PHYSICAL REVIEW D VOLUME 40, NUMBER 2 15 JULY 1989

Dissipation of quantum fields from particle creation

E. Calzetta
Department of Physics, University of Alberta, Edmonton, Alberta, Canada T6G 2JI
B. L. Hu
Department of Physics and Astronomy, University of Maryland, Coilege Park, Maryland 20742
and Netvman Laboratory of Nuclear Studies, Cornell University, Ithaca, Netv Pork 14853-5001
(Received 15 February 1989)
%e discuss the nature and origin of the dissipation of quantum fields due to the back reaction of
particle creation. We derive the efFective action of a scalar gP theory in the closed-time-path-
integral formalism. From the real and causal equation of motion for the background field we
deduce a dissipative function for this process and for the cosmological anisotropy damping problem
studied earlier. This model illustrates that the appearance of dissipative behavior from the back re-
action of particle creation in quantum fields is a general feature. It also suggests that the role of
gravity in the display of dissipative behavior in semiclassical processes is not unique.

In this paper we ask the question: Is there dissipation ties of quantum processes. We want to see if there is
associated with particle creation and its back-reaction similar dissipative behavior in this back-reaction process
effects depicted in quantum field theory, specifically in and what are the basic assumptions entering into the
the effective-action formalism? A well-known example is analysis which leads to such behavior.
Euler and Heisenberg's 1936 semiclassical calculation The criteria we use to determine whether a system
and Schwinger's 1951 field-theoretical calculation of elec-
tron pair creation in a strong electromagnetic field. Our statistical mechanics. ''
shows dissipative behavior are based on first principles in
They involve (a) the separation
interest in problems of this nature stemmed from earlier of a part whose behavior we are interested in, which we
work in the back-reaction effect of particle creation in call the system, and the rest, which we call the bath or the
cosmological spacetimes, especially the so-called "anisot- environment, and their coupling (in the language of
ropy dissipation" problem. Despite our ability to do subsystems — the relevant and irrelevant parts); (b) the
full-scale calculations and produce the by now well- choice of boundary conditions (e.g. , the in-out vacuum
known results, the nature of "dissipation" in its most persistence amplitude or the in-in vacuum expectation
foundational and elemental statistical-mechanical sense value) which includes the stipulation of the phase relation
remains elusive. This prompted one of us to begin ques- of the initial states (such as pure, random, or thermal
tioning the notion of entropy defined for quantum states); and (c) a way to average out some degrees of free-
fields, ' and the entropy generated in particle production dom of the bath, known as coarse graining. All three
processes.
and others ' ' The recent work of
Hu, Kandrup, Sorkin,
provided much insight into the role played
steps need be stipulated for one to see dissipative behav-
ior in the system. These points can be illustrated by a
by correlation and coherence' in the statistical nature of simple example, that of coupled harmonic oscillators. '
dynamical quantum fields. However, because the pro- Referring to a single oscillator as the system and the rest
cesses studied by these authors are modeled by time- as the bath, by introducing certain averaging measure
dependent coupled harmonic-oscillator systems and par- (coarse graining) in the bath variables and choosing some
ticle production is mimicked by the process of parametric specific initial condition, one can see the dissipative be-
amplification, only entropy change from changes in the havior arising from the otherwise time-reversal-invariant
correlations due to interactions or from the changes in the dynamics of a classical system. Equivalently, one can use
phase relation of the states (coherence) of the system as it an effective action to describe these results, following the
evolves are considered. The problem of "dissipation" of same set of procedures as notably demonstrated recently
a field due to the back reaction of created particles re- by Caldeira and Leggett based on earlier work of Feyn-
quires instead a full quantum-field-theoretical descrip- man and Vernon. '" Dissipation in interacting quantum
tion. In Ref. 11 we have studied in detail the problem of field theory was also discussed in a recent paper of ours. '
back reaction of particle creation for free conformal fields There we presented the field-theoretical version of
in an anisotropic (Bianchi type-I) universe by way of the the Bogoliubov-Born-Green-Kirkwood- Yvon (BBGKY)
closed-time-path-integral formalism. ' We were able to hierarchy and used these criteria to show how dissipative
deduce a real, causal equation of motion for the back™ effects appear when this series is truncated and certain
ground field and from it identify the source of dissipation. averaging conditions on the higher-order correlation
In this paper we choose to analyze one of the simplest functions are introduced.
model fields theories, the gP theory, under the simplest Here we shall use these basic criteria to show that the
conditions — i.e., static, flat-space background. Putting back reaction of particle creation exhibits dissipative be-
aside the effect of dynamics and curvature enables us to havior even in the simplest setting. In a background-field
focus only on the field-theoretical and statistical proper- splitting one separates the quantum field into a back-

40 1989 The American Physical Society

 BRIEF REPORTS 657

ground field which is our system, and a fluctuation field tation the equation of motion for the system has the form
which acts as the bath. In setting up the problem one of a damped oscillator system or an LCR circuit. One
gives the appropriate boundary conditions (in-out or in- can thus easily identify the respective dissipative and
in), e.g. , by choosing the appropriate propagators which reactive components. So, as a by-product we also derive
contain the causal information of the system. Then in the dissipative function (or the related spectral density)
calculating the effective action one implicitly introduces a for the gP theory which has a structure similar to that of
coarse-graining procedure when the quantum Auctua- the cosmological anisotropy-damping process studied be-
tions are integrated out. ' We shall illustrate how these fore.
steps are taken in a typical field-theoretical calculation. The classical action of a scalar field with g(t interac-
This analysis serves to clarify the origin of dissipation in tion is
these field-theoretical processes. We find that both the
anisotropy-damping problem and the P theory embody S[C)]= d"x[ '(M&) —'m
f —, —, N —)gN —J@],
—, (l)
the same dissipative mechanism. In so doing we ascribe
the dissipation in the cosmological anisotropy-damping
where J
is an external source. In a background-field
decomposition N=P+y we may view the background
problem to the same set of basic assumptions and mea-
field P as our system and the Auctuation field y as our
sures introduced in a much broader context than that as-
bath variables. The closed-time-path effective action is
sociated with gravitational fields. In fact, we see that the
obtained after integrating out the Auctuating field
role of gravity in these semiclassical processes is not more
than providing the dynamics of the background field and I [P+, P ]= —i lnf Dy+Dy exp[i(S[P++y+]
the energy for particle creation. '
In this calculation we shall derive the effective action —S[P +y ])],
in the closed-time-path-integral formalism. ' The energy
density associated with particle production is measured (2)
by the expectation value of the energy-momentum tensor where the + superscripts
on P and y denote the positive
with respect to the same "in" vacuum. The advantage is and negative time branches. To order only one-loop g,
that the equation of motion for the background field is graphs contribute which contain only linear Auctuations
real and causal. When written in the frequency represen- around P+ and P
I

r[y+, y-]=S[y+] —S[y-] —i ln Dy+Dy f exp —f d x [[(By+ ) —m (y+) ] —[(By )

—m (P ) ]]

——
X l f d x g[P+(y+) —P (y ) ] —,' f d x d —x'g [P+(y+) P(y ) ](x—
)

X[/+(y+) —P (y ) ](x') . (3)

Since the measure is Gaussian, we may use the Wick theorem to get

+(ig'/8) f d x d x'([P+(y+)' P(y )']( P+( y+)'

x)[— P(y '))„„„„„g
)'](x—
The Feynman rules are
(y+(x)y+(x')) =i f [d p/(2~)"]e'~' ' '/(p — m +ie)=(0~ T[y(x)y(x')]~)0),
(y+(x)y (x')) = i f [d "p/(2—
m)"]e'~'" '2~i5(p m)8(p )=(— O~y(x')y(x)~0) . (5b)
The part which contributes to the linearized equations for P is then I
l,„, [(t+, P ]= ,' d x— f P—+(x)(CI+m )P+(x)
~
2
'P+( ) P+( ') j 71

(2~)"
P (P(X X )
' —m '+ i e) [(p —k)' —m '+ i e]
(2~)" (&

gn
eip(x —x') 0 "k
(x ) f (2'p)" f (2~)" [2~i g(i(. 2 m 2)g(i 0)]

X[2vri6((p —k) —m )0((p —k) )] . (6)

functions are to vanish. For p ) )

It is easy to see that the second k integral vanishes if p (0, and that p —k 0 for p 0 if the arguments of both 6
0, the second integral is given by twice the imaginary part of the first. ' The first in-
)
658 BRIEF REPORTS

tegral is evaluated by introducing a Feynman integration parameter x, rotating k to Euclidean space, developing in
powers of a=n — 4, and choosing the cut along the negative real axis to ensure positivity in the argument of the loga-
rithm. We get the one-loop effective action to order g:
dx x +m x

4(4~) f d x d x'P+(x) f (2')' e'~'" " 'f dx P+(x') ln

4~p
i~—
8(p x(l —x) —m )

—2P (x')[ 2i—

~8(p x(1 —x) —m )]9(p ) . (7)

The equation of motion is found by taking the variation with respect to P+(x) and identifying P+ and P (see Ref.
11). In terms of the Fourier function P(p) = d x e 't'"P(x) f we get

p2 — — + ,'i~0—(p x(1 —x) —m —J(p) .

m2
(4~)' f dx 'ln
—,
4' )sgn(p ) P(p)=

With the above choice of the branch for the argument of a complex number, we can write =p

—,
(co )

—m'z g' ' [(~—ie)' —Ikl']x (I —x) I '"

Im —
co' —lkl'2
'
)'
dx ln P(p) = —J(p) .
(47r &4vrp

We see that the equation for P(p) has an imaginary term given in Ref. 15. We can show that the energy dissipated
given by in the background field as represented by this complex
1/2 term is exactly equal to the energy of the coherent pairs
g l&
sgn(co) 1—4m 0(p —4m ) . (10) emitted by the source.
(4~) 2 p The total energy dissipated over the whole history is
@= — f dt F, P, where F, is the dissipative force. In
This corresponds to a viscous force term F, = yP in the
equation of motion for a damped harmonic oscillator, or
Fourier representation 8= f
dco cog*(co)1mb, '(co)P(co),
I
the resistance term in a CR circuit which signifies the where 6 is the propagator in Eq. (9). Using the optical
appearance of dissipation. In fact, if we write the imagi- theorem one can show that
nary term as y(p)coP(p), the dissipative function y is
given by
@= 2 f dco(2co) g I
A( I~a)A(co)P(co)l~, (12)
/(4~ )][P(p)IIcoI ]8(p
y'(p) =[7rg —4m ), (11)
where 3 (1 — +a) is the transition amplitude from a one-
where P(p) = [1 —(4m /p )]'~ . particle state to any pair state a. The integrand gives the
Dissipation arises because by looking at the expecta- total probability for the creation of a particle pair with
tion value of the field we are only considering one- energy 2', which is the only mode up to order g .
particle states. (The expectation value of the field is relat- Another way to see the dissipative effect of particle
ed through the reduction formula to the amplitude for creation is to examine the impulsive response of the sys-
the source to emit one particle. ) But the source can emit tem to see if the background field will damp away in time.
also pairs, triplets, etc. , whose energy are not accounted Thus, we introduce a 6-function source into the equation
for if only the background-field evolution is observed. of motion (9) and consider the behavior of any spatial
This is related to the explanation of dissipation arising Fourier component of the field pk(t) (system being spa-
from truncating the higher-order correlation functions tially homogeneous):
I

g' [(co —ie)' —Ikl']x (1 —x)

Pk(co) = —
(4~)' f 1,
~

CO CO
I (13)
Ul

where cok = lkl +m . The short-time behavior of pk(t) is ft /2COk

determined by the large-co behavior of pk(co). For large Pk(t)-(sincokt/cok )e
co (co — lkl &)4m ), the inverse propagator (9) acquires valid for t (((4cok/g2). We see that pk(t) is indeed
an imaginary part iy — [g /2(4') ]sgn(co). We may ap- damped, at least for short times. At late times, the tran-
proximate Eq. (9) as' sient contribution dies out entirely, and pk(t) becomes a
I [~k (~ 1'/2cok ) ]' —~k I o—
k(~)— simple oscillation with frequency cok (we are neglecting
the difference between the minimally substracted mass
which leads immediately to m and the physical mass). In this regime
 BRIEF REPORTS 659

Pk ( t ) -B '( sincoi, t /co i, ), (16) In this paper, through a simple model problem, we
have demonstrated the following points.
where B = 1+ C (g /m ) and C is a positive constant. (1) The existence of dissipation to the background field
p —m,
B ' is the residue of the propagator at the pole
due to particle creation. The back reaction of particles
being related to the dissipative function through the
dispersion relation for the propagator. B 1 implies that
the system is damped, and vice versa.
) created; this form of Lenz-law' ' '
created tends to diminish the source where they are
behavior noticed in
cosmological back-reaction problems thus acquires a
Using the same line of reasoning, one can also deduce more general field-theoretical explanation.
the dissipative function for the cosmological anisotropy (2) A theoretical explanation of the nature of dissipa-
damping problem. Earlier, we derived an in-in effective tion in these processes in terms of basic statistical-
action and a real, causal equation of motion for the rate mechanical premises. Their physical meaning is clarified
of change Ir;~. =/3,'~ (with respect to conformal times) of by relating the amount of particle creation to the energy
anisotropy P; in the metric [Eq. (3.24) of Ref. 11]. In the dissipation. As a concrete result the dissipative functions
frequency domain, if we write the imaginary term of both the gP model field theory and the anisotropy-
F, (co);~ =iy(co)a;J(co), we can identify the dissipative damping problem are derived.
function y(co) to be (3) The dissipative nature of back-reaction problems in
field theory is not special to gravitational fields. This ex-
y(co) =[~/60(4m ) ]ico~ (17)
ample helps to correctly identify the role of gravity in the
Since the conformal scalar field is massless there is no class of semiclassical theories. '
threshold and thus this term is nonzero for all frequen- Further discussion on the theoretical implication of
cies. We can show that the total anisotropy energy dissi- these and related issues can be found in Ref. 20. Applica-
pated using the expression for damped harmonic oscilla- tion of this method to a derivation of the viscosity func-
tors tion associated with particle creation in the reheating
epoch of the new inAationary cosmology is being carried
E= I dt +,,)(t)lr,', = J (dco/2')(colr, *) )(ycolr'J) out by Stylianopoulos. '
is indeed equal to the total energy of the particle pairs
created [Eq. (3.29) of Ref. 11]. This provides a clear This research was supported in part by the U. S. Na-
physical meaning of dissipation of anisotropy by particle tional Science Foundation under Grant No. PHY87-
creation. 17155 and the Canadian Research Council.

For cosmological particle creation see, e.g. , L. Parker, Phys. (1965)]. For a recent review, see K. C. Chou et al. , Phys. Rep
Rev. 1S3, 1057 (1969); Ya. B. Zel'dovich, Pis'ma. Zh. Eksp. 118, 1 (1985).
Teor. Fiz. 12, 443 (1970) [JETP Lett. 12, 307 (1970)]. See, e.g. , R. Zwanzig, in Lectures in Theoretical Physics III,
For an introduction to cosmological back-reaction problems, edited by W. E. Britten (Interscience, New York, 1961), p.
see, e.g. , B. L. Hu, in Proceedings of the Second Marcel 106; R. Rubin, J. Math. Phys. 1, 309 (1960); 2, 373 (1961); G.
Grossmann Meeting on General Relativity, edited by R. W. Ford, M. Mac, and P. Mazur, ibid. 6, 504 (1963); R. Bales-
Ruffini (North-Holland, Amsterdam, 1982). cu, Equilibrium and Nonequilibriurn Statistical Mechanics
See, e.g. , B. S. DeWitt, Phys. Rep. 19C, 297 (1975). (Wiley, New York, 1975).
4W. Heisenberg and H. Euler, Z. Phys. 98, 714 (1936); J. A. O. Caldeira and A. J. Leggett, Ann. Phys. (N. Y.) 149, 374
Schwinger, Phys. Rev. 82, 664 (1951). (1983); R. Feynman and F. L. Vernon, ibid. 24, 118 (1963).
5Ya. B. Zel'dovich and A. A. Starobinsky, Zh. Eksp. Teor. Fiz. '5E. Calzetta and B. L. Hu, Phys. Rev. D 37, 2878 (1988).
61, 2161 (1971) [Sov. Phys. JETP 34, 1159 (1972)]; B. L. Hu B. L. Hu and Y. Zhang (unpublished).
and L. Parker, Phys. Rev. D 17, 933 (1978); J. B. Hartle and For a discussion of the role of gravity in these processes, see
B. L. Hu, ibid. 20, 1772 (1979); 21, 2756 (1980). B. L. Hu, in Proceedings of the Fifth Marcel Grossmann Meet
68. L. Hu, in Cosmology of the Early Universe, edited by L. Z. ing on General Relativity, Perth, Australia, 1988, edited by D.
Fang and R. Ruffini (World Scientific, Singapore, 1984); Phys. Blair and M. J. Buckingham (World Scientific, Singapore,
Lett. 90A, 375 (1982); 97A, 368 (1983). 1989).
7H. E. Kandrup, Class. Quantum Gravit. 3, L55 (1986). '~For a similar calculation see, e.g, Schwinger (Ref. 12) and J.
L. Bombelli et al. , Phys. Rev. D 34, 373 (1986). Schwinger, Ann. Phys. (N. Y.) 9, 169 (1960).
B. L. Hu and H. E. Kandrup, Phys. Rev. D 35, 1776 (1987); H. '9L. Parker, in Quantum Theory of Gravity, edited by S.
E. Kandrup, J. Math. Phys. 28, 1398 (1987). Christensen (Hilgar, Bristol, 1984).
B. L. Hu and D. Pavon, Phys. Lett. B 180, 329 (1986); H. E. B. L. Hu, in Proceedings of the Conference on Thermal Fields
Kandrup, Phys. Rev. D 37, 3505 (1988). and Their Applications, Cleveland, 1988 [Physica A (to be
'E. Calzetta and B. L. Hu, Phys. Rev, D 35, 495 (1987). published)].
' J. Schwinger, J. Math. Phys. 2, 407 (1961); L. V. Keldysh, Zh. A. Stylianopoulos, University of Maryland report (unpub-
Eksp. Teor. Fiz. 47, 1515 (1964) [Sov. Phys. JETP 20, 1018 lished).