5.23).
It can be seen from Fig. E.6, that temperature across a given PhE is practi-
cally constant. Actually, the radial coordinates of the PhE change only twice
between = 0 and = 1/. That is converted by (E.6.5) to even smaller tem-
perature interval: from the temperature at the line of sight T
LOS
to T
LOS
/
3
2
at = 1/.
Integrating these instantaneous spectra over t
a
a non thermal spectrum is
obtained. It may be represented as Band function (Band et al., 1993) with low-
energy power law index 0.5 and high-energy power law index = 4,
see Fig. E.8. There is an additional cutoff in that spectrum at a very high
frequency corresponding to initial temperature of the outow T
0
.
Evolution of instantaneous spectrum with arrival time depends on as-
sumed outow density prole (E.3.11). Time evolution of observed tempera-
ture is determined by entropy conservation so it does not depend on assumed
density prole. But emitting area dependence on arrival time actually does
depend on it due to change in the blooming velocity. Low-energy power law
index is quite insensitive to the form of both density and temperature pro-
les. However high-energy part of the time-integrated spectrum of photon
thin outows strongly depends on it. In particular, for power law density
proles (E.2.6) with slope the high-energy part of the spectrum is
F
3
1+
. (E.6.9)
The resulting parameter of the Band spectrum may vary from (expo-
nential cutoff) for = 1 to 1 for = .
134
E.6. Instantaneous and time-integrated spectra from the photosphere
0.01 1 100 10
4
10
6
0
10
15
10
12
10
9
10
6
0.001
1
F
_
1
T
tr
_
3E
0
R
0
4a
_
1/4
= 4.4 10
10
(E
54
R
8
)
1/4
cm,
E
54
4.8 10
20
B
4
2
R
1
8
,
_
E
0
BR
0
24m
p
c
2
_
1/3
= 1.8 10
12
(E
54
B
2
R
8
)
1/3
cm,
4.8 10
20
B
4
2
R
1
8
E
54
3.2 10
8
B
4
2
R
2
8
,
E
0
B
3
8m
p
c
2
R
0
= 1.8 10
17
E
54
B
3
2
R
1
8
cm,
3.2 10
8
B
4
2
R
2
8
E
54
1.1 10
5
B
5
2
R
2
8
,
_
E
0
B
4m
p
c
2
_
1/2
= 5.9 10
14
(E
54
B
2
)
1/2
cm,
E
54
1.1 10
5
B
5
2
R
2
8
.
(E.7.1)
For very small baryonic loading, or in other words for a pure electron-
positron plasma, the transparency radius does not depend on B parameter,
then it increases as B
1/3
(accelerating photon thick solution), then it steeply
increases as B
3
(coasting photon thick solution), and nally it increases as
B
1/2
(coasting photon thin solution), see Fig. E.12. In Fig. E.12 we also
show as function of the baryonic loading parameter the following quanti-
ties computed at the transparency radius: the Lorentz factor, the observed
and comoving LOS temperatures, fraction of energy emitted from the pho-
tosphere to the total energy, for different values of the total energy E
0
. It is
clear that the highest Lorentz factors at transparency radius are attained in
138
E.7. Application to GRBs
photon thick asymptotics. The largest transparency radii are reached instead
in photon thin asymptotics.
Fig. E.11 shows the energy-baryonic loading diagram, where the regions of
validity of the asymptotics discussed above are indicated explicitly for typical
parameters of GRBs. For all the relevant range of GRBs parameters 10
48
erg <
E
0
< 10
55
erg and 10
6
cm < R
0
< 10
12
cm all four asymptotics are present in
the interval 10
10
< B < 10
1
.
Using (E.3.12) we obtain for the electron-positron contribution to opacity
n
/n = 2.5 10
5
R
3/4
8
B
1
2
E
1/4
54
, (E.7.2)
after their decoupling, so it is clear that this contribution may be safely ne-
glected for, say, B > 10
4
.
In photon thin case, expressing the peak arrival time (E.5.22) in units of
total energy and baryonic loading, we nd
t
p
1E
1/2
54
B
5/2
2
s, (E.7.3)
This expression gives an estimate for duration of photospheric emission in
arrival time, and we conclude that typically this duration should be smaller
than one second, given that typically measured bulk Lorentz factors exceed
10
2
. Notice that t
p
does not depend on R
0
and, consequently, on the light
crossing time of the outow. Currently time resolved spectra of GRBs with
good statistics cannot resolve such small time scales, and therefore we expect
that only time integrated spectra may be observed from GRBs which have
their parameters corresponding to the photon thin case. Thus, starting from
comoving thermal spectrum for the photospheric emission we obtain for the
rst time an observed spectrum which may be well described by the Band
function with high energy power law index being determined by the den-
sity prole of the outow. We nd this result quite remarkable. Notice that
the idea of convolution over time has been introduced for GRBs by Blinnikov
et al. (1999). Double convolution over EQTS and arrival time is also one of
the key ideas in the reshell model (Rufni et al., 2003a).
It is even more remarkable that GRBs appear to be the only objects in nature
able to reach the photon thin asymptotics in their ultrarelativistic expansion.
For thermally accelerated relativistic plasmas which are discussed in connec-
tion with their possible synthesis in ground based laboratories (e.g. Rufni
et al. (2010)) it is unreachable. In order to quantify what should be initial op-
tical depth
0
=
T
n
0
l in order to reach the photon thin asymptotics one may
use (E.3.23) to get
0
4
4
= 4 10
8
B
4
2
. (E.7.4)
139
E. Photospheric emission from relativistic outows
GRBs clearly satisfy this constraint with the contribution of baryons only
0
=
3
T
BE
0
4m
p
c
2
l
2
10
14
E
54
B
2
R
2
8
. (E.7.5)
which is increased even further by the presence of electron-positron pairs.
In contrast, in photon thick case the corresponding observed spectrum, as it
is well known in the literature, is a black body with small deviations present
at both low and high energy parts of the spectrum. It remains close to ther-
mal also when variation of the Lorentz factor through the outow is present
(Daigne and Mochkovitch, 2002).
When the outow becomes transparent in the transition from photon thick
to photon thin conditions, the observed spectrum will contain the Band com-
ponent produced by the photon thin layer, with an almost black body coming
from the photon thick part superimposed. As we already mentioned, in any
outow there is always a photon thin layer. So when the photon thick part
becomes dominant, the spectrum is expected to be dominated by the black
body component. This may be the reason why in most cases analysed by
Ryde and Peer (2009) there are both power law and black body components:
transparency occurs at the boundary between photon thin and photon thick
conditions.
In Sec. E.5.3 we give analytic expressions for the photon ux for the sim-
ple model of the portion of wind (E.3.11). With more complex density pro-
le composed of presumably many shells the light curve is expected to be
variable and arbitrarily complex. It is necessary to emphasize however, that
the decaying part of the light curve is determined solely by the geometry
of the limiting surfaces in photon thick (E.4.2) and photon thin (E.4.4) cases,
respectively. So when the photon ux is observed with faster than t
3
de-
cay it should be concluded that the emission does not come from the pho-
tosphere. The photospheric emission may be additionally identied by the
spectral analysis. What we have shown here, though, is that the observed
spectrum may not necessarily be close to thermal, but in the photon thin case
it may look very different.
E.7.2. Wind model
In the case of gradual energy release resulting in relativistic wind an addi-
tional parameter is present, that is the duration of energy release, which we
parametrize as t = t
1
s. Instead of the total energy E
0
the luminosity
L = 10
50
L
50
erg/s will be used.
From formulae (E.3.9), (E.3.19), and (E.3.25) we obtain for the transparency
140
E.7. Application to GRBs
10
4
10
3
10
2
10
48
10
50
10
52
10
54
10
4
10
3
10
2
10
48
10
50
10
52
10
54
B
E
0
,
e
r
g
Accelerating
Photon Thick
Coasting
Photon Thin C
o
a
s
t
i
n
g
P
h
o
t
o
n
T
h
i
c
k
Figure E.11.: The energy-baryonic loading diagram showing the validity of
the various asymptotic solutions for the transparency radius for typical pa-
rameters of GRBs with l = R
0
= 10
8
cm.
141
E. Photospheric emission from relativistic outows
radius
R
tr
=
_
_
1
T
tr
_
LR
2
0
4ca
_
1/4
= 8.1 10
8
L
1/4
50
R
1/2
8
cm,
L
50
5.3 10
15
B
4
2
R
2
8
,
_
LBR
2
0
24m
p
c
3
_
1/3
= 1.3 10
10
_
L
50
B
2
R
2
8
_
1/3
cm,
5.3 10
15
B
4
2
R
2
8
L
50
9.8 10
2
B
4
2
R
8
,
LB
3
8m
p
c
3
= 5.9 10
10
L
50
B
3
2
cm,
9.8 10
2
B
4
2
R
8
L
50
10
5
B
5
2
t
1
,
_
LBt
4m
p
c
2
_
1/2
= 5.9 10
10
(L
50
t
1
B
2
)
1/2
cm,
L
50
10
5
B
5
2
t
1
.
(E.7.6)
In Fig. E.13 we show as function of the baryonic loading parameter the
following quantities computed at the transparency radius: the Lorentz factor,
the observed and comoving LOS temperatures, fraction of energy emitted
from the photosphere to the total energy, for different duration of the wind
with the total energy E
0
= 10
51
erg, and inner boundary radius R
0
= 10
8
cm.
Wind duration ranges from 10 ms to 10 s. The corresponding wind lumi-
nosity varies from 10
53
erg/s to 10
50
erg/s. Fig. E.14 shows the luminosity-
baryonic loading diagram where the regions of validity of the asymptotics
discussed above are indicated. Photon thin asymptotics is not common for
the typical parameters of wind considered in the literature, but still can be
relevant for energetic winds of short duration.
All the discussion on the complexity of light curves for shell models is valid
for wind models as well. Due to smaller range of applicability of photon
thin asymptotics we expect that relativistic winds produce spectrum which
is close to thermal one.
Generally speaking, for both shell and wind models with more complex
number density, energy density and Lorentz factor distributions within the
142
E.7. Application to GRBs
10 8 6 4 2
5
4
3
2
1
0
1
2
3
4
5
6
7
8
1
2
3
4
10
11
12
13
14
15
16
5
4
3
2
1
0
1
2
3
4
5
6
7
8
1
2
3
4
10
11
12
13
14
15
16
log
10
B
l
o
g
1
0
E
P
h
E
0
l
o
g
1
0
T
t
r
,
l
o
g
1
0
T
o
b
s
,
e
V
l
o
g
1
0
t
r
l
o
g
1
0
R
t
r
,
c
m
10
49
erg
10
55
erg
10
55
erg
10
49
erg
10
49
erg
10
55
erg
10
49
erg
10
55
erg
10
49
erg
10
55
erg
Figure E.12.: Transparency radius R
tr
, Lorentz factor
tr
, observed T
obs
and
comoving T
tr
temperatures, and ratio of the energy emitted to the total energy
E
Ph
/E
0
at transparency radius as functions of baryonic loading B for shells
with different total energy E
0
but the same width l = R
0
= 10
8
cm. All
four regimes with different asymptotics are clearly visible and dashed lines
corresponding to their domain of validity from Eq. (E.7.1) are shown. Curves
are drawn for E
0
equal to 10
49
erg (green), 10
51
erg (blue), 10
53
erg (violet),
and 10
55
erg (red).
143
E. Photospheric emission from relativistic outows
10 8 6 4 2
5
4
3
2
1
0
1
2
3
4
5
6
7
8
1
2
3
9
10
11
12
13
14
15
5
4
3
2
1
0
1
2
3
4
5
6
7
8
1
2
3
9
10
11
12
13
14
15
log
10
B
l
o
g
1
0
E
P
h
E
0
l
o
g
1
0
T
t
r
,
l
o
g
1
0
T
o
b
s
,
e
V
l
o
g
1
0
t
r
l
o
g
1
0
R
t
r
,
c
m
10 s
0.01 s
10 s
0.01 s
10 s
0.01 s
10 s
0.01 s
10 s
0.01 s
Figure E.13.: The same as in Fig. E.12 for winds with different duration, but
the same total energy E
0
= 10
51
erg and radius of origin R
0
= 10
8
cm. All
four regimes with different asymptotics are clearly visible. Curves are drawn
for t from to 10
2
s (green) to 10 s (red) in steps of one order of magnitude.
144
E.7. Application to GRBs
10
4
10
3
10
2
10
48
10
49
10
50
10
51
10
52
10
53
10
54
10
4
10
3
10
2
10
48
10
49
10
50
10
51
10
52
10
53
10
54
B
L
,
e
r
g
s
Accelerating
Photon Thick
C
o
a
s
t
i
n
g
P
h
o
t
o
n
T
h
i
c
k
Figure E.14.: The luminosity-baryonic loading diagram showing the valid-
ity of the various asymptotic solutions for transparency radius of wind with
duration t = 0.1 s. Notation is the same as on Fig. E.11.
145
E. Photospheric emission from relativistic outows
outow various light curves and observed spectra of photospheric emission
may be obtained, with the geometrical constraints discussed in previous sec-
tions. The theory developed in this paper may be used for the calculation
of these observed quantities using the corresponding proles obtained from
e.g. hydrodynamic simulations. We plan to report on this study elsewhere
(Rufni et al., 2011).
E.8. Conclusions
In summary, in this paper we readdressed the issue of photospheric emission
from spherically symmetric relativistic outows. Two types of outows are
discussed: winds resulting from gradual energy release, and shells produced
by instant energy release. We evaluated the optical depth and the correspond-
ing photospheric radius for each type of the outow. We demonstrated that
there are two asymptotic solutions for a coasting shell, similarly to a coasting
wind. In particular we showed that due to ultrarelativistic motion a geomet-
rically thin shell may appear as thick for photons propagating inside it. For
this reason we suggest to refer to any relativistic outow as photon thick,
when the photon spends enough time inside it to feel its density decreasing
with time or radius. In the opposite case we refer to the outow as photon
thin.
We also studied geometry and dynamics of photospheres of relativistic out-
ows. As we are interested in appearance of the photosphere to a distant ob-
server, we introduced the notion of photospheric equitemporal surface, the
PhE. While in general the PhE is dynamically evolving, we showed that it
is well approximated by a part of an ellipsoid in the case of the photon thin
outow. The photon thick PhE initially looks the same, due to the presence of
photon thin layer in any outow. Then it becomes concave and approaches
asymptotically the photosphere of innitely long wind.
Then, assuming that photons are trapped in the outow with thermal and
isotropic distribution for optical depth greater than one, and are release to
freely stream when opacity decreases below one, we computed both photon
ux and observed spectra of photon thick and photon thin outows. Our
adiabatic approximation corresponds to an assumption that the last scatter-
ing surface of photons is well dened. This assumption is shown to work in
the photon thin case. As for photon thick case, our results for observed ux
are shown to agree with other results in the literature, obtained with more
accurate methods (Beloborodov, 2011), see also (Peer and Ryde, 2011).
We nd that instantaneous spectra of both photon thick and photon thin
outows are close to the thermal one. This result is in agreement with the
corresponding results in the literature for relativistic winds, since such winds
with long enough duration must indeed be photon thick. However, the time
integrated spectrum from photon thin outows appears to be non thermal,
146
E.8. Conclusions
due to strong temperature and luminosity evolution in arrival time. This
spectrum may be described by the Band function (Band et al., 1993) with
some specic low energy and high energy power law indices, and an ex-
ponential cutoff corresponding to the temperature in the source of energy
release.
Our results are then applied to GRBs. We show in particular, that instanta-
neous spectra cannot be observed in the case of photon thin outows, and we
predict therefore that typical observed spectrum originating from the photo-
spheric emission should be the Band spectrum. Our analysis for simple den-
sity prole gives values for the low energy power law index 0.5 and the
high-energy power law index = 4, which are different from the typical
values observed in GRBs. The low energy index is insensitive to both density
and temperature proles adopted. The high energy index strongly depends
on the density prole, and it may vary from to 1.
147
E. Photospheric emission from relativistic outows
148
F. Correlation dynamics in
cosmology
Introduction
According to the modern cosmology gravitational instability of dark matter
is the main mechanism for generation of large-scale structure of the Universe,
see e.g. Dodelson (2003). This structure is thought to be formed dynamically
from initial Gaussian seeds seen in cosmic microwave background radiation
e.g. by WMAP satellite Komatsu et al. (2011) all the way to the present day
observed galaxies, clusters and superclusters of galaxies along with voids as
seen by galaxy surveys such as 2dF Cole et al. (2005) and SDSS Percival et al.
(2007).
While the structure around us up to superclusters scales is strongly in-
homogeneous, the distribution of matter becomes homogeneous on larger
scales Hogg et al. (2005),Yadav et al. (2005). This point supports the cosmo-
logical principle which is needed to match theoretical cosmology with well
established background dynamics to observations. It is in fact believed that
even highly nonlinear structures seen on scales from galaxies to clusters do
not inuence global dynamics expressed by the (simple) Friedmann back-
gound solution of (complex) Einstein equations. However, phenomenolog-
ical CDM model, being in agreement with observations, lacks predictive
power and cannot neither explain values of numerous cosmological param-
eters nor answer more fundamental questions such as what is the nature of
recently discovered cosmic acceleration. There is growing concern in the lit-
erature Wiltshire (2007),Mattsson (2010),Ellis (2008) that the picture described
above is an adequate one. We believe that in the epoch when the Universe is
dominated by highly nonlinear structures the concept of background space-
time dynamics may have limited physical meaning. Dynamical theory of
structure formation is required, but not yet developed.
Standard cosmological structure formation model distinguishes two dif-
ferent physical regimes. Initially uctuations of matter density are small at
all scales and they are considered as linear perturbations to a background
homogeneous isotropic solution. This linear regime of structure formation
is well understood. On the contrary, the second, nonlinear, regime of the
structure formation lacks a self-consistent theoretical treatment. The only
existing approaches are strictly limited to N-body simulations Davis et al.
(1985),Springel et al. (2005),Klypin et al. (2011). Recent study Einasto et al.
(2011) of the large-scale structure using wavelet technique provides impor-
149
F. Correlation dynamics in cosmology
tant insights into clustering phenomenon. However, such approach is not
suitable for dynamical description of structure formation.
The key concept of the structure formation theory is that of the average
matter density which is assumed to be a source for the underlying back-
ground spacetime cosmological solution. Therefore, an averaged description
of the cosmological evolution is a critical issue in cosmology, see also Sylos
Labini et al. (2009). A consistent covariant spacetime averaging of the gravi-
tational eld within this approach naturally leads to appearence of the gravi-
tational correlation terms which modify the structure of the averaged matter
source Zalaletdinov (1992),Zalaletdinov (1993),Zalaletdinov (1997).
In this connection it is important to stress that observed nonlinear struc-
tures are usually described not by dark matter density contrast, but by its
spatial correlations as observed e.g. in the two-point correlation function (r)
Peebles (1993). This two-point correlation function is interpreted as manifes-
tation of the fractal structure Rufni et al. (1988). There is no satisfactory
dynamical theory of cosmological correlations describing in particular evo-
lution of the correlation function (r, t).
The aim of this Letter is to propose a possible direction for the development
of the theoretical model which describes the evolution of the Universe from
the beginning of the gravitational instability up to the currently observed
nonlinear structures which is spatially homogeneous and isotropic on large
scales as dynamics of gravitational correlations.
Basic equations and initial conditions
Dark matter is assumed to consist of weakly interacting massive particles
which are non-relativistic when decouple from the thermodynamic equilib-
rium with cosmic plasma. Therefore, it should be described at kinetic level,
as collisionless self-gravitating system of particles. On scales of interest, from
galaxies to superclusters, curvature effects are negligible and Newtonian
approximation is adopted. The fundamental equations consist of coupled
Boltzmann-Vlasov and Poisson equations Binney and Tremaine (1987),Lon-
gair (1998)
f
t
+ v
f
r
f
v
= 0, (F.0.1)
2
= 4G
_
f d
3
v, (F.0.2)
where there are two unknown functions: f (r, v, t) is the distribution function
of dark matter particles and (r, t) is the gravitational potential; r and v are
phase space coordinates, G is Newtons constant.
Initially the distribution function is assumed to be Maxwellian with small
uctuations f (r, v, t) = f
0
([v[) [1 + (r, v, t)], 1. These perturbations
are assumed to be Gaussian, having Harrison-Zeldovich power spectrum
[(k, t
0
)[
2
= Ak, where A is a normalization constant, k is a wavenumber,
150
(r, v, t) =
V
(2)
3
_
(k, v, t) exp(ik r) d
3
k. In comoving coordinates the
distribution function depends implicitly on time due to cosmic expansion.
Distribution function, spatial averaging and coarse graining
The whole dynamics can, in principle, be extracted from the solution of
(F.0.1),(F.0.2) or equivalent N-body problem Ma and Bertschinger (2004). In
practice, however, analytic solution of the system (F.0.1),(F.0.2) is hardly pos-
sible, while N-body approach provides limited information.
On the linear phase of the structure formation, when < 1, uctuations of
the distribution function grow essentially in the conguration space since
(r, v, t)
v
v
(r, v, t)
r
, (F.0.3)
This phase is well studied and is described in a hydrodynamic-like ap-
proximation Gilbert (1966),Stewart (1972),Bisnovatyi-Kogan and ZelDovich
(1971), where the term
1
f
0
f
0
v
r
plays the role of the source term in linearized
equation (F.0.1). In the short wavelengths limit uctuations experience Lan-
dau damping, while in the opposite limit perfect uid oscillatory behav-
ior is recovered. The critical wavelength separating these two regimes, is
k
J
(m) = 4Gm
_
+
1
v
f
0
v
dv. This characteristic scale depends on parti-
cles mass. However, candidates for CDM particles have so large masses
m >GeV that their Jeans mass is irrelevant for cosmological structure forma-
tion. In contrast, during nonlinear phase, when > 1, the condition (F.0.3)
does not hold, distortions of the distribution function in velocity space be-
come signicant, so that the distribution function is driven out of equilib-
rium. This out of equilibrium function is then subject to the mean eld re-
laxation Lynden-Bell (1967),Severne and Luwel (1980),White (1996),Arad and
Lynden-Bell (2005),Chavanis and Bouchet (2005) in strongly time-varying
gravitational potential. From the one hand, during nonlinear phase equa-
tion (F.0.1) is still satised, but it becomes untractable since f (r, v, t) remains
always out of equilibrium and time dependent. From the other hand, the
coarse-grained distribution function relaxes locally to a stationary function
f
LB
(r, v) characterizing Lynden-Bell equilibrium. Relaxed distribution func-
tion f
LB
(r, v) is thought to describe virialized objects which are in deep non-
linear regime, such as elliptical galaxies, however it certainly does not apply
to unvirialized object such as superclusters of galaxies, which are still in lin-
ear regime today.
Existence of several types of objects with the corresponding different scales,
which are in different stages of their evolution calls for development of a new
dynamical scheme in the theory of structure formation. This scheme should
be designed in order to disentangle two manistations of the gravitational
interaction in collisionless system of particles: gravitational instability and
mean eld relaxation.
151
F. Correlation dynamics in cosmology
Important observation, which helps to construct such a scheme, is that
the distribution function f is not directly accessible to measurements Bin-
ney and Tremaine (1987). Potentially measurable are, instead, mass density
m
_
f d
3
v, m is the mass of dark matter particle, as well as the grav-
itational potential averaged on some spatial scale d. Consequently, key
dynamical variables are spatially averaged distribution function and gravita-
tional potential
f (r, v, t)
1
V
S
_
S
f (r + r
/
, v, t)d
3
r
/
, (F.0.4)
(r, t)
1
V
S
_
S
(r + r
/
, t)d
3
r
/
, (F.0.5)
where V
S
is the volume space region S (see Zalaletdinov and Coley (2002),
Zalaletdinov and Coley (2002) for dention and properties of space volume
average tensor elds). A compact space regions S of the space volume aver-
aging is determined by its characteristic size d, so that the phase space cell
volume is V
S
d
3
.
Substituting (F.0.4) and (F.0.5) into (F.0.1),(F.0.2) one nds
f
t
+ v
f
r
r
f
v
= , (F.0.6)
2
= 4G
_
f d
3
v, (F.0.7)
where the right hand side of the averaged Boltzmann equations contains now
a correlation operator, dened as
_
f
v
_
r
f
v
. (F.0.8)
The correlation operator depends on the averaging scale d as a free parameter.
While equations (F.0.6),(F.0.7) are consistent with the linear phase of struc-
ture formation, where the correlation operator is negligible, the spatial
averaging (F.0.4),(F.0.5) loses the statistical meaning in the nonlinear phase
since the distribution function f (r, v, t) is driven out of equilibrium. In or-
der to restore the statistical meaning of the spatial averaging, additional step
is required. It is proposed that the dynamical variables should be further
152
coarse-grained
f (r, v, t)
_
1
V
f (r + r
/
, v + v
/
, t)
_
d
3
r
/
d
3
v
/
, (F.0.9)
(r, t)
1
V
(r + r
/
, t)
_
d
3
r
/
. (F.0.10)
The properties of a typical cell of the coarse grained phase space are deter-
mined by its characteristic size in coordinate space scale
x
and velocity
space scale
v
, so that the phase space cell volume is V
(
x
)
3
(
v
)
3
.
The fundamental relation between the three characteristic space scales,
x
, d and L which represents a coarse-grained self-gravitating collisionless
system of particles under a space volume averaging is given by the relation
x
d L. (F.0.11)
This is conceptual relaization of description for a classical physical system
with the microscopic scale
x
and the macroscopic scale L.
For a scalar quantity p(r, v, t) dened on the 6-dimensional phase-space ,
the operations of the space volume averaging over a compact space region S
and the coarse graining over a phase-space cell commute, p = p. This
fact is expected to play a fundamental role in the theory of structure forma-
tion. It is worth noting that the integration over velocity space also commutes
with the above mentioned operations. This property readily permits building
up moments of the distribution function
f
_
.
As the result of application of the proposed averaging scheme, dark mat-
ter is assumed to be described by the coupled space-volume averaged and
coarse-grained system of the Boltzmann-Vlasov and Poisson equations
f
_
t
+ v
f
_
r
r
f
_
v
=
, (F.0.12)
2
= 4G
_
f
_
d
3
v. (F.0.13)
The structure of the correlation operator, being a functional of the coarse
grained gravitational potential and distribution fuction
_
,
f
, and depend-
ing on two free parameters
x
and d, completely determines dynamics of
structure formation on both linear and nonlinear phases.
Structure formation as correlation dynamics
The main task in the scheme presented above is the construction of the cor-
relation operator . An attempt to study correlation dynamics via BBGKY
hierarchy Prigogine and Severne (1966),Fall and Saslaw (1976) fails since un-
like a plasma, weak coupling approximation cannot be adopted for correla-
tions in self-gravitating system. Other assumptions such as simple relations
153
F. Correlation dynamics in cosmology
between three-point and two-point correlation functions Davis and Peebles
(1977) do not work. There are two possible approaches in counstructing the
correlation operator. Firstly, it can be found from rst principles. Secondly, a
phenomenological approach can be followed. The structure of the correlation
operator has been studied by applying the rst and second laws of thermody-
namics and attempting a representation by means of Fokker-Planck diffusion
coefcients Chavanis and Bouchet (2005). In such an approach interpretation
of the diffusion coefcients should be given.
On the linear phase of structure formation is negligible, since uctu-
ations are assumed to be Gaussian, and equations (F.0.6),(F.0.7) reduce to
(F.0.1),(F.0.2). Therefore, linear theory of structure formation is recovered in
the proposed scheme as the limit in which the correlation operator (F.0.8) van-
ishes and
f
_
f . As expected, both spatial averaging and coarse-graining
have no role since gradients of density and gravitational potential are small,
along with time variations of the gravitational potential.
Spatial correlations, represented by the correlation operator , become
important once uctuations grow to become nonlinear on the dynamical
timescale (G)
1/2
. Since mean eld relaxation operates essentially on
the same timescale, description of nonlinear structures should be given in
terms of
f
_
, satisfying equations (F.0.12),(F.0.13). It is then natural to identify
the scale where uctuations reach nonlinearity R
NL
(t) with the microscopic
scale
x
R
NL
(t). (F.0.14)
We stress that the dynamics of ne-grained distribution function f remains
essentailly of non-equilibrium nature. Therefore, only
f
_
can reach station-
ary form f
LB
(r, v) which results from smearing out microscopic evolution of
the distribution function f on scales l <
x
, v <
v
. Severne and Luwel
Severne and Luwel (1980) in an attempt to construct dynamical theory of col-
lisionless relaxation, derived an equation for
f by linearizing (F.0.12),(F.0.13).
Instead of our spatial averaging they used an ensemble averaging. Their
equation, containing the source term proportional to
f
3
, gives as a solution
the Linden-Bell distribution function Kadomtsev and Pogutse (1970).
There are two issues here of fundamenal signicance. Firstly, upon a mean
eld relaxation the distribution function remain inhomogeneous in space,
that is, the particles of a self-gravitating collisionless system should be ex-
pected to form a stationary spatially distributed structure. Secondly, such
a stationary spatial structure must be of universal nature since there is no
separation by masses of particles by virtue of their absence in the Boltzmann-
Vlasov-Poisson equations. The universal character of simulated dark matter
halo proles Navarro et al. (1997) velocity distributions Merrall and Henrik-
sen (2003), observed two-point correlation function Jones et al. (2004) and
numerical experiments Levin et al. (2008) seems to give a clear evidence of
that fundamental phenomenon. Both these facts should be reected in the
154
form and structure of the correlation operator.
The phenomemon of mean eld relaxation is usually attributed to the for-
mation process of elliptical galaxies. Unlike a single galaxy, the large scale
structure is thought to originate as the interrelated process of hierarchical
halo formation and subsequent merging Press and Schechter (1974), the lat-
ter operate simultaneously with the mean eld relaxation. In other words,
at scales where uctuations of dark matter density are nonlinear the gravita-
tional instability ends and dark matter experiences mean eld relaxation. It
is important to stress therefore, that both gravitational instability and mean
eld relaxation are generic, universal features of collisionless self-gravitating
matter. They are just different manifestations of the same physical law. De-
spite many models of nonlinear clustering have been proposed, for review
see Jones et al. (2004),Bernardeau et al. (2002),Malik and Wands (2009) up to
now there is no model bridging the gap between these two phenomena. The
scheme proposed in this paper naturally closes this gap, suggesting unied
description of the structure formation, from initial small perturbations all the
way to presently observed structures.
We were focusing on the Newtonian cosmology in this paper, giving useful
insight into mechanisms underlying structure formation. However, in such
an approach there remain several limitations. In particular, 1) Jeans swin-
dle should be adopted; 2) average density cannot be dened and should be
treated as free parameter; 3) effects of General Relativity, and topology are
neglected.
Conclusions
Dynamical theory of structure formation is still missing. Various ap-
proaches to nonlinear structure formation can capture limited number of phe-
nomena and do not provide a unied picture of the structure formation from
seed perturbations all the way to the presently observed large scale struc-
ture. It is generally accepted that on such large scales the only important
interaction is the gravitational one. In this work we propose to generalize the
widely accepted treatment of large scale clustering based on Vlasov-Poisson
equations for collisionless dark matter particles, by combining two averaging
procedures: spatial averaging and coarse graining. We show that in this way
both phenomena of gravitational instability and mean eld relaxation nd
natural and unied description.
155
F. Correlation dynamics in cosmology
156
G. Semidegenerate self-gravitating
system of fermions as a model
for dark matter halos and
universality laws
G.1. Introduction
Dark matter properties are constrained both from cosmology and astro-
physics. Cosmological bounds are based on the assumptions on cross sec-
tions of interaction between the dark matter particles and usual matter plus
interaction between dark matter particles themselves. Usually it appears that
dark matter decouple from normal matter at thermal equilibrium and at the
same time or earlier interaction between dark matter particles themselves
come to a halt, so that they form collisionless dark matter. In this case mass
density of the particles can be found and compared to the known dark mat-
ter cosmological density. This provides different limits on the mass of dark
matter particles, two of them being Gershtein-Zeldovich limit for the sum of
neutrino masses (Gershtein and ZelDovich, 1966)
m < (G.1.1)
Different approach to the study of properties of dark matter is coming from
cosmology and specially from theories of galaxy formation and evolution.
While studies of CMB allow to constrain number and masses of light neu-
trinos representing hot dark matter (Giusarma et al., 2011), properties and
masses of warm and cold dark matter are constrained by the total mass den-
sity of the Universe (Lee and Weinberg, 1977) and galactic halos structure
(Tremaine and Gunn, 1979) and substructure (Polisensky and Ricotti, 2011).
The Lee-Weinberg bound (Lee and Weinberg, 1977) limits the mass of dark
matter particles with given coupling constant from above implying that par-
ticle was in thermal equilibrium in early Universe. For typical weak inter-
action coupling G
F
mass is constrained to be more than 2 GeV/c
2
and
such particles is referred to as Weakly Interacting Massive Particles, WIMPs.
Bound by Tremaine and Gunn (1979) and its improvement by many authors
(Hogan and Dalcanton, 2000; de Vega and Sanchez, 2010) are based on as-
157
G. Semidegenerate self-gravitating system of fermions as a model for dark
matter halos and universality laws
sumption of fermionic nature of dark matter and nondegeneracy of galactic
haloes of dwarf galaxies, that leads to lower limit on mass 0.41 keV/c
2
.
The problem of dark matter distribution in galactic halos has traditionally
been treated in the realm of newtonian physics in view of the low velocities of
the stars in the galaxies, like the simulations from Navarro, Frenk and White
(1997). In the meantime, phenomenological proles of dark matter have been
advanced by Einasto (1965); Burkert (1995), and universal properties of the
dark matter distribution have been inferred from dwarf galaxies and proba-
bly globular clusters all the way to very massive galaxies (Gentile et al., 2009;
Donato et al., 2009; Walker et al., 2010; Boyarsky et al., 2009). However, a
problem arises: while simulations like those from NFW point to a cusped
halo, observations from various types of galaxies seem to show cored halos
(Salucci et al., 2011). This discrepancy between theory and observations is not
yet fully understood, but could show a problem with the simulations done
so far.
In a completely unrelated eld (as of yet), the physics of Active Galactic
Nuclei (AGN) and quasars has been recognized for more than 50 years as
dominated by relativistic gravitational effects of a black hole. The formation
of these black holes is not yet fully understood, although different black holes
mass estimates for AGNs and quasars show masses up to 10
10
M
2m
3
c
3
_
1 + /2mc
2
(1 + /mc
2
)
_
/mc
2
d(/mc
2
).
(G.2.6)
The particle energy is a constant of motion, so
( + mc
2
)e
/2
= const, (G.2.7)
while thermodynamical equilibrium (Tolman condition and Klein integral
159
G. Semidegenerate self-gravitating system of fermions as a model for dark
matter halos and universality laws
(Rueda et al., 2011)) implies
( + mc
2
)e
/2
= (
R
+ mc
2
)e
R
/2
(G.2.8)
Te
/2
= T
R
e
R
/2
, (G.2.9)
where the quantities with subscript R refer to the boundary of the congu-
ration. For the cutoff energy we have
(
c
+ mc
2
)e
/2
= mc
2
e
R
/2
, (G.2.10)
since
c
(R) = 0.
Introducing the function W =
c
/kT and the temperature parameter at the
boundary
R
= kT
R
/mc
2
, and using eqs. (G.2.9) and (G.2.10) we can nd that
mc
2
kT
=
1
R
W
R
. (G.2.11)
Note that the condition 0
R
W < 1 has to be fullled. Using eq. (G.2.9)
to substitute the temperature in eq. (G.2.8) we get the relation between the
metric function and W:
e
= e
R
(1
R
W)
2
(G.2.12)
so now the spacetime metric is completely determined:
e
= e
R
[1
R
W]
2
, e
=
_
1
2GM
rc
2
_
1
(G.2.13)
with
R
+
R
= 0.
Differentiating eq. (G.2.12) and using the conservation of the energy mo-
mentum tensor
dP
dr
=
1
2
(P + c
2
)
d
dr
(G.2.14)
gives
dP
dr
=
R
(P + c
2
)
1
R
W
dW
dr
(G.2.15)
and we can write eq. (G.2.2) as
dW
dr
=
G
c
2
_
1
R
W
R
_
Mc
2
+ 4Pr
3
r(rc
2
2GM)
(G.2.16)
In order to numerically integrate the nal set of equations (G.2.3) and
(G.2.16) with initial conditions W(0) = W
0
and M(0) = 0, it is useful to
160
G.2. Model
transform all of our physical variables into dimensionless ones:
=
c
2
G
2
(G.2.17)
P =
c
4
G
2
P (G.2.18)
M =
c
2
G
M (G.2.19)
r = r, (G.2.20)
where
=
h
mc
_
m
p
m
_
_
8
3
g
_
1/2
(G.2.21)
has dimension of length and m
p
= ( hc/G)
1/2
is the Planck mass.
It is instructive to write down characteristic length , that is inversely pro-
portional to square of the mass of the particle, in conventional units
= 0.870m
2
pc, (G.2.22)
where m is measured in keV/c
2
, and unit of mass is
c
2
G
= 1.820 10
13
m
2
M
, (G.2.23)
where M
= 1.989 10
33
g is mass of the Sun.
We then obtain the dimensionless equations
dW
d r
=
_
1
R
W
R
_
M(r) + 4
P r
3
r( r 2
M(r))
dM(r)
dr
= 4 r
2
, (G.2.24)
where
= 4
2
_
R
1
R
W
_
3/2
_
W
0
_
1 +
R
x/2
1
R
W
_
1/2
_
1 +
R
x
1
R
W
_
2
1 e
xW
e
x
+ 1
x
1/2
dx
P =
8
2
3
_
R
1
R
W
_
5/2
_
W
0
_
1 +
R
x/2
1
R
W
_
3/2
1 e
wW
e
x
+ 1
x
3/2
dx, (G.2.25)
where = /kT is the degeneracy parameter and we introduced the variable
x = /kT. We have for this variable
mc
2
=
R
x
1
R
W
. (G.2.26)
161
G. Semidegenerate self-gravitating system of fermions as a model for dark
matter halos and universality laws
The relation between the degeneracy parameter and W is
W =
R
, (G.2.27)
so that W(R) = 0, where
R
is the value of the degeneracy parameter at the
boundary. We can relate the parameters in the boundary with those in the
center
R
=
0
W
0
R
=
0
1 +
0
W
0
(G.2.28)
so that
R
0
for
0
<< 1. Besides that we have
1
R
W
r
=
1
0
(WW
0
)
0
. (G.2.29)
Now the system can be completely solved (numerically) by solving the
equations eq. (G.2.24) together with
=
0
+WW
0
(G.2.30)
and using eq. (G.2.25) with three independent parameters: W
0
,
0
and
0
.
The only remaining free parameter is the mass of the particle, which occurs
only in the denition of and the characteristic length .
G.2.1. Properties of the equilibrium congurations
We have solved numerically the system of integral-differential equations
given by (G.2.16), the two equations corresponding to and and (G.2.25),
with a set of initial conditions M
0
, W
0
,
0
and
0
. Galactic dark matter halos
have asymptotic rotation velocities of the order of ten to thousands km/s, i.e.,
they are not relativistic. As that velocities are of the same order as thermal
velocities of fermionic particles forming the halo, this means that
R
1
and consequently
0
1. For semidegenerate congurations
0
20, and in
this case we have three regions of halo (g. G.1): a degenerate core of almost
constant density, an inner halo also with almost constant density and a tail
where density scales as r
2
until the cutoff.
On the velocity curve, we can see 4 characteristic regions (g. G.2):
Part I: The core with constant density, where v r;
Part II: The rst part of the inner halo, where the mass of the core pre-
vails over the mass of the halo and v r
1/2
;
Part III: Second part of the inner halo, where now the mass of the halo
prevails and again v r;
162
G.3. Comparison with other DM proles
Figure G.1.: Density prole of the model for
0
= 10
8
,
0
= 32 and W
0
= 92.
Figure G.2.: Velocity curve for the same parameter values as before
Part IV: The outer halo, where the velocity tends to a constant value v
0
after some oscillations of diminishing magnitude.
G.3. Comparison with other DM proles
To compare results obtained with known Dark Matter properties we need
to nd out the correspondence between ts of circular velocity, much like it
was suggested in Boyarsky et al. (2009). There is some controversy in cur-
rent literature about the undisturbed prole of dark matter in Galaxies and
163
G. Semidegenerate self-gravitating system of fermions as a model for dark
matter halos and universality laws
clusters. Cold dark matter simulations suggest the so-called Navarro-Frenk-
White prole (Navarro et al., 1997)
=
NFW
r/r
NFW
(1 + r/r
NFW
)
2
(G.3.1)
and Einasto prole (Navarro et al., 2010) introduced by J. Einasto for model-
ing of matter distribution in Milky Way (Einasto, 1965)
=
E
exp
_
[(r/r
E
)
1]
_
(G.3.2)
while phenomenological pseudoisothermal sphere
=
iso
1 + (r/r
iso
)
2
(G.3.3)
and Burkert prole (Burkert, 1995)
=
B
(1 + r/r
B
)(1 + (r/r
B
)
2
)
. (G.3.4)
are commonly used for tting. Comparing proles of circular velocity for all
these proles with the one of the semidegenerate solution, we came to con-
clusion that NFW and Burkert proles, having wrong asymptotics as r ,
better reproduce the characteristic bump in the circular velocity near the
edge of inner halo (g. G.3). The best reproduction is obtained for Burkert
prole. As most of papers dealing with rotational curve tting nd out that
Burkert or other cored proles are the best ts for dark matter distribution,
and that characteristic scale r
B
of the tted prole is comparable to the full
length of tting range (see, e.g., Gentile et al. (2004)), that means that semide-
generate fermion halo can provide the same quality of ts for that galaxies.
As we move outside from the border of inner/outer halo, the ts by pseu-
doisothermal sphere became better than that of other proles (g. G.4). The
ts by other proles suffer due to their different outer slope, so constant circu-
lar velocity can be only approximated in a nite range of radii by a decreasing
function. As a result we have systematic deviations from the real at curve
in the beginning and the end of tting range.
The result obtained means that all ts of rotational curves by Burkert, NFW
and pseudoisothermal sphere proles could be tted as well by semidegen-
erate fermion halo.
164
G.3. Comparison with other DM proles
10000 20000 30000 40000
r, pc
1.0
1.2
1.4
1.6
10
5
v
c
c
Figure G.3.: Dependence of v
c
/c on radius r (black thick line) for =
10
10
,
0
= 20, m = 9.3 keV/c
2
near the edge of the inner halo and its ts by
NFW (red dashed line), Burkert (blue thin line) and pseudoisothermal sphere
(green dot-dashed line) proles. The radius range used for tting is shown
by grey shading (r = 10
3
to r = 10
4
pc).
100000 200000 300000 400000
r, pc
0.8
1.0
1.2
1.4
1.6
10
5
v
c
c
Figure G.4.: Dependence of v
c
/c on radius r (black thick line) for =
10
10
,
0
= 20, m = 9.3 keV/c
2
in the outer halo and its ts by NFW (red
dashed line), Burkert (blue thin line) and pseudoisothermal sphere (green
dot-dashed line) proles. The radius range used for tting is shown by grey
shading (r = 10
4
to r = 10
5
pc).
165
G. Semidegenerate self-gravitating system of fermions as a model for dark
matter halos and universality laws
G.4. Scaling Laws
The solutions obtained show remarkable self-similarity properties. The char-
acteristics of solutions obey ve scaling laws against the free parameters of
the model
0
,
0
, W
0
and m
f
. These are laws for the asymptotic velocity of
the rotation curves, for the core mass, for the core radius, for the halo mass
and for the halo radius.
Asymptotic rotation velocity scaling law
v
0
= 4.47 10
5
_
0
km/s (G.4.1)
show dependence on the temperature of the conguration only. This is es-
sentially the same scaling law that appears in the case of isothermal sphere
(Binney and Tremaine, 1987).
Core is dened as region from the center of the system till the rst maxi-
mum of the rotation velocity curve (region I in g. G.2). Near that point the
density of fermions decreases fast. Scalings of mass and radius of the core are
M
c
= 1.96 10
12
(
0
0
)
0.75
m
2
f
M
, (G.4.2)
r
c
= 0.180(
0
0
)
0.25
m
2
f
pc, (G.4.3)
where m
f
is in units of keV/c
2
. Characteristics of core is dependent not on
0
and
0
separately, but only on their product
0
0
, i.e. on the chemical
potential at the semidegenerate center of conguration, in accordance with
results of Narain et al. (2006). From the laws an important relation could be
obtained, involving only M
c
, r
c
and m
M
c
r
3
c
= 1.15 10
10
m
4
f
M
pc
3
. (G.4.4)
Halo is dened as region up r
0
, where density drops to 1/4 of that in min-
imum rotational velocity point = (r
h
) =
0
4
(regions II and III in g. G.2).
Scaling laws of halo properties are
M
h
= 1.49 10
13
0
0.77
10
0.16
0
m
2
f
M
, (G.4.5)
r
h
= 0.35
0
0.25
10
0.16
0
m
2
f
pc. (G.4.6)
Notice that instead of power-law dependance on
0
in core properties, halo
properties depend on central degeneracy of conguration exponentially.
The equations above are exact in m
f
and hold in the following physical
range of the other parameters: log
10
0
[10, 5],
0
[20, 200], W
0
166
G.4. Scaling Laws
[110, 200]. Note that W
0
does not appear in the scaling laws, so that its value
does not change inner structure of conguration.
G.4.1. Solving the scaling laws system of equations
In order to nd out the order of semidegenerate halo parameters correspond-
ing to observed ones, we adopted the four most reliable observed character-
istics of the Milky Way, i.e. its asymptotic rotational velocity v
0
220 km/s,
mass of the central object in the Galactic center M
c
4 10
6
M
, radius of
Galactic halo r
h
r
0
14 10
3
pc, (r
0
: one dark halo scale length) and its
mass M
h
2 10
11
M
.
We take three equations corresponding to the number of three unknown
parameters, namely (G.4.1), (G.4.5), and (G.4.2). Then solving the system
using the data presented above we arrive to
0
= 10
6.6
, (G.4.7)
0
= 32.4 (W
0
= 130), (G.4.8)
m
f
= 12.6 KeV. (G.4.9)
Finally we can take the law (G.4.3) to obtain r
c
= 2.2 10
2
pc. Although
this value is quite far from the size of very compact region known as SgrA*,
it is still in the very inner region of the bulge, in accordance with the ob-
servations of Ghez et al. (2008) and Gillessen et al. (2009) made for orbits of
S2(blue) stars, where can be seen that at a radius of around 1 10
2
pc the
enclosed mass for the orbit must be around 4 10
6
M
.
G.4.2. Application to spiral, elliptical and group of galaxies
The application of the model to galaxies has been inspired by the recent ob-
servations showing that the mean dark matter surface density within one
dark halo scale/length r
0
(within this radius the volume density prole of
dark matter remains approximately at) is constant across a wide range of
galaxies (Gentile et al., 2009; Donato et al., 2009), or equivalently, an almost
constant DM acceleration at r
0
: a
DM
1
km
2
s
2
pc
is observed (Walker et al., 2010).
This Universality laws have also been extended by Boyarsky et al. (2009) for
matter distributions at all observed scales.
Spiral galaxies
We based our analysis within the phenomenological work made by Walker
et al. (2010) from the MO7 data, consisting in HI rotation curves, and then
isolating the DM contribution to an observed rotation curve substracting
167
G. Semidegenerate self-gravitating system of fermions as a model for dark
matter halos and universality laws
contributions from stellar and HI regions. This data is based on 686 inde-
pendent resolved rotation velocities measurements for 60 galaxies, spanning
radii between 1 kpc r
h
75 kpc. The important feature is that this sam-
ple covers virtually the entire range of spiral properties; ie. circular velocities
50 km/s v
0
300 km/s within others. With this data they could nd
a constant acceleration due to DM a
DM
= 0.9
+0.9
0.5
km
2
s
2
pc
= 3
+3
2
10
9 cm
s
2
, in
complete accordance with the Universality law found by Gentile et al. (2009).
In the context of our model, and in order to reproduce the a
DM
with the
corresponding scatter in a plot a
DM
vs r
h
, we proceeded as follows:
i) From (G.4.1) we obtained the corresponding b
0
for the lowest observed
value of the circular velocity v
0
= 50
km
s
; log
10
0
= 7.9;
ii) From (G.4.6) and using the lowest limit for
0
, log
10
0
= 7.9, m
f
=
13.5 KeV and the two observed border values for r
h
(1 kpc till 75 kpc) we
obtained the entire range of validity for
0
, namely 23.6
0
35.3;
iii) The same as in i), but now for the highest value of the circular velocity
v
0
= 300 km/s, and then nding the corresponding log
10
0
= 6.34;
iv) The same as in ii), but now with the highest limit of
0
, obtaining now
a new range of validity for
0
, 26
0
38;
v) Considering that a
DM
=
GM
h
r
h
2
and making use of the equations (G.4.5)
and (G.4.6), we obtained the corresponding scaling law for the acceleration
due to DM as
log
10
a
DM
km
2
/(s
2
pc)
= 11.786 + 1.27 log
10
0
0.16
0
+ 2 log
10
m
keV/c
2
(G.4.10)
vi) Finally, with (G.4.6) and (G.4.10) and the physical ranges for
0
and
0
obtained above, we made a plot a
DM
vs r
h
, that shows that all the data points
of the Universality law for a
DM
made by Walker et al. (2010) are included by
the physical range of parameters of our model. The two sloped lines indicates
the
0
range (ie. the velocity range), and the length of these lines indicates
the
0
range; while the two horizontal lines are the limits related to the scatter
from the observations.
The comparison between the two graphics is shown in Fig. G.5.
Spiral, elliptical and group of galaxies
The application of the model in this section was inspired in the phenomeno-
logical work made by Boyarsky et al. (2009) based on analogous observations
as in the former case mentioned in the work. The objects covered in this case
are: 124 spiral galaxies, 10 elliptical ones and 25 groups of galaxies (within
specic selection criteria explained in that paper).
168
G.4. Scaling Laws
0.01
0.1
1
10
10 100 1000 10000 100000
a
D
M
(
k
m
2
/(
s
2
p
c
1
)
)
r (pc)
model bounds
lower observational limit
upper observational limit
Figure G.5.: The acceleration due to dark matter vs the halo scale-length r
h
.
The left bound corresponds to v
0
= 50 km/s (log
0
= 7.9), while the right
corresponds to v
0
= 300 km/s (log
0
= 6.34).
Authors present the Universality law in terms of the DM column density
S =
2
r
2
0
_
r
0
0
rdr
_
dz
DM
_
r
2
+ z
2
M
cyl
r
2
0
, (G.4.11)
so that the column density can be approximated as the mass within a cylin-
der of radius r
0
(meaning this as always for our purpose, the one halo scale-
length). This quantity is also proportional to
0
r
0
, so it should be constant for
a wide span of magnitudes.
Interpreting M
cyl
as the mass of the halo, and using the formula mentioned
in (Boyarsky et al., 2009), S
M
cyl
.r
2
0
and noting that S G = a
DM
we nd from
(G.4.5) and (G.4.6) the corresponding scaling law in this case as
log
10
S
M
/pc
2
= 16.115 + 1.27 log
10
0
0.16
0
+ 2 log
10
m
f
keV/c
2
. (G.4.12)
Then, in order to reproduce the log
10
S with the corresponding scatter in
a plot log
10
S vs M
h
, we proceeded in a complete analogous way as in the
points i)-iv) for the former case.
The comparison between the two graphics is shown in Fig. G.6.
Two important remarks concerning the richness of the model are: i) we
dont take into account the data of Dwarf Galaxies due to the very high error
bars in the observations, and ii) as it can be seen from making a direct com-
parison between the two graphics for the second case, although the groups of
galaxies are covered by the range of parameters mentioned above, the model
exclude the clusters of galaxies as a part of a Universality law.
So we have showed within the model that always exists a physical range
of parameters
0
,
0
and W
0
that reproduces the graphics showing the Uni-
169
G. Semidegenerate self-gravitating system of fermions as a model for dark
matter halos and universality laws
0
1
2
3
4
5
6
10
7
10
8
10
9
10
10
10
11
10
12
10
13
10
14
10
15
10
16
D
M
c
o
l
u
m
n
d
e
n
s
i
t
y
,
l
g
(
S
/
M
s
u
n
p
c
-
2
)
DM halo mass [M
sun
]
Clusters of galaxies
Groups of galaxies
Spiral galaxies
Elliptical galaxies
dSphs
Isolated halos from CDM N-body simulations
Subhalos from Aquarius simulation
0
1
2
3
4
5
6
1e+07 1e+08 1e+09 1e+10 1e+11 1e+12 1e+13 1e+14 1e+15 1e+16
lo
g
S
(
M
s
u
n
/
p
c
2
)
M
h
(Msun)
model bounds
upper observational limit
lower observational limit
Figure G.6.: The dark matter column density S vs the halo mass M
h
. The left
bound from the model corresponds to v
0
= 50 km/s (log
0
= 7.9) and the
right one to v
0
= 1400 km/s (log
0
= 5). The dot is the milky way.
versality laws in the different scopes.
170
H. Features in the primordial
spectrum: new constraints from
WMAP7+ACT data and
prospects for Planck
H.1. Introduction
Current cosmological observations can be explained in terms of the so-called
concordance CDM model in which the primordial uctuations are created
during an early period of inationary expansion of the Universe. In par-
ticular, the spectrum of anisotropies of the cosmic microwave background
(CMB) is in excellent agreement with the inationary prediction of adiabatic
primordial perturbations with a nearly scale-invariant power spectrum Ko-
matsu et al. (2011); Larson et al. (2011); Das et al. (2011); Dunkley et al. (2011);
Hlozek et al. (2011). In its simplest implementation, ination is driven by
the potential energy of a single scalar eld, the inaton, slowly rolling down
towards a minimum of its potential; curvature perturbations, that constitute
the primordial seeds for structure formation, are originated during the slow
roll from quantum uctuations in the inaton itself. The scale invariance of
the spectrum is directly related to the atness and smoothness of the ina-
ton potential, that are necessary to ensure that the slow-roll phase lasts long
enough to solve the paradoxes of the Big Bang model.
However, in more general inationary models, there is the possibility that
slow roll is briey violated. This naturally happens in theories with many
interacting scalar elds, as it is the case, for example, in a class of multield,
supergravity-inspired models Adams et al. (1997b,a), where supersymmetry-
breaking phase transitions occur during ination. These phase transitions
correspond to sudden changes in the inaton effective mass and can be mod-
eled as steps in the inationary potential. If the transition is very strong, it
can stop the inationary phase as it happens in the usual hybrid ination
scenario; on the contrary, ination can continue but the inationary pertur-
bations and thus the shape of the primordial power spectrum are affected.
Departures from the standard power-law behaviour can also be caused by
changes in the initial conditions due to trans-planckian physics Branden-
berger and Martin (2001); Easther et al. (2002); Martin and Brandenberger
171
H. Features in the primordial spectrum: new constraints from
WMAP7+ACT data and prospects for Planck
(2003) or to unusual initial eld dynamics Burgess et al. (2003); Contaldi et al.
(2003)
A violation of slow-roll will possibly lead to detectable effects on the cos-
mological observables, or at least to the opportunity to constraint these mod-
els by the absence of such effects. In particular, step-like features in the pri-
mordial power spectrum have been shown Adams et al. (2001); Hunt and
Sarkar (2004) to lead to characteristic localized oscillations in the power spec-
trum of the primordial curvature perturbation. Such oscillations have been
considered as a possible explanation to the glitches observed by the Wilkin-
son Microwave Anisotropy Probe (WMAP) in the temperature anisotropy
spectrum of the CMB, although the WMAP team notes that these could be
caused simply by having neglected beam asymmetry, the gravitational lens-
ing of the CMB, non-gaussianity in the CMB maps and other small ( 1%)
contributions to the covariance matrix. In the following we will assume that
these features have indeed a cosmological origin as in the class of extended
models described above and we will use CMB data to constrain the phe-
nomenological parameters describing the step in the inaton potential.
Constraints on oscillation in the primordial perturbation spectrum, as well
as best-t values for the step parameters, have been previously derived in
Refs. Peiris et al. (2003); Covi et al. (2006); Hamann et al. (2007); Mortonson
et al. (2009); Hazra et al. (2010). Here we improve on the previous analyses in
several aspects. First, we use more recent CMB data, in particular the WMAP
7-year and the Atacama Cosmology Telescope (ACT) data. This allows us
to derive tighter constraints on the parameters; in particular we get an up-
per limit on the step height (related to the amplitude of oscillations) that is
independent on the position of the step itself in the prior range considered.
We also nd a clear correlation between the position and the height of the
step. Secondly, we generate mock data corresponding to the model provid-
ing the best-t to the WMAP data, and use these data to assess the ability
of the Planck satellite to detect the presence of oscillations in the primordial
spectrum.
The paper is organized as follows: in Section II we describe the evolution of
perturbations in interrupted slow roll and the phenomenological model used
to describe a step in the inationary potential. In Section III we discuss the
analysis method adopted. In Section IV we present the results and in Section
V we derive our conclusions.
H.2. Inationary perturbations in models with
interrupted slow roll
Steps in the potential can naturally appear in multiple ination models,
where the inaton eld is gravitationally coupled to a at direction eld
172
H.2. Inationary perturbations in models with interrupted slow roll
(belonging to the visible sector of the theory), i.e. a direction in eld space
along which the potential vanishes. The -eld can undergo a symmetry-
breaking phase transition and acquire a vacuum expectation value . The
gravitational coupling between the and the inaton eld will cause the ef-
fective mass-squared of the latter to change; for example, in the case in which
the coupling between the two elds is described by a term
2
2
/2 in the
Lagrangian, the inaton mass-squared after the phase transition will become
m
2
eff
() = m
2
0
+
2
. It is worth noticing that the presence of at eld direc-
tions also opens the possibility to have ination with a curved trajectory in
eld space; however, in the following, we will disregard this scenario.
The exact behaviour of the inaton mass will depend by the dynamics of
the phase transition; however, this is so fast that the -eld reaches the min-
imum of its potential very rapidly. It is then very reasonable to model the
inaton mass in a phenomenological way as
m
2
eff
() = m
2
_
1 + c tanh
_
b
d
__
. (H.2.1)
Here, the parameter b is of the order of the critical value of the inaton eld
for which the phase transition occurs, c is the height of the step (related to
the change in the inaton mass) and d is its width (related to the duration of
the phase transition). In the following we shall work in reduced Planck units
(c = h = 8G = 1), so that all dimensional quantities like m, b and d should
be multiplied by the reduced Planck mass M
p
= 2.435 10
18
GeV in order to
get their values in physical units.
Let us now briey recall how to compute the spectrum of primordial per-
turbations, as discussed in details by Adams et al. (2001). For the moment,
we do not specify the exact form of the inaton potential V(); we will return
on this in the next Section. In the case of scalar perturbations, it is useful to
dene the gauge-invariant quantity Stewart and Lyth (1993) u zR, where
z = a
/H, a is the scale factor, H is the Hubble parameter, R is the curvature
perturbation, and dots denote derivatives with respect to the cosmological
time t. The Fourier components of u evolve according to:
u
//
k
+
_
k
2
z
//
z
_
u
k
= 0 , (H.2.2)
where k is the wavenumber of the mode, and primes denote derivatives with
respect to conformal time . When k
2
z
//
/z, the solution to the above
equation tends to the free-eld solution u
k
= e
ik
/
2k.
In the slow-roll approximation, z
//2
H
2
. However, in the models considered
here this expectation can be grossly violated near the phase transition, and
the time evolution of z has to be derived by solving the equations for the
173
H. Features in the primordial spectrum: new constraints from
WMAP7+ACT data and prospects for Planck
inaton eld and for the Hubble parameter:
+ 3H
+
dV
d
= 0 , (H.2.3)
3H
2
=
2
2
+ V() . (H.2.4)
Once the form of the potential is given, these can be integrated to get H and
, and thus z, as a function of time. At this point, it is possible to integrate Eq.
(H.2.2) to get u
k
() for free-eld initial conditions when k
2
z
//
/z. Finally,
knowing the solution for the mode k, the power spectrum of the curvature
perturbation P
R
can be computed by means of
P
R
=
k
3
2
u
k
z
2
(H.2.5)
evaluated when the mode crosses the horizon. The resulting spectrum for
models with a step in the potential is essentially a power-law with superim-
posed oscillations; thus, asymptotically, the spectrum will recover the famil-
iar k
n
s
1
form typical of slow-roll inationary models.
In practice, however, one has to relate the horizon size at the step with
a physical wavenumber. For a general wavenumber k
= a
end
e
N
, where a
and H
=
k
0
= 0.0025 Mpc
1
. A different choice would correspond to a translation in
the position of the step in and would thus be highly degenerate with b. For
this reason we do not treat N
)
/
(
1
0
1
6
G
e
V
)
4
/M
Pl
Figure H.1.: Inationary potential (H.3.1) for m = 7.5 10
6
. The solid thin
black line corresponds to a smooth (c = 0) chaotic potential m
2
2
/2. The
long-dashed red curve has b = 14.23, c = 0.001 and d = 0.025 and roughly
corresponds to the spectrum giving the best t to the WMAP7 data (see Sec.
H.4 below). The other curves correspond to b = 14.23, c = 0.005, d = 0.025
(blue short-dashed), b = 14.23, c = 0.005, d = 0.05 (green dotted) and b =
14.25, c = 0.005, d = 0.025 (magenta dot-dashed).
m
2
eff
()
2
/2. Using Eq. (H.2.1), this corresponds to a potential
V() =
1
2
m
2
2
_
1 + c tanh
_
b
d
__
. (H.3.1)
In Fig. H.1 we show the shape of this potential for m = 7.5 10
6
and differ-
ent values of the step parameters (close to the best-t values), compared to a
smooth m
2
2
/2 potential (c = 0).
The potential (H.3.1) uniquely denes the spectrum of perturbations P
R
.
The parameters that dene the primordial spectrum and the initial conditions
for the evolution of cosmological perturbations are then the inaton mass m
and the step parameters b, c and d. The inaton mass sets the overall scale for
the potential and consequently for the amplitude of the perturbations; it can
then be traded, in the Monte Carlo analysis, for the more familiar parameter
A
s
, i.e., the amplitude of the primordial spectrum at the pivot wavenumber
k
0
= 0.0025 Mpc
1
. On the other hand, as already noted above, a step in the
175
H. Features in the primordial spectrum: new constraints from
WMAP7+ACT data and prospects for Planck
0
1e-09
2e-09
3e-09
4e-09
5e-09
6e-09
0.0001 0.001 0.01 0.1
2
R
(
k
)
k [Mpc
-1
]
Figure H.2.: Primordial power spectrum for an inationary potential of the
form (H.3.1). The values of the step parameters are the same as in Fig. (H.1),
namely: b = 14.23, c = 0.001, d = 0.025 (red long-dashed), b = 14.23, c =
0.005, d = 0.025 (blue short-dashed), b = 14.23, c = 0.005, d = 0.05 (green
dotted) and b = 14.25, c = 0.005, d = 0.025 (magenta dot-dashed).
potential produces a perturbation spectrum with oscillations superimposed
over a smooth power law. In the case of the potential (H.3.1), the underlying
power-law has a xed spectral index n
s
= 0.96. In Fig. H.2 we show the
primordal spectrum for different values of the step parameters.
The results obtained in the case of a specic potential will be, by denition,
model-dependent. However, as argued in Ref. Hamann et al. (2007), the issue
of model dependence can be alleviated in a phenomenological way by restor-
ing the spectral index as a free parameter, i.e., by dening the generalized
spectrumP
gen
R
as
P
gen
R
(k) = P
ch
R
(k)
_
k
k
0
_
n
s
0.96
, (H.3.2)
where P
ch
R
(k) is the spectrum induced by the chaotic potential (H.3.1). Since
the latter has a overall tilt of 0.96, n
s
will describe the overall tilt of the gener-
alized spectrum.
Summarising, we consider two classes of models. Models belonging to
the rst class (referred to as class A) corresponding to the potential (H.3.1),
176
H.3. Analysis Method
are described by eight parameters: the physical baryon and cold dark matter
densities
b
=
b
h
2
and
c
=
c
h
2
, the ratio between the sound hori-
zon and the angular diameter distance at decoupling, the optical depth to
reionization , the parameters b, c and d of the step-ination model, and the
overall normalization of the primordial power spectrum A
s
(equivalent to
specifying m
2
as discussed above). Models in the second class, referred to as
class B, correspond to the generalized step model (H.3.2) and are described
by the effective tilt n
s
in addition to the eight parameters of the rst class. In
both cases, we consider purely adiabatic initial conditions, impose atness
and neglect neutrino masses. We limit our analysis to scalar perturbations.
Priors. Apart from the hard-coded priors of CosmoMC on H
0
(40 km s
1
Mpc
1
< H
0
<100 km s
1
Mpc
1
) and the age of the Universe (10 Gyr < t
0
<
20 Gyr), we impose at priors on
b
,
c
, , and, when considered, n
s
and
a logarithmic prior on A
s
. As we shall see, for these parameters the width
of the posterior is much smaller than the prior range, so that the latter is
not really relevant. For the step parameters, the situation is complicated by
the fact that the likelihood (and the posterior) does not go to zero in certain
directions of the subspace. This happens in particular for very small values of
c, for which the spectrum becomes indistinguishable from a power law, and
for values of b either too large or too small so that the features in the spectrum
are moved outside the range of observable scales. Then we choose for b a at
prior 13 b 15, that roughly encompasses said range. In the case of c and
d, since we do not have any a priori information on these parameters, not even
on their order of magnitude, we nd convenient to consider a logarithmic
prior on both of them. Hence, we take (in the following, log x denotes the
base 10 logarithm) 6 log c 1 and 2.5 log d 0.5. Additionally,
since the combination c/d
2
is better constrained by the data than d alone, we
also impose a prior 5 log(c/d
2
) 3. Finally, we recall that, since the
posteriors for b, log c and log d do not necessarily vanish at the edge of the
prior range, all integrals of the probability density function depend on the
extremes of integration and are thus somewhat ill-dened. Care should then
be taken when quoting condence limits in the b, log c and log d subspaces.
Datasets. We performthe statistical analysis for each of the models by com-
paring the theoretical predictions to two different datasets. The rst includes
the WMAP 7-year temperature and polarization anisotropy data (WMAP7).
The likelihood is computed using the the WMAP likelihood code publicly
available at the LAMBDAwebsite
1
. We marginalize over the amplitude of the
Sunyaev-Zeldovich signal. The second dataset includes the WMAP7 data
with the addition of the small-scale CMB temperature anisotropy data from
the ACT experiment. For the ACT dataset we also consider two extra param-
eters accounting for the Poisson and clustering point sources foregrounds
components. The ACT dataset is considered up to
max
= 2500.
1
http://lambda.gsfc.nasa.gov/
177
H. Features in the primordial spectrum: new constraints from
WMAP7+ACT data and prospects for Planck
Other than deriving the limits on the models from existing data, we also
assess the ability of future experiments, in particular of the Planck satellite,
to improve these constraints. In order to do this, we simulate mock data
corresponding to the step model that yields the best-t to the WMAP 7 and
then perform a statistical analysis on these data as if they were real. The
forecast method we use is identical to the one presented in Galli et al. (2010)
and we refer to this paper for further details and references. The synthetic
dataset is generated by considering for each C
= w
1
exp(( + 1)8 ln2/
2
b
), (H.3.3)
where
b
is the full width at half maximum (FWHM) of the beam assuming
a Gaussian prole and where w
1
is the experimental power noise related to
the detectors sensitivity by w
1
= (
b
)
2
. The experimental parameters are
reported in Table H.1.
Experiment Channel FWHM T/T P/T
Planck 70 14 4.7 6.7
f
sky
= 0.85 100 10 2.5 4.0
143 7.1 2.2 4.2
Table H.1.: Planck Collaboration (2006) experimental specications of Planck
satellite. Channel frequency is given in GHz, FWHM in arcminutes and noise
per pixel for the Stokes I (T/T), Q and U parameters (P/T) is in [K/K],
where T = T
CMB
= 2.725 K. In the analysis, we assume that beam uncertain-
ties and foreground uncertainties are smaller than the statistical errors.
Together with the primary anisotropy signal we also take into account in-
formation from CMB weak lensing, considering the power spectrum of the
deection eld C
dd
.
Analysis. We derive our constraints from parallel chains generated using
the Metropolis-Hastings algorithm. We use the Gelman and Rubin R param-
eter to evaluate the convergence of the chains, demanding that R 1 < 0.03.
The one- and two-dimensional posteriors are derived by marginalizing over
the other parameters.
H.4. Results and Discussion
We rst consider the WMAP7 and WMAP7+ACT datasets. We nd that the
CDM t to both datasets can be improved by the inclusion of a step in the
inationary potential, in both cases when the scalar spectral index is being
xed to n
s
= 0.96 (Model A), and when it is being treated as a free parameter
178
H.4. Results and Discussion
0
0.2
0.4
0.6
0.8
1
13.6 13.8 14 14.2 14.4 14.6 14.8 15
L
/
L
m
a
x
b
Figure H.3.: Model likelihood as a function of b for model A (thin curves) and
B (thick curves) using WMAP7 data (dashed curves) and the WMAP7+ACT
dataset (solid curves)
Table H.2.: Best-t values for the parameters of the primordial spectrum.
Model A Model A Model B Model B
Parameter WMAP7 WMAP7+ACT WMAP7 W7+ACT
b 14.23 14.25 14.24 14.25
log c -3.11 -2.71 -2.97 -2.67
log d -1.58 -1.60 -1.65 -1.45
n
s
0.953 0.959
ln[10
10
A
s
] 3.08 3.06 3.07 3.08
2
7469.4 7489.6 7467.9 7491.4
179
H. Features in the primordial spectrum: new constraints from
WMAP7+ACT data and prospects for Planck
13 13.5 14 14.5 15
P
o
s
t
e
r
i
o
r
P
r
o
b
a
b
i
l
i
t
y
b
Model A, WMAP7
Model B, WMAP7
Model A, WMAP7+ACT
Model B, WMAP7+ACT
-4 -3.5 -3 -2.5 -2 -1.5 -1
P
o
s
t
e
r
i
o
r
P
r
o
b
a
b
i
l
i
t
y
Log[c]
Model A, WMAP7
Model B, WMAP7
Model A, WMAP7+ACT
Model B, WMAP7+ACT
-2.5 -2 -1.5 -1 -0.5
P
o
s
t
e
r
i
o
r
P
r
o
b
a
b
i
l
i
t
y
Log[d]
Model A, WMAP7
Model B, WMAP7
Model A, WMAP7+ACT
Model B, WMAP7+ACT
Figure H.4.: One-dimensional posterior probability density for b (left panel),
log c (middle panel) and log d (right panel) for model A (thin curves) and B
(thick curves) using WMAP7 data (dashed curves) and the WMAP7+ACT
dataset (solid curves)
(Model B). The best-t values for the step parameters are reported in Table
H.2. We also show the full likelihood for b in Fig. H.3. It can be seen that in all
cases the maximum in the likelihood occurs for b 14.2; as we show below,
this is due to oscillations placed in correspondence to the WMAP glitches at
20 and 40 and thus able to improve, for suitable values of the other
parameters, the goodness-of-t with respect to the vanilla CDM model. We
found that in case of the WMAP7 analysis the best-t vanilla -CDM model
is at about
2
eff
6 from the global best t with features.
As long as bayesian statistics is concerned, the actual probability density
distribution for a parameter is not given by the likelihood (the probability of
the data given the parameters) but instead by the posterior (the probability
of the parameters given the data). In Fig. H.4 we show the one-dimensional
posterior distributions for the step parameters b, log c and log d. It can be
noted that the posterior for b has a peculiar shape, presenting a peak for
b 14.2 and a fairly wide dip for b 14. The peak traces the peak in the
likelihood discussed above. The decrease for b < 14 is instead due to the fact
that, lowering b, the oscillations are moved to larger multipoles where they
tend to spoil the CDM t unless c is set to a very small value.
This is clearly illustrated in Fig. H.5, where we compare the WMAP7
data with three realizations of the CMB spectrum: the CDM best t to
the WMAP data, the generalized step model best t to the same data (cor-
responding to the third column of Table H.2), and a generalized step model
with the same parameters as the best t, with the exception of b that is set to
b = 13.9. It is clear, especially from the second panel, that for b = 14.2 the os-
cillations improve the t in the region 20 50. On the other hand, when
b = 13.9 the height of the rst peak is diminished so that the predicted spec-
trum is completely at variance with the data. The posterior does not drop to
zero because it still exist a fair amount of parameter space, i.e., models with
low c, than can t the data even with the oscillations placed in the wrong
place. The posterior going to a constant value at the edges of the prior range
is instead related to the oscillations being moved out of the observable scales.
The inclusion of the ACT data in addition to WMAP7 helps in constraining
180
H.5. Concluding Remarks
small values of b, i.e., oscillations at small scales (large s).
The shape of the log c posterior is typical of a quantity parametrizing the
amplitude of a non-standard effect: it is constant for small values of the
parameter (when the step model becomes indistinguishable from standard
CDM), and then rapidly vanishes above a critical value. It can be seen
that the probability density becomes half of its asymptotic value at c = 0
for c 10
2
. Finally, the posterior for log c clearly shows that this parameter
is largely unconstrained by data.
We do not quote one-dimensional condence limits on the parameters be-
cause, as noted in Sec. H.3, the posteriors do not vanish at the edge of the
prior range and in this case the condence limits depend on the integration
range chosen. However, for illustrative purposes, in Fig. H.6 we show the 2-
dimensional 95% condence regions, computed assuming that the posterior
vanishes outside the prior range, in the (b log c) plane. It is clear from the
plots that there is a region below b = 14 where the data are more sensitive to
the value of c; this is related as noted above to the oscillations being placed in
the region where the data are more accurate and favour a smooth spectrum
over one with oscillations.
The results presented here are fully compatible with the analysis made
by Mortonson et al. (2009) where the WMAP5 dataset was considered. The
apparently different value for the best-t b parameter found in that paper
is due to the different choice of the pivot scale (k
0
= 0.05 Mpc
1
instead
of k
0
= 0.0025 Mpc
1
as assumed in our analysis). We have checked that
performing the analysis on the WMAP7 dataset with the assumption of
k
0
= 0.05 Mpc
1
results in a best-t value of b 14.7 in agreement with
the results of Mortonson et al. (2009).
Finally, we show our results on the sensitivity of Planck to the step param-
eters. We have assumed as a ducial model a generalized step model with
b = 14.2, log c = 2.97, log d = 1.65, n
s
= 0.953, A
s
= 2.16 10
9
(basi-
cally corresponding to the Model B best-t to the WMAP7 data, i.e, the third
column of Tab. H.2). The one-dimensional posteriors for b, log c and log d
are shown in Fig. H.7, while in Tab. H.3 we report the mean values for the
primordial spectrum parameters together with their 2 error. As we can see
the prior range dependence goes away with Planck data and we can quote
marginalised credible intervals. We also show the two-dimensional posteri-
ors for the step parameters in Fig. H.8. It is evident that the Planck data will
greatly increase the precision to which the step parameters can be measured;
in particular, a detection of oscillations will be possible.
H.5. Concluding Remarks
We have considered ination models with a small-amplitude step-like fea-
ture in the inaton potential. Features of these kind can be due for example
181
H. Features in the primordial spectrum: new constraints from
WMAP7+ACT data and prospects for Planck
0
1000
2000
3000
4000
5000
6000
1 10 100 1000
l
(
l
+
1
)
C
l
/
2
l
CDM best fit
Step model best fit
b=13.9
WMAP data
600
800
1000
1200
1400
1600
10 20 30 40 50 60
l
(
l
+
1
)
C
l
/
2
l
CDM best fit
Step model best fit
b=13.9
WMAP data
Figure H.5.: Upper panel: CMB anisotropy spectrumfor the CDM(red solid
line) and generalized step model (blue long dashed line) best ts, and for
a step model with b = 13.9 (black short dashed line), compared with the
WMAP7 data. Lower panel: Zoom of the region 60, showing the im-
proved t of the step model.
182
H.5. Concluding Remarks
b
l
o
g
c
12 13 14 15 16
6
5
4
3
2
b
l
o
g
c
12 13 14 15 16
6
5
4
3
2
b
l
o
g
c
12 13 14 15 16
6
5
4
3
2
1
b
l
o
g
c
12 13 14 15 16
6
5
4
3
2
1
Figure H.6.: 95% two-dimensional condence region in the (blog c) plane.
The four panels correspond to, from left to right and from top to bottom: class
A, WMAP7+ACT; class B, WMAP7+ACT; class A, WMAP7; class B, WMAP7.
0
0.2
0.4
0.6
0.8 1 14.18
14.185
14.19
14.195
14.2
14.205
14.21
14.215
14.22
14.225
Posterior probability
b
0
0.2
0.4
0.6
0.8 1 -3.8
-3.6
-3.4
-3.2
-3
-2.8
-2.6
-2.4
Posterior probability
logc
0
0.2
0.4
0.6
0.8 1 -2.1
-2
-1.9
-1.8
-1.7
-1.6
-1.5
-1.4
-1.3
-1.2
Posterior probability
logd
Figure H.7.: One-dimensional posterior probability density for b (left panel),
log c (middle panel) and log d (right panel) derived from the mock Planck
data, for models of class A (dashed curves) and B (solid curves).
Table H.3.: Parameter constraints from Planck.
Parameter Model A Model B
b 14.200 0.010 14.200 0.011
log c 3.00 0.32 3.00 0.34
log d 1.66 0.22 1.64 0.23
n
s
0.96 (xed) 0.957 0.007
ln[10
10
A
s
] 3.073 0.016 3.074 0.016
183
H. Features in the primordial spectrum: new constraints from
WMAP7+ACT data and prospects for Planck
b
l
o
g
d
14.2 14.22
2
1.8
1.6
1.4
log d
l
o
g
c
2 1.5
3.6
3.4
3.2
3
2.8
2.6
b
l
o
g
c
14.2 14.22
3.6
3.4
3.2
3
2.8
2.6
b
l
o
g
c
14.1914.214.21
3.6
3.4
3.2
3
2.8
2.6
log d
l
o
g
c
2 1.5
3.6
3.4
3.2
3
2.8
2.6
b
l
o
g
d
14.1914.214.21
2
1.8
1.6
1.4
Figure H.8.: Two-dimensional posteriors in the (blog c) (left), (log dlog c)
(middle) and (blog d) (right) planes, for models of class A (top) and B (bot-
tom), derived from the mock Planck data. The shaded areas correspond to
the 68% (light blue) and 95% (dark blue) condence regions.
184
H.5. Concluding Remarks
to phase transitions occurring during the slow roll in multi-eld inationary
models. In these models the primordial perturbation spectrum has the form
of a power-law (as in the standard featureless case) with superimposed oscil-
lations, localized in a nite range of scales that basically depends on the posi-
tion of the step in the potential. We have compared the theoretical predictions
of a specic model, i.e., chaotic ination, and of a more general phenomeno-
logical model to the WMAP7 and ACT data, in order to nd constraints on
the parameter describing the model. We have also studied the possibility
of detecting the oscillations with the upcoming Planck data in the case they
really exist.
We have found that models with features can improve the t to the
WMAP7 data when the step in the potential is placed in such a way as to
produce oscillations in the region 20 60, where the WMAP7 data
shows some glitches. We found no further evidence for small scales glitches
from the recent ACT data, this is fully consistent with the recent analysis of
Hlozek et al. (2011). We have also found that models with too high a step are
excluded by the data. Finally, assuming as a ducial model the generalized
step model that provides the best t to the WMAP7 data, we have found that
the Planck data will allow to measure the parameters of the model with re-
markable precision, possibly conrming the presence of glitches in the region
20 60.
185
H. Features in the primordial spectrum: new constraints from
WMAP7+ACT data and prospects for Planck
186
Bibliography
ABRAMOWICZ, M.A., NOVIKOV, I.D. AND PACZYNSKI, B.
The appearance of highly relativistic, spherically symmetric stellar winds.
ApJ , 369, pp. 175178 (1991).
doi:10.1086/169748.
ADAMS, J.A., CRESSWELL, B. AND EASTHER, R.
Inationary perturbations from a potential with a step.
Phys. Rev., D64, p. 123514 (2001).
doi:10.1103/PhysRevD.64.123514.
ADAMS, J.A., ROSS, G.G. AND SARKAR, S.
Multiple ination.
Nucl.Phys., B503, pp. 405425 (1997a).
doi:10.1016/S0550-3213(97)00431-8.
ADAMS, J.A., ROSS, G.G. AND SARKAR, S.
Natural supergravity ination.
Phys.Lett., B391, pp. 271280 (1997b).
doi:10.1016/S0370-2693(96)01484-0.
AKSENOV, A.G., MILGROM, M. AND USOV, V.V.
Structure of Pair Winds from Compact Objects with Application to Emission from
Hot Bare Strange Stars.
ApJ , 609, pp. 363377 (2004).
doi:10.1086/421006.
AKSENOV, A.G., RUFFINI, R. AND VERESHCHAGIN, G.V.
Thermalization of Nonequilibrium Electron-Positron-Photon Plasmas.
Phys. Rev. Lett. , 99(12), pp. 125003+ (2007).
doi:10.1103/PhysRevLett.99.125003.
AKSENOV, A.G., RUFFINI, R. AND VERESHCHAGIN, G.V.
Thermalization of the mildly relativistic plasma.
Phys. Rev. D, 79(4), p. 043008 (2009).
doi:10.1103/PhysRevD.79.043008.
AKSENOV, A.G., RUFFINI, R. AND VERESHCHAGIN, G.V.
Pair plasma relaxation time scales.
Phys. Rev. E, 81, p. 046401 (2010).
187
Bibliography
ALCOCK, C., FARHI, E. AND OLINTO, A.
Strange stars.
ApJ , 310, pp. 261272 (1986).
doi:10.1086/164679.
ARAD, I. AND LYNDEN-BELL, D.
Inconsistency in theories of violent relaxation.
MNRAS , 361, pp. 385395 (2005).
doi:10.1111/j.1365-2966.2005.09133.x.
ARBOLINO, M.V. AND RUFFINI, R.
The ratio between the mass of the halo and visible matter in spiral galaxies and
limits on the neutrino mass.
A&A , 192, pp. 107116 (1988).
ARCHIDIACONO, M., MELCHIORRI, A. AND PANDOLFI, S.
The impact of reionization modelling on CMB neutrino mass bounds.
Nucl.Phys.Proc.Suppl., 217, pp. 6567 (2011).
doi:10.1016/j.nuclphysbps.2011.04.070.
ARCHIDIACONO, M., COORAY, A., MELCHIORRI, A. AND PANDOLFI, S.
CMB Neutrino Mass Bounds and Reionization.
Phys. Rev., D82, p. 087302 (2010).
doi:10.1103/PhysRevD.82.087302.
BAND, D., MATTESON, J., FORD, L., SCHAEFER, B., PALMER, D., TEEGAR-
DEN, B., CLINE, T., BRIGGS, M., PACIESAS, W., PENDLETON, G. ET AL.
BATSE observations of gamma-ray burst spectra. I - Spectral diversity.
ApJ , 413, pp. 281292 (1993).
doi:10.1086/172995.
BELOBORODOV, A.M.
Collisional mechanism for gamma-ray burst emission.
MNRAS , 407, pp. 10331047 (2010).
doi:10.1111/j.1365-2966.2010.16770.x.
BELOBORODOV, A.M.
Radiative Transfer in Ultrarelativistic Outows.
ApJ , 737, p. 68 (2011).
doi:10.1088/0004-637X/737/2/68.
BENEDETTI, A., HAN, W.B., RUFFINI, R. AND VERESHCHAGIN, G.V.
On the frequency of oscillations in the pair plasma generated by a strong electric
eld.
Physics Letters B, 698, pp. 7579 (2011).
doi:10.1016/j.physletb.2011.02.050.
188
Bibliography
BENETTI, M., LATTANZI, M., CALABRESE, E. AND MELCHIORRI, A.
Features in the primordial spectrum: new constraints from WMAP7+ACT data
and prospects for Planck.
Phys.Rev., D84, p. 063509 (2011).
doi:10.1103/PhysRevD.84.063509.
BERESTETSKII, V.B., LIFSHITZ, E.M. AND PITAEVSKII, V.B.
Quantum Electrodynamics (Elsevier, 1982).
BERNARDEAU, F., COLOMBI, S., GAZTA
NAGA, E. AND SCOCCIMARRO, R.
Large-scale structure of the Universe and cosmological perturbation theory.
Phys. Rep. , 367, pp. 1248 (2002).
doi:10.1016/S0370-1573(02)00135-7.
BIANCO, C.L., RUFFINI, R. AND XUE, S.
The elementary spike produced by a pure e
+
e
AGI, L.J.,
EINASTO, M., M
-gamma winds.
MNRAS , 300, pp. 11581180 (1998).
GUILBERT, P.W. AND STEPNEY, S.
Pair production, Comptonization and dynamics in astrophysical plasmas.
195
Bibliography
MNRAS , 212, pp. 523544 (1985).
HALL, G. AND WATT, J.M.
Modern Numerical Methods for Ordinary Differential Equations (New York,
Oxford University Press, 1976).
HAMANN, J., COVI, L., MELCHIORRI, A. AND SLOSAR, A.
New Constraints on Oscillations in the Primordial Spectrum of Inationary Per-
turbations.
Phys.Rev., D76, p. 023503 (2007).
doi:10.1103/PhysRevD.76.023503.
HAZRA, D.K., AICH, M., JAIN, R.K., SRIRAMKUMAR, L. AND SOURADEEP,
T.
Primordial features due to a step in the inaton potential.
JCAP, 1010, p. 008 (2010).
doi:10.1088/1475-7516/2010/10/008.
HLOZEK, R., DUNKLEY, J., ADDISON, G., APPEL, J.W., BOND, J. ET AL.
The Atacama Cosmology Telescope: a measurement of the primordial power spec-
trum (2011).
* Temporary entry *.
HOGAN, C.J. AND DALCANTON, J.J.
New dark matter physics: Clues from halo structure.
Phys. Rev. D, 62(6), 063511 (2000).
doi:10.1103/PhysRevD.62.063511.
HOGG, D.W., EISENSTEIN, D.J., BLANTON, M.R., BAHCALL, N.A.,
BRINKMANN, J., GUNN, J.E. AND SCHNEIDER, D.P.
Cosmic Homogeneity Demonstrated with Luminous Red Galaxies.
ApJ , 624, pp. 5458 (2005).
doi:10.1086/429084.
HUNT, P. AND SARKAR, S.
Multiple ination and the WMAP glitches.
Phys.Rev., D70, p. 103518 (2004).
doi:10.1103/PhysRevD.70.103518.
INGROSSO, G., MERAFINA, M., RUFFINI, R. AND STRAFELLA, F.
System of self-gravitating semidegenerate fermions with a cutoff of energy and
angular momentum in their distribution function.
A&A , 258, pp. 223233 (1992).
JONES, B.J., MART
ESZ
ESZ
ESZ
AMPFER, B.
Gamma ashes from relativistic electron-positron plasma droplets.
Phys. Rev. A, 79(2), pp. 020103+ (2009).
doi:10.1103/PhysRevA.79.020103.
MYATT, J., DELETTREZ, J.A., MAXIMOV, A.V., MEYERHOFER, D.D., SHORT,
R.W., STOECKL, C. AND STORM, M.
Optimizing electron-positron pair production on kilojoule-class high-intensity
lasers for the purpose of pair-plasma creation.
Phys. Rev. E, 79(6), pp. 066409+ (2009).
doi:10.1103/PhysRevE.79.066409.
NAKAR, E., PIRAN, T. AND SARI, R.
Pure and Loaded Fireballs in Soft Gamma-Ray Repeater Giant Flares.
ApJ , 635, pp. 516521 (2005).
doi:10.1086/497296.
NARAIN, G., SCHAFFNER-BIELICH, J. AND MISHUSTIN, I.N.
Compact stars made of fermionic dark matter.
Phys. Rev. D, 74(6), 063003 (2006).
doi:10.1103/PhysRevD.74.063003.
NASELSKY, P.D., NOVIKOV, D.I. AND NOVIKOV, I.D.
The Physics of the Cosmic Microwave Background.
Cambridge Astrophysics (Cambridge Univ. Press, 2011).
NAVARRO, J.F., FRENK, C.S. AND WHITE, S.D.M.
A Universal Density Prole from Hierarchical Clustering.
ApJ , 490, p. 493 (1997).
doi:10.1086/304888.
201
Bibliography
NAVARRO, J.F., LUDLOW, A., SPRINGEL, V., WANG, J., VOGELSBERGER, M.,
WHITE, S.D.M., JENKINS, A., FRENK, C.S. AND HELMI, A.
The diversity and similarity of simulated cold dark matter haloes.
MNRAS , 402, pp. 2134 (2010).
doi:10.1111/j.1365-2966.2009.15878.x.
PACZYNSKI, B.
Gamma-ray bursters at cosmological distances.
ApJ , 308, pp. L43L46 (1986).
doi:10.1086/184740.
PACZYNSKI, B.
Super-Eddington winds from neutron stars.
ApJ , 363, pp. 218226 (1990).
doi:10.1086/169332.
PANDOLFI, S., FERRARA, A., CHOUDHURY, T., MELCHIORRI, A. AND MI-
TRA, S.
Data-constrained reionization and its effects on cosmological parameters (2011).
* Temporary entry *.
PANDOLFI, S., GIUSARMA, E., LATTANZI, M. AND MELCHIORRI, A.
Ination with primordial broken power law spectrum as an alternative to the con-
cordance cosmological model.
Phys. Rev., D81, p. 103007 (2010a).
doi:10.1103/PhysRevD.81.103007.
PANDOLFI, S. ET AL.
Harrison-Zeldovich primordial spectrum is consistent with observations.
Phys. Rev., D81, p. 123509 (2010b).
doi:10.1103/PhysRevD.81.123509.
PANDOLFI, S. ET AL.
Impact of general reionization scenarios on extraction of inationary parameters
(2010c).
PATRICELLI, B., BERNARDINI, M.G., BIANCO, C.L., CAITO, L., IZZO, L.,
RUFFINI, R. AND VERESHCHAGIN, G.
a New Spectral Energy Distribution of Photons in the Fireshell Model of GRBS.
International Journal of Modern Physics D, 20, pp. 19831987 (2011).
doi:10.1142/S0218271811020056.
PAULI, W.
Theory of relativity.
Dover books on relativity and related areas (Pergamon Press, 1958).
ISBN 9780486641522.
202
Bibliography
PEEBLES, P.J.E.
Recombination of the Primeval Plasma.
ApJ , 153, pp. 1+ (1968).
doi:10.1086/149628.
PEEBLES, P.J.E.
Principles of Physical Cosmology (Princeton University Press, 1993).
PEER, A.
Temporal Evolution of Thermal Emission from Relativistically Expanding Plasma.
ApJ , 682, pp. 463473 (2008).
doi:10.1086/588136.
PEER, A. AND RYDE, F.
A Theory of Multicolor Blackbody Emission from Relativistically Expanding Plas-
mas.
ApJ , 732, pp. 49+ (2011).
doi:10.1088/0004-637X/732/1/49.
PEER, A., RYDE, F., WIJERS, R.A.M.J., M
ESZ
ESZ
AROS, P.
Photosphere-internal shock model of gamma-ray bursts: case studies of Fermi/LAT
bursts.
MNRAS , 415, pp. 16631680 (2011).
doi:10.1111/j.1365-2966.2011.18807.x.
TREMAINE, S. AND GUNN, J.E.
Dynamical role of light neutral leptons in cosmology.
Phys. Rev. Lett. , 42, pp. 407410 (1979).
UEHLING, E.A.
Transport Phenomena in Einstein-Bose and Fermi-Dirac Gases. II.
Physical Review, 46, pp. 917929 (1934).
doi:10.1103/PhysRev.46.917.
UEHLING, E.A. AND UHLENBECK, G.E.
Transport Phenomena in Einstein-Bose and Fermi-Dirac Gases. I.
Physical Review, 43, pp. 552561 (1933).
doi:10.1103/PhysRev.43.552.
USOV, V.V.
Bare Quark Matter Surfaces of Strange Stars and e
+
e
Emission.
Phys. Rev. Lett. , 80, pp. 230233 (1998).
VESTERGAARD, M.
Determining Central Black Hole Masses in Distant Active Galaxies.
ApJ , 571, pp. 733752 (2002).
208
Bibliography
doi:10.1086/340045.
VESTERGAARD, M. AND PETERSON, B.M.
Determining Central Black Hole Masses in Distant Active Galaxies and Quasars.
II. Improved Optical and UV Scaling Relationships.
ApJ , 641, pp. 689709 (2006).
doi:10.1086/500572.
VIOLLIER, R.D., TRAUTMANN, D. AND TUPPER, G.B.
Supermassive neutrino stars and galactic nuclei.
Physics Letters B, 306, pp. 7985 (1993).
doi:10.1016/0370-2693(93)91141-9.
WALKER, M.G., MCGAUGH, S.S., MATEO, M., OLSZEWSKI, E.W. AND
KUZIO DE NARAY, R.
Comparing the Dark Matter Halos of Spiral, Low Surface Brightness, and Dwarf
Spheroidal Galaxies.
ApJ , 717, pp. L87L91 (2010).
doi:10.1088/2041-8205/717/2/L87.
WEINBERG, S.
Cosmology (Oxford University Press, April 2008., 2008).
WHITE, S.D.M.
Violent Relaxation in Hierarchical Clustering.
In O. Lahav, E. Terlevich, & R. J. Terlevich (ed.), Gravitational dynamics, p.
121 (1996).
WILKS, S.C., KRUER, W.L., TABAK, M. AND LANGDON, A.B.
Absorption of ultra-intense laser pulses.
Physical Review Letters, 69, pp. 13831386 (1992).
doi:10.1103/PhysRevLett.69.1383.
WILTSHIRE, D.L.
Cosmic clocks, cosmic variance and cosmic averages.
New Journal of Physics, 9, p. 377 (2007).
doi:10.1088/1367-2630/9/10/377.
WOODS, E. AND LOEB, A.
Constraints on Off-Axis X-Ray Emission from Beamed Gamma-Ray Bursts.
ApJ , 523, pp. 187191 (1999).
doi:10.1086/307738.
YADAV, J., BHARADWAJ, S., PANDEY, B. AND SESHADRI, T.R.
Testing homogeneity on large scales in the Sloan Digital Sky Survey Data Release
One.
MNRAS , 364, pp. 601606 (2005).
209
Bibliography
doi:10.1111/j.1365-2966.2005.09578.x.
YARESKO, R., MUSTAFA, M.G. AND K
AMPFER, B.
Relativistic expansion of electron-positron-photon plasma droplets and photon
emission.
Physics of Plasmas, 17(10), p. 103302 (2010).
doi:10.1063/1.3499410.
ZALALETDINOV, R. AND COLEY, A.
Averaging out Inhomogeneous Newtonian Cosmologies: III. The Averaged
Navier-Stokes-Poisson Equations.
ArXiv General Relativity and Quantum Cosmology e-prints (2002).
ZALALETDINOV, R.M.
Averaging out the Einstein equations.
General Relativity and Gravitation, 24, pp. 10151031 (1992).
doi:10.1007/BF00756944.
ZALALETDINOV, R.M.
Towards a theory of macroscopic gravity.
General Relativity and Gravitation, 25, pp. 673695 (1993).
doi:10.1007/BF00756937.
ZALALETDINOV, R.M.
Averaging problem in general relativity, macroscopic gravity and using Einsteins
equations in cosmology.
Bulletin of the Astronomical Society of India, 25, p. 401 (1997).
ZELDOVICH, Y.B., KURT, V.G. AND SYUNYAEV, R.A.
Recombination of Hydrogen in the Hot Model of the Universe.
Zhurnal Eksperimental noi i Teoreticheskoi Fiziki, 55, pp. 278286 (1968).
210