1: \documentclass[preprint]{revtex4}

2: 

3: \usepackage{amsmath}

4: \usepackage{latexsym}

5: \usepackage{epsfig}

6: \usepackage{psfrag}

7: \usepackage{here}

8: 

9: 

10: % our layout

11: \renewcommand{\baselinestretch}{1.2}

12: 

13: \setlength{\evensidemargin}{0.2in}

14: \setlength{\oddsidemargin}{0.2in}

15: \setlength{\textwidth}{6in}

16: \setlength{\topmargin}{-0.5in}

17: \setlength{\textheight}{9in}

18: 

19: \setlength{\parindent}{0em}

20: \setlength{\parskip}{0.5\baselineskip}

21: 

22: 

23: 

24: % general command definitions

25: \newcommand{\beq}{\begin{equation}}

26: \newcommand{\eeq}{\end{equation}}

27: \newcommand{\beqa}{\begin{eqnarray}}

28: \newcommand{\eeqa}{\end{eqnarray}}

29: \newcommand{\del}{\partial}

30: 

31: 

32: % command definitions specific to this document

33: \newcommand{\vp}{``$+$''}

34: \newcommand{\vm}{``$-$''}

35: 

36: 

37: 

38: 

39: 

40: 

41: 

42: 

43: \begin{document}

44: 

45: 

46: 

47: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

48: %

49: %

50: %  Title

51: %

52: %

53: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

54: 

55: 

56: \preprint{BI-TP 2005/47}

57: 

58: \title{Boltzmann Collision Term} 

59: 

60: \author{Steffen~Weinstock\footnote{e-mail: steffen@physik.uni-bielefeld.de}}

61: \affiliation{Fakult\"at f\"ur Physik, Universit\"at Bielefeld, 33615 Bielefeld, Germany\vspace{0.2cm}}

62: 

63: 

64: 

65: 

66: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

67: 

68: 

69: 

70: 

71: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

72: %Create the title page

73: 

74: \begin{abstract}

75: 

76: We derive the Boltzmann equation for scalar fields using the Schwinger-Keldysh

77: formalism. The focus lies on the derivation of the collision term. We show that

78: the relevant self-energy diagrams have a factorization property. The collision

79: term assumes the Boltzmann-like form of scattering probability times

80: statistical factors for those self-energy diagrams which correspond to tree

81: level scattering processes. Our proof covers scattering processes with any

82: number of external particles, which come from self-energy diagrams with any

83: number of loops.

84: 

85: \end{abstract}

86: 

87: \maketitle

88: 

89: 

90: 

91: 

92: 

93: 

94: 

95: 

96: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

97: %

98: %

99: %  Introduction

100: %

101: %

102: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

103: 

104: \section{Introduction}

105: 

106: %

107: % General introduction

108: %

109: 

110: Kinetic theory has proven to be a very successful tool for the description of

111: systems out of equilibrium.  Important applications can be found in many fields

112: of physics,

113: %

114: current examples are early cosmology, or the theory of heavy

115: ion collisions, where one aims to understand how the matter produced in the

116: collision evolves, and in particular whether it thermalizes or not.

117: %

118: A very famous kinetic equation is the Boltzmann equation for the evolution of

119: the particle distribution functions in a sufficiently dilute system.

120: %

121: A first derivation of non-relativistic kinetic theory from field theory was

122: given by Kadanoff and Baym~\cite{Kad62}.  Using the Closed-Time-Path formalism,

123: also called Schwinger-Keldysh formalism, this derivation becomes simpler and

124: can also be extended to relativistic theories, see for

125: example~\cite{Li:1982gk,Calzetta:1986cq,Mrowczynski:1992hq,Klevansky:1997wm,PSW}.

126: %

127: In early treatments the collision term of the Boltzmann equation was simply

128: modeled heuristically: consider each reaction the particle can undergo, compute

129: the probability for this reaction by using the free particle cross section, and

130: multiply with the appropriate statistical factors, that is the Bose-enhanced or

131: Fermi-suppressed phase space distribution functions, respectively.

132: %

133: The collision term for a scalar particle that can undergo 2-to-2 scattering,

134: for example, is

135: %

136: \beqa

137:    d_t f_p

138:  &=&

139:    \frac{1}{4\omega_p}

140:    \int

141:    \frac{d^3k}{(2\pi)^32\omega_k}

142:    \frac{d^3p'}{(2\pi)^32\omega_{p'}}

143:    \frac{d^3k'}{(2\pi)^32\omega_{k'}}

144:    (2\pi)^4\delta^4(p' +  k' - p - k)

145: %

146: \label{Boltzmann}

147: \\[1ex]&&\hphantom{XXX}

148: %

149:    \times

150:    {\cal P}_{pkp'k'}

151:    \times

152:    \big[ f_{p'} f_{k'} (1+f_p) (1+f_k)

153:         -(1+f_{p'}) (1+f_{k'}) f_p f_k

154:    \big]

155: .

156: \nonumber

157: \eeqa

158: %

159: We refer to a collision term of this form as a standard Boltzmann collision

160: term.

161: %

162: In

163: references~\cite{Li:1982gk,Calzetta:1986cq,Mrowczynski:1992hq,Klevansky:1997wm}

164: the collision term of the resulting Boltzmann equation was derived for a number

165: of specific theories from the right hand side of the Kadanoff-Baym equation

166: considering self-energy diagrams with up to two loops.

167: %

168: To our knowledge the first paper with a computation that went beyond the 2-loop

169: self-energy is~\cite{Carrington:2004tm}.  The authors explicitly computed a

170: selection of self-energy diagrams with up to three or four loops in real,

171: scalar $\phi^3$- and $\phi^4$-theory, respectively, and managed to bring them

172: into a form like~(\ref{Boltzmann}). This computation was quite involved and

173: required the assistance of a computer system to handle the big number of terms

174: appearing in intermediate steps.

175: 

176: 

177: %

178: % This paper

179: %

180: 

181: In the present paper we consider a scalar theory with an unspecified

182: non-derivative self-interaction. We first show that the self-energies appearing

183: in the collision term have a useful factorization property.  In the strict

184: on-shell limit a certain class of self-energy diagrams then indeed leads to a

185: collision term of the form~(\ref{Boltzmann}), namely those that correspond to

186: tree level scattering diagrams with any number of external particles.

187: %

188: Not all contributions to the self-energy fit into this picture, which can

189: already be expected from the problems that arise when extending the vacuum

190: Cutkosky rules to finite temperature~\cite{Kobes,Gelis:1997zv}.

191: 

192: 

193: 

194: %

195: % Outline

196: %

197: 

198: In section~\ref{sec:basics} we give the basics of nonequilibrium field theory

199: as far as required for our purpose. We also show briefly how the flow term of

200: the Boltzmann equation is obtained from the Schwinger-Dyson equation.

201: %

202: This procedure is standard and has extensively been covered in literature

203: already, so only the basic steps are given here. For more details see for

204: example~\cite{Carrington:2004tm} and references therein.

205: %

206: Section~\ref{sec:collision-term} is the main part of this paper, where we study

207: the collision term for a real scalar field.

208: %

209: A remark on the extension to charged scalar fields can be found in

210: section~\ref{sec:complex}, and in section~\ref{sec:discussion} we finally

211: discuss our results.

212: 

213: 

214: 

215: 

216: 

217: 

218: 

219: 

220: 

221: 

222: 

223: 

224: 

225: 

226: 

227: 

228: 

229: 

230: 

231: 

232: 

233: 

234: 

235: 

236: 

237: 

238: 

239: 

240: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

241: %

242: %

243: %  Basics

244: %

245: %

246: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

247: 

248: \section{Basics}

249: \label{sec:basics}

250: 

251: An appropriate framework to study the time evolution of physical quantities in

252: nonequilibrium situations is given by the Schwinger-Keldysh or Closed-Time-Path

253: formalism~\cite{Schwinger:1960qe,Keldysh:1964ud,Chou:1984es}. The basic

254: technical point is that the time variable of all objects is defined on a path

255: $C$ on the real axis that leads from $-\infty$ to $+\infty$, and then back to

256: $-\infty$. The definition of the Green function is still

257: %

258: \beq

259:   \Delta(x, y) = -i \left< T_C \phi(x) \phi^\dagger(y) \right>

260: ,

261: \label{Green_function:def:path}

262: \eeq

263: %

264: but time ordering is performed along the path $C$ here. If we split the path

265: into a \vp-branch from $-\infty$ to $+\infty$ and a \vm-branch going back to

266: $-\infty$, we can distinguish four real time Green functions, differing by the

267: branches on which the time arguments are placed:

268: %

269: \beqa

270:       i\Delta^{++}(x, y)

271:   &=& i\Delta^t(x, y)

272:    =  \left< T \phi(x) \phi^\dagger(y) \right>

273: ,

274: \label{Green_function:def:index}

275: \\

276:       i\Delta^{+-}(x, y)

277:   &=& i\Delta^<(x, y)

278:    =  \left< \phi^\dagger(y) \phi(x) \right>

279: ,

280: \nonumber\\

281:       i\Delta^{-+}(x, y)

282:   &=& i\Delta^>(x, y)

283:    =  \left< \phi(x) \phi^\dagger(y) \right>

284: ,

285: \nonumber\\

286:       i\Delta^{--}(x, y)

287:   &=& i\Delta^{\bar{t}}(x, y)

288:    =  \left< \bar{T} \phi(x) \phi^\dagger(y) \right>

289:  .

290: \nonumber

291: \eeqa

292: %

293: The $\bar{T}$ in the last line denotes anti time ordering.  These functions are

294: not independent of each other, but are connected via the relation

295: %

296: \beq

297:   \Delta^t + \Delta^{\bar{t}} = \Delta^< + \Delta^>

298: .

299: \label{Green_function:relation}

300: \eeq

301: %

302: In addition, one defines the retarded and advanced Green functions:

303: %

304: \beq

305:    \Delta^{R,A}(x, y)

306:  = \Delta^t(x, y) - \Delta^{<,>}(x, y)

307: .

308: \label{Green_function:def:ret_adv}

309: \eeq

310: %

311: In contrast to ordinary perturbation theory in vacuum, here each internal

312: vertex can be either of type \vp or of type \vm. The lines between vertices

313: represent the Green functions defined in~(\ref{Green_function:def:index}),

314: depending on the types of the incident vertices. For each vertex of type \vm

315: there is an additional factor $-1$.

316: 

317: 

318: 

319: %

320: % Schwinger-Dyson equation 

321: %

322: 

323: The Schwinger-Dyson equation for the contour Green

324: function~(\ref{Green_function:def:path}) leads to the following equation of

325: motion:

326: %

327: \beq

328:     \left( \del^2_x + m^2 \right) \Delta(x,y)

329:   = - \delta^4_C(x-y)

330:     - \int_C d^4z \, \Pi(x,z) \Delta(z,y)

331: .

332: \label{Green_function:EOM:path}

333: \eeq

334: %

335: The self-energy $\Pi$ is defined as $i$ times the sum of all one-particle

336: irreducible two-point functions. We switch to index notation and obtain the

337: equation of motion for $\Delta^{<,>}$:

338: %

339: \beq

340:     \left( \del^2_x + m^2 \right) \Delta^{<,>}(x,y)

341:   = \int d^4z \,

342:      \left(\Pi^{<,>}(x,z) \Delta^A(z,y) - \Pi^R(x,z) \Delta^{<,>}(z,y) \right)

343: .

344: \label{Green_function:EOM:index}

345: \eeq

346: %

347: For simplicity we omit a potential tadpole contribution to the self-energy

348: $\Pi$. We assume that its only effect is a shift in the mass of the particles

349: which possibly introduces a force on the left hand side of the resulting

350: Boltzmann equation, but that doesn't change the argumentation concerning the

351: collision term.

352: 

353: 

354: %

355: % Wigner transform

356: %

357: 

358: Our goal is an equation that describes the evolution of the phase space density

359: of particles, which takes place on a macroscopical scale. In order to separate

360: this from the quantum evolution on a microscopical scale, we perform a Wigner

361: transformation. For any two-point function we introduce the average coordinate

362: %

363: \beq

364:   X = \frac 12 (x+y)

365: \label{averagw_coordinate:def}

366: \eeq

367: %

368: and carry out the Fourier transformation with respect to the relative

369: coordinate:

370: %

371: \beq

372:   \Delta(X, p) = \int d^4(x-y) \, \text{e}^{ip\cdot(x-y)} \Delta(x, y)

373: .

374: \label{Wigner_transform:def}

375: \eeq

376: %

377: Note that the functions $i\Delta^{<,>}(X, p)$, called Wigner functions, are

378: real, while $i\Delta^t(X,p)$ is the complex conjugate of

379: $i\Delta^{\bar{t}}(X,p)$. The respective self-energies have the same

380: properties.

381: 

382: 

383: We apply the Wigner transform to the equation of

384: motion~(\ref{Green_function:EOM:index}). The fact that we are dealing with

385: two-point functions is reflected in the appearance of an infinite series of

386: derivatives

387: %

388: \beq

389:    \diamond \{(1)\} \{(2)\}

390:  = \frac 12 \left(  \del^{(1)}   \cdot \del_p^{(2)}

391:                   - \del_p^{(1)} \cdot \del^{(2)}

392:             \right) \{(1)\} \{(2)\}

393: \label{diamond:def}

394: \eeq

395: %

396: with respect to $X$ and $p$:

397: %

398: \beqa

399: &&

400:   \left( - p^2 + m^2 -ip\cdot\del

401:          + \frac 14 \del^2

402:   \right) \Delta^{<,>}(X, p)

403: \label{Wigner:EOM}

404: \\

405: &&\hphantom{XXX}

406:   = \text{e}^{-i\diamond}

407:    \Big(   \{ \Pi^{<,>}(X, p) \} \{ \Delta^A(X, p) \}

408:           - \{ \Pi^R(X, p) \} \{ \Delta^{<,>}(X, p) \}

409:    \Big)

410: .

411: \nonumber

412: \eeqa

413: %

414: 

415: 

416: %

417: % Gradient expansion

418: %

419: 

420: Now some approximations are necessary.

421: %

422: We assume that the functions of interest have a smooth macroscopic behavior,

423: more precisely, we assume that the scale on which the functions change with

424: respect to the average coordinate $X$ is much bigger than the microscopical

425: scale set by the de Broglie wavelength of the particles. Consequentially, the

426: mixed derivative $\del\cdot\del_p$ is a small quantity and kept only up to

427: linear order.

428: %

429: We furthermore assume that the macroscopical scale is also large compared to

430: the particles' Compton wavelength, so that the second order derivative on the

431: left hand side of~(\ref{Wigner:EOM}) is negligible as well.

432: %

433: Since the Wigner functions $i\Delta^{<,>}$ are real, we can extract the real

434: and imaginary part of the equation of motion and find

435: %

436: \beqa

437:      \left( p^2 - m^2 \right) i\Delta^{<,>}

438:  &=& -\frac i2

439:        \left(   \Pi^{<,>} \, \text{Re}\,\Delta^R

440:               + \text{Re}\,\Pi^R \, \Delta^{<,>}

441:        \right)

442:       - \frac 12 \diamond \left( \Pi^< \Delta^> - \Pi^> \Delta^< \right)

443: ,\,\,

444: \label{EOM:real}

445: \\

446:      \left(-p \cdot \del\right) i\Delta^{<,>}

447:  &=& \frac 12 \left( \Pi^> \Delta^< - \Pi^< \Delta^> \right)

448:      - \frac i2 \diamond

449:         \left(   \Pi^{<,>} \, \text{Re}\,\Delta^R

450:                + \text{Re}\,\Pi^R \, \Delta^{<,>}

451:         \right)

452: .

453: \label{EOM:imaginary}

454: \eeqa

455: %

456: The real part has the form of a constraint, the imaginary part has the form of

457: a kinetic equation.

458: 

459: 

460: 

461: 

462: 

463: %

464: % Free field

465: %

466: 

467: For a free field we can write down the solutions of these equations in the form

468: %

469: \beqa

470:   i\Delta_0^<(X, p) &=& 2\pi \delta(p^2-m^2) \text{sgn}(p_0) n_0(X, p)

471: ,

472: \label{Green_free_<>}

473: \\

474: %

475:   i\Delta_0^>(X, p) &=& 2\pi \delta(p^2-m^2) \text{sgn}(p_0)

476:                           \big(1+ n_0(X, p)\big)

477: \nonumber

478: .

479: \eeqa

480: %

481: The solutions for the chronological and antichronological Green functions in

482: the free case are

483: %

484: \beqa

485:   i\Delta_0^t(X, p) &=& \frac{i}{p^2-m^2+i\text{sgn}(p_0)\epsilon}

486:              + 2\pi \delta(p^2-m^2) \text{sgn}(p_0) n_0(X, p)

487: ,

488: \label{Green_free_ttbar}

489: \\

490: %

491:   i\Delta_0^{\bar{t}}(X, p) &=& \frac{-i}{p^2-m^2+i\text{sgn}(p_0)\epsilon}

492:              + 2\pi \delta(p^2-m^2) \text{sgn}(p_0) \big(1 + n_0(X, p)\big)

493: .

494: \nonumber

495: \eeqa

496: %

497: In thermal equilibrium the KMS relation determines $n_0$ to be the

498: Bose-Einstein distribution, but in a nonequilibrium situation this function is

499: not known a priori.

500: 

501: 

502: 

503: %

504: % Small coupling expansion

505: %

506: 

507: The next simplification is a small coupling expansion. Later this will be used

508: for a detailed analysis of the collision term, here we need it to get rid of

509: those terms on the right hand side which are suppressed by both the mixed

510: derivative $\del\cdot\del_p$ and the coupling constant.

511: %

512: At this point we also demand that our system can be described in terms of

513: (quasi-)particles. To this end we assume that the right hand side of the

514: constraint equation~(\ref{EOM:real}) vanishes, turning this equation into a

515: mass-shell condition.

516: %

517: In the end, equations~(\ref{EOM:real}) and~(\ref{EOM:imaginary}) simplify to

518: %

519: \beqa

520:      \left( p^2 - m^2 \right) i\Delta^{<,>}

521:  &=& 0

522: ,

523: \label{constraint}

524: \\

525:      \left(-p \cdot \del\right) i\Delta^{<,>}

526:  &=& \frac 12 \left( \Pi^> \Delta^< - \Pi^< \Delta^> \right)

527: ,

528: \label{kinetic}

529: \eeqa

530: %

531: and we can make an on-shell ansatz for the Wigner functions:

532: %

533: \beqa

534:   i\Delta^<(X, p) &=& 2\pi \delta(p^2-m^2)

535:                   \Big[  \theta(p_0)  f_+(X, \vec{p}\,)

536:                         +\theta(-p_0) \big( 1+f_-(X, -\vec{p}\,) \big)

537:                   \Big]

538: ,

539: \label{final:ansatz:Wigner}

540: \\

541:   i\Delta^>(X, p) &=& 2\pi \delta(p^2-m^2)

542:                   \Big[  \theta(p_0)  \big( 1+f_+(X, \vec{p}\,) \big)

543:                         +\theta(-p_0) f_-(X, -\vec{p}\,)

544:                   \Big]

545: .

546: \nonumber

547: \eeqa

548: %

549: Spectral sum rules that follow from the basic commutator relations for the

550: scalar field operators and make a connection between $i\Delta^<$ and

551: $i\Delta^>$ ensure that this ansatz is consistent.

552: %

553: By comparison with the equilibrium functions and also by inserting the

554: ansatz~(\ref{final:ansatz:Wigner}) into the expressions for the expectation

555: values of current

556: %

557: \beq

558:   j^\mu(X) = 2\int \frac{d^4p}{(2\pi)^4} \, p^\mu \, i\Delta^<(X, p)

559: \label{current:def}

560: \eeq

561: %

562: and energy-momentum

563: %

564: \beq

565:   T^{\mu\nu}(X) = \int \frac{d^4p}{(2\pi)^4} \, p^\mu p^\nu \, i\Delta^<(X, p)

566: ,

567: \label{energy-momentum:def}

568: \eeq

569: %

570: we finally can identify $f_+$ and $f_-$ with the phase space densities of

571: particles and anti-particles, respectively.

572: %

573: In the case of real scalar fields there is an additional relation,

574: $i\Delta^<(-p)=i\Delta^>(p)$, which leads to $f_+=f_-$.

575: 

576: 

577: 

578: %

579: % Final flow term

580: %

581: 

582: The constraint equation is satisfied identically with this ansatz, and all that

583: is left over is the kinetic equation~(\ref{kinetic}).  We

584: insert~(\ref{final:ansatz:Wigner}) and integrate over positive frequencies. The

585: resulting equation reproduces the flow term of a relativistic Boltzmann

586: equation for particles with phase space density $f_+$:

587: %

588: \beq

589:   \left( \del_t + \frac{\vec{p}}{\omega}\cdot\vec{\del} \right) f_+(X, \vec{p})

590:   = \int_0^\infty \frac{dp_0}{\pi} \, C(X, p)

591: .

592: \label{kinetic-equation:final}

593: \eeq

594: %

595: Integration over negative momenta results in a similar equation for the

596: corresponding antiparticles. The most important part for us is the right hand

597: side: the collision term so far is

598: %

599: \beq

600:   C(X, p) = \frac{1}{2}

601:           \Big(   i\Pi^>(X, p) \, i\Delta^<(X, p)

602:                 - i\Pi^<(X, p) \, i\Delta^>(X, p)

603:           \Big) 

604: .

605: \label{collision-term:def}

606: \eeq

607: %

608: In the next section we try to re-express this in terms of particle densities

609: and scattering amplitudes.

610: 

611: 

612: 

613: 

614: 

615: 

616: 

617: 

618: 

619: 

620: 

621: 

622: 

623: 

624: 

625: 

626: 

627: 

628: 

629: 

630: 

631: 

632: 

633: 

634: 

635: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

636: %

637: %

638: %  Collision term

639: %

640: %

641: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% 

642: 

643: \section{Collision term}

644: \label{sec:collision-term}

645: 

646: 

647: Now we come to the main part of this paper, the collision term. We make a

648: perturbative expansion of the self-energies in $C$ and try to bring it to a

649: form resembling the collision term of a Boltzmann equation like shown in the

650: introduction.

651: %h

652: The collision term is local in our approximation, so we drop the argument $X$

653: from now on in order to simplify the notation.

654: 

655: 

656: 

657: \subsection{Self-energy}

658: \label{subsec:self-energy}

659: 

660: %

661: % Classification

662: %

663: 

664: In a perturbative expansion the self-energy $i\Pi^<(p)=i\Pi^{+-}(p)$ is

665: expressed as the sum of all amputated one-particle irreducible graphs with

666: momentum $p$ entering at a \vp-vertex and leaving at a \vm-vertex. We first

667: classify the graphs in this expansion in the following way:

668: %

669: take any graph and imagine all lines connecting a \vp-vertex with a \vm-vertex

670: were cut. This would split the graph into a number of connected subgraphs, each

671: containing only \vp- or \vm-vertices, respectively. We call these subgraphs

672: {\em clusters}.

673: %

674: Obviously there are at least two clusters, namely one which is connected to the

675: incoming line, called \vp-base, and one connected to the outgoing line, called

676: \vm-base.

677: %

678: Now we can distinguish two types of graphs:

679: %

680: \begin{enumerate}

681: \item

682: %

683:   Graphs of type 1 have only direct connections between the \vp- and the

684:   \vm-base, i.e. there are no paths leading from the \vp-base to the \vm-base

685:   via some other clusters (see Fig.~\ref{fig:type1}).  The simplest example for

686:   this are graphs with only two clusters.

687: %

688: \item

689: %

690:   Graphs of type 2 have connections between the \vp-base and the \vm-base via

691:   other clusters (see Fig.~\ref{fig:type2}).

692: \end{enumerate}

693: %

694: %

695: %

696: \begin{figure}[!tb]

697: \unitlength=1in

698: %

699: \begin{minipage}[t]{2.8in}

700: \begin{center}

701: \psfrag{A}[cc]{$\scriptstyle -$}    

702: \psfrag{B}[cc]{$\scriptstyle -$}    

703: \psfrag{C}[cc]{$\scriptstyle +$}    

704: \psfrag{D}[cc]{$\scriptstyle -$}    

705: \psfrag{E}[cc]{$\scriptstyle +$}    

706: \psfrag{F}[cc]{$\scriptstyle +$}    

707: \psfrag{a}[cc]{$\scriptstyle p$}    

708: \psfrag{b}[cc]{$\scriptstyle p$}    

709: \includegraphics[width=2.2in,height=1in]{Figures/type1.eps}

710: \end{center}

711: \caption{\small A graph with only direct connections between

712: \vp- and \vm-base (type 1).

713: There is a unique complete cut (dashed line).}

714: \label{fig:type1}

715: \end{minipage}

716: %

717: \hfill

718: %

719: \begin{minipage}[t]{2.8in}

720: \begin{center}

721: \psfrag{A}[cc]{$\scriptstyle -$}    

722: \psfrag{B}[cc]{$\scriptstyle -$}    

723: \psfrag{C}[cc]{$\scriptstyle +$}    

724: \psfrag{D}[cc]{$\scriptstyle -$}    

725: \psfrag{E}[cc]{$\scriptstyle +$}    

726: \psfrag{F}[cc]{$\scriptstyle -$}    

727: \psfrag{G}[cc]{$\scriptstyle +$}    

728: \psfrag{H}[cc]{$\scriptstyle +$}    

729: \psfrag{a}[cc]{$\scriptstyle p$}    

730: \psfrag{b}[cc]{$\scriptstyle p$}    

731: \includegraphics[width=2.2in,height=1in]{Figures/type2.eps}

732: \end{center}

733: \caption{\small A graph with indirect connections between

734: \vp- and \vm-base (type 2).

735: Two different complete cuts are possible (dashed lines).}

736: \label{fig:type2}

737: \end{minipage}

738: %

739: \end{figure}

740: %

741: %

742: %

743: There is another way to classify the graphs contributing to $i\Pi^{+-}(p)$:

744: %

745: try to divide a given graph into two parts, one attached to the incoming line,

746: the other part attached to the outgoing line, by cutting a suitable set of

747: ``$+-$''-lines. Only cut lines which are attached to a \vp-vertex in the first

748: part and to a \vm-vertex in the second part, and which are effective in

749: separating the two parts.

750: %

751: This works with every graph, since the incoming line ends in a \vp-vertex,

752: while the outgoing line starts from a \vm-vertex.  We call this a complete cut.

753: %

754: It is not hard to see that a graph belongs to type 1 if and only if there is

755: exactly one such complete cut. In this case the cut lines are exactly the ones

756: connecting the \vp-base with the \vm-base.  Otherwise the graph belongs to type

757: 2.

758: 

759: 

760: 

761: 

762: 

763: 

764: 

765: 

766: 

767: \subsubsection*{Unique complete cut}

768: 

769: 

770: Let us first concentrate on the graphs of type 1, i.e. graphs which have a

771: unique complete cut. We further classify these diagrams according to the number

772: of cut lines:

773: %

774: \beq

775:   i\Pi^{+-}_{(\text{type 1})}(p) = \sum_{n=2}^\infty \, i\Pi^{+-}_n(p)

776: .

777: \label{classify-by-n}

778: \eeq

779: %

780: Now we claim that

781: %

782: \beqa

783:       -i\Pi^{+-}_n(p)

784:   &=& \frac{1}{n!}

785:       \int \frac{d^4k_1}{(2\pi)^4} \, \ldots \, \frac{d^4k_n}{(2\pi)^4}

786:       (2\pi)^4 \delta^4(p-k_1-\ldots-k_n)

787: \label{basic-claim}

788: \\[1ex]

789:   &&\hphantom{x}

790:        \times

791:        {\cal M}^+_{(1)}(p, -k_1, \ldots , -k_n)

792:        \,

793:        i\Delta^{+-}_0(k_1) \ldots i\Delta^{+-}_0(k_n)

794:        \,

795:        {\cal M}^-_{(1)}(-p, k_1, \ldots, k_n)

796: .

797: \nonumber

798: \eeqa

799: %

800: Symbolically we can write this equation as

801: %

802: \beq

803: %

804: \begin{minipage}{1.3in}

805: \begin{center}

806: \psfrag{a}[cc]{$\scriptstyle p$}

807: \psfrag{b}[cl]{$\scriptstyle p$}

808: \psfrag{c}[cc]{$\scriptscriptstyle +$}

809: \psfrag{d}[cc]{$\scriptscriptstyle -$}

810: \psfrag{e}[cc]{$\scriptscriptstyle (n)$}

811: \includegraphics[width=1in]{Figures/S.eps}

812: \end{center}

813: \end{minipage}

814: %

815:   = \frac{1}{n!} \int \{ dk_i \}

816: %  \hphantom{x}

817: %

818: \begin{minipage}{1.3in}

819: \begin{center}

820: \psfrag{a}[cc]{$\scriptstyle p$}

821: \psfrag{b}[cl]{$\scriptstyle k_1$}

822: \psfrag{c}[cl]{$\scriptstyle k_2$}

823: \psfrag{d}[cl]{$\scriptstyle k_n$}

824: \psfrag{e}[br]{$\scriptscriptstyle +$}

825: \psfrag{f}[bl]{$\scriptscriptstyle +$}

826: \psfrag{g}[bl]{$\scriptscriptstyle +$}

827: \psfrag{h}[cc]{$\scriptscriptstyle +$}

828: \includegraphics[width=1in]{Figures/Splus.eps}

829: \end{center}

830: \end{minipage}

831: %

832: %\hphantom{xx}

833: \{i\Delta^{+-}_0(k_i)\}

834: \hphantom{xx}

835: %

836: \begin{minipage}{1.3in}

837: \begin{center}

838: \psfrag{a}[cc]{$\scriptstyle p$}

839: \psfrag{b}[cr]{$\scriptstyle k_1$}

840: \psfrag{c}[cr]{$\scriptstyle k_2$}

841: \psfrag{d}[cr]{$\scriptstyle k_n$}

842: \psfrag{e}[bl]{$\scriptscriptstyle -$}

843: \psfrag{f}[br]{$\scriptscriptstyle -$}

844: \psfrag{g}[br]{$\scriptscriptstyle -$}

845: \psfrag{h}[cc]{$\scriptscriptstyle -$}

846: \includegraphics[width=1in]{Figures/Sminus.eps}

847: \end{center}

848: \end{minipage}

849: %

850: .

851: \nonumber

852: \eeq

853: %

854: The infinite sum of diagrams that contribute to the self-energy on the left

855: hand side factorizes into two other infinite sums which are identified as the

856: diagrammatic expansions of certain $n$-point functions.

857: %

858: The totally symmetric function ${\cal M}^+_{(1)}$ is the scattering amplitude

859: belonging to the amputated, connected, out-of-equilibrium $(n+1)$-point

860: function; positive momenta are entering, and all external momenta are attached

861: to \vp-vertices.  Basically it is the Fourier transform

862: %

863: \beq

864:     \Delta_{(1)}^+(q_1, \ldots, q_{n+1})

865:   = \prod_{j=1}^{n+1}

866:     \left[

867:     \int d^4x_j \,

868:     \text{e}^{iq_jx_j} \,

869:     \Box_{x_j} 

870:     \right]

871:     \Delta_{(1)}^+(x_1, \ldots , x_{n+1})

872: \eeq

873: %

874: of the out-of-equilibrium $(n+1)$-point function

875: (we use the same symbol for the function and its Fourier transform)

876: with time ordered fields

877: %

878: \beq

879:     \Delta_{(1)}^+(x_1, \ldots, x_{n+1})

880:   = \left< T \phi(x_1) \ldots \phi(x_{n+1}) \right>_\text{conn}

881: ,

882: \eeq

883: %

884: however with the additional restrictions that the graphs contributing to this

885: function must not have any corrections on the $p$-line, and for each $k_i$

886: there must be a path connecting it to $p$ which only contains \vp-vertices.

887: The derivative operators $\Box = \del^2 + m^2$ remove the external legs and we

888: explicitly take out the momentum conservation $\delta$-function,

889: %

890: \beq

891:     \Delta_{(1)}^+(q_1, \ldots, q_{n+1})

892:   = (2\pi)^4 \delta^4(q_1 + \ldots + q_{n+1})

893:     i {\cal M}_{(1)}^+(q_1, \ldots, q_{n+1})

894: ,

895: \eeq

896: %

897: so that ${\cal M}_{(1)}^+$ has the form of a scattering amplitude.

898: %

899: The functions ${\cal M}^-_{(1)}$ and $\Delta^-_{(1)}$ are defined analogously

900: with anti-time ordering.

901: 

902: 

903: 

904: 

905: In order to prove~(\ref{basic-claim}), we have to show two things:

906: %

907: first, the set of diagrams appearing in the perturbative expansion on the left

908: hand side is the same as the set of diagrams one finds on the right hand

909: side. Here we don't care about statistical factors of diagrams or how often they

910: appear.

911: %

912: This is done in a second step: it may be possible to obtain the same

913: diagram by putting together several different contributions from $\Delta^+$ and

914: $\Delta^-$. The number of such combinations together with the factor $1/n!$ and

915: the symmetry factors of these contributions must match the symmetry factor of

916: the resulting diagram on the left hand side.

917: 

918: 

919: \subsubsection*{Diagrams}

920: 

921: Any graph $G$ contributing to $\Pi^{+-}_n$ has a unique complete cut consisting

922: of $n$ lines.

923: %

924: The free propagators $i\Delta^{+-}_0$ associated to the cut lines and the

925: integrals over their momenta, named $k_1, \ldots, k_n$, are taken out of $G$

926: and written down explicitly on the right hand side of~(\ref{basic-claim}).

927: %

928: The remainders of $G$ to the ``left'' and to the ``right'' of the cut are

929: called $G^+$ and $G^-$, respectively.

930: %

931: Obviously $G^+$ is a Feynman diagram with $n+1$ external momenta attached to

932: \vp-vertices. We took away the $k_i$-lines, so it is amputated.

933: %

934: Since $G$ is one-particle irreducible, there can be no corrections on the

935: $p$-line, while there may be corrections on the $k_i$-lines.

936: %

937: Because $G$ is of type 1, the cut lines are exactly the ones that connect the

938: \vp-base to the \vm-base, so all external momenta in $G^+$ are attached to the

939: \vp-base and thus are connected with each other via paths that only include

940: \vp-vertices.

941: %

942: This shows that $G^+$ is a Feynman diagram that contributes to

943: $\Delta_{(1)}^+$, and likewise $G^-$ is a part of $\Delta_{(1)}^-$.

944: 

945: 

946: 

947: Conversely, take any contributions $G^+$ form $\Delta_{(1)}^+$ and $G^-$ from

948: $\Delta_{(1)}^-$. Together with the free propagators $i\Delta^{+-}_0$ and the

949: integrals over their momenta they make up an amputated Feynman diagram $G$ with

950: a unique complete cut and with momentum $p$ entering at a \vp-vertex and

951: leaving at a \vm-vertex.

952: %

953: Since there are at least two $k_i$-lines, the only way for $G$ not to be

954: one-particle irreducible would be to consist of a $G^+$ or a $G^-$ which can be

955: split into two parts, one connected to the $p$-line and one connected to the

956: $k_i$-lines, by cutting a single line. But this is impossible, because $G^+$

957: and $G^-$ don't have corrections on the $p$-line.

958: %

959: So the diagram $G$ contributes to the diagrammatic expansion of $\Pi^{+-}_n$,

960: which eventually proves that the same diagrams are appearing on both sides

961: of~(\ref{basic-claim}).

962: 

963: 

964: %

965: % Delta - functions

966: %

967: 

968: A comment about the momentum conservation $\delta$-function is in order here.

969: If one performs as many integrals over internal momenta as possible in the

970: diagrams $G^+$ and $G^-$, in both cases a $(2\pi)^4\delta^4(p-k_1-\ldots-k_n)$

971: is left over which is not part of the scattering amplitude.  One appears

972: explicitly in~(\ref{basic-claim}), the other one reduces to

973: $(2\pi)^4\delta^4(0)$ and is dropped according to the usual definition of the

974: self-energy.

975: 

976: 

977: 

978: 

979: \subsubsection*{Numbers}

980: 

981: Let us first have a closer look at the symmetry factor of a diagram $G$

982: contributing to $\Pi^{+-}_n$.  The symmetry factor of a diagram is the order of

983: the graph's symmetry group, which contains all permutations of lines and

984: vertices that do not alter the diagram. $G$ has a unique cut, and accordingly

985: we can distinguish lines and vertices in the part $G^+$ left of the cut from

986: lines and vertices in the part $G^-$ to the right of the cut, and they all are

987: topologically different from the cut lines. This means that there are no

988: symmetries exchanging lines or vertices in $G^+$ with those in $G^-$, and

989: neither are there symmetries that exchange cut lines with uncut ones.

990: %

991: Therefore the symmetry group of $G$ is the direct product of three groups: the

992: symmetry group of $G^+$, the symmetry group of $G^-$, and the symmetry group

993: $S$ of the cut lines (in the context of $G$, i.e. permutations of cut lines

994: that do not change $G$).

995: %

996: We call the orders of these groups $s_+$, $s_-$ and $s$, respectively, and so

997: the symmetry factor of $G$ is $1/(s_+ss_-)$.

998: %

999: In fact, things are a bit more complicated, since in graphs with tadpole-like

1000: structures, i.e. where both ends of some lines are attached to the same vertex,

1001: the above definition of the symmetry group is too narrow. However, such lines

1002: will never be cut, and therefore their symmetry properties are part of the

1003: symmetry groups of $G^+$ or $G^-$ and don't change this discussion.

1004: 

1005: 

1006: 

1007: Given a graph $G$ contributing to $\Pi^{+-}_n$, how many different combinations

1008: of a $G^+ \in \Delta^+$ and a $G^- \in \Delta^-$ produce this $G$?

1009: %

1010: First of all, the topologies of $G^+$ and $G^-$ are completely determined by

1011: $G$ and its cut. But there are several possibilities to name their external

1012: lines, which determines how they are put together. Some namings result in

1013: topologies different from $G$, so they contribute to a different graph and are

1014: not relevant here.

1015: %

1016: The rest of the namings corresponds exactly to the inequivalent renamings of

1017: the cut lines $k_i$ of $G$, where inequivalent means that they are not

1018: symmetries of $G$.

1019: %

1020: So the number of combinations of a $G^+$ and a $G^-$ that produce $G$ is equal

1021: to the number of inequivalent renamings or permutations of the cut lines in

1022: $G$.

1023: 

1024: 

1025: 

1026: 

1027: The group $P$ of all permutations $\sigma_i$ of cut lines in $G$, i.e.  all

1028: possibilities of renaming them, has the order $n!$

1029: %

1030: If we build the right cosets of $P$ relative to $S$,

1031: %

1032: \beq

1033:   S\sigma_1, S\sigma_2, \ldots, S\sigma_{n!}

1034: ,

1035: \label{coset:a}

1036: \eeq

1037: %

1038: then two sets $S\sigma_i$ and $S\sigma_j$ are either identical or

1039: disjoint~\cite{GroupTheory}.  So in fact there are only $f$ different sets

1040: %

1041: \beq

1042:   S\sigma_{i_1}, \ldots , S\sigma_{i_f}

1043: ,

1044: \label{coset:b}

1045: \eeq

1046: %

1047: each of which contains $s$ elements.

1048: %

1049: Since all elements within one $S\sigma_i$ belong to equivalent permutations

1050: of cut lines, the number of inequivalent permutations must be equal to $f$.

1051: %

1052: But the step from~(\ref{coset:a}) to~(\ref{coset:b}) only removes redundant

1053: elements and leaves their total number unchanged: $n!=f \cdot s$. Together with

1054: the symmetry factors of the parts $G^+$ and $G^-$ the complete combinatorial

1055: factor on the right hand side then is

1056: %

1057: \beq

1058:   \frac{f}{n!}\frac{1}{s_+s_-} = \frac{1}{s_+ss_-}

1059: \eeq

1060: %

1061: and exactly matches the symmetry factor of $G$.

1062: 

1063: 

1064: 

1065: 

1066: 

1067: 

1068: 

1069: \subsubsection*{Example}

1070: 

1071: As an example we consider two 2-loop contributions to the self-energy of a real

1072: scalar theory with a $\phi^3$-interaction, displayed in Fig.~\ref{fig:phi3}.

1073: %

1074: %

1075: \begin{figure}[!tb]

1076: %

1077: \begin{center}

1078: %

1079: \begin{minipage}[t]{4in}

1080: %

1081: \psfrag{a}[rt]{$\scriptstyle (a)$}

1082: \psfrag{v1}[cc]{$\scriptstyle +$}

1083: \psfrag{v2}[cc]{$\scriptstyle -$}

1084: \psfrag{v3}[rc]{$\scriptstyle +$}

1085: \psfrag{v4}[lc]{$\scriptstyle -$}

1086: \includegraphics[width=1.2in]{Figures/1to1a.eps}

1087: %

1088: \hspace*{1in}

1089: %

1090: \psfrag{a}[rt]{$\scriptstyle (b)$}

1091: \psfrag{v1}[cc]{$\scriptstyle +$}

1092: \psfrag{v2}[cc]{$\scriptstyle -$}

1093: \psfrag{v3}[cc]{$\scriptstyle +$}

1094: \psfrag{v4}[cc]{$\scriptstyle -$}

1095: \includegraphics[width=1.2in]{Figures/1to1b.eps}

1096: %

1097: \end{minipage}

1098: %

1099: \caption{\small

1100: Two 2-loop contributions to the self-energy of real scalar $\phi^3$-theory with

1101: a unique complete cut comprising $n=3$ lines. The symmetry factors of these

1102: diagrams are $1/2$ and $1$, respectively.  }

1103: \label{fig:phi3}

1104: %

1105: \end{center}

1106: %

1107: \end{figure}

1108: %

1109: %

1110: There is only one possible topology for the parts $G^+$ and $G^-$, namely a

1111: 2-to-2-scattering diagram. There are three possibilities for naming the

1112: external lines of this diagram, so that the relevant contribution is\\[1ex]

1113: %

1114: \beq

1115: %

1116: \left(\,

1117: %

1118: \begin{minipage}{0.6in}

1119: \begin{center}

1120: \psfrag{o1}[cl]{$\scriptstyle k_1$}

1121: \psfrag{o2}[cl]{$\scriptstyle k_2$}

1122: \psfrag{o3}[cl]{$\scriptstyle k_3$}

1123: \includegraphics[width=0.5in, height=0.4in]{Figures/1to3.eps}

1124: \end{center}

1125: \end{minipage}

1126: %

1127:   +

1128: %

1129: \begin{minipage}{0.6in}

1130: \begin{center}

1131: \psfrag{o1}[cl]{$\scriptstyle k_2$}

1132: \psfrag{o2}[cl]{$\scriptstyle k_1$}

1133: \psfrag{o3}[cl]{$\scriptstyle k_3$}

1134: \includegraphics[width=0.5in, height=0.4in]{Figures/1to3.eps}

1135: \end{center}

1136: \end{minipage}

1137: %

1138:   +

1139: %

1140: \begin{minipage}{0.6in}

1141: \begin{center}

1142: \psfrag{o1}[cl]{$\scriptstyle k_3$}

1143: \psfrag{o2}[cl]{$\scriptstyle k_1$}

1144: \psfrag{o3}[cl]{$\scriptstyle k_2$}

1145: \includegraphics[width=0.5in, height=0.4in]{Figures/1to3.eps}

1146: \end{center}

1147: \end{minipage}

1148: %

1149: \,\right)

1150: \times

1151: \left(\,

1152: %

1153: \begin{minipage}{0.65in}

1154: \begin{flushright}

1155: \psfrag{i1}[cr]{$\scriptstyle k_1$}

1156: \psfrag{i2}[cr]{$\scriptstyle k_2$}

1157: \psfrag{i3}[cr]{$\scriptstyle k_3$}

1158: \includegraphics[width=0.5in, height=0.4in]{Figures/3to1.eps}

1159: \end{flushright}

1160: \end{minipage}

1161: %

1162:   +

1163: %

1164: \begin{minipage}{0.65in}

1165: \begin{flushright}

1166: \psfrag{i1}[cr]{$\scriptstyle k_2$}

1167: \psfrag{i2}[cr]{$\scriptstyle k_1$}

1168: \psfrag{i3}[cr]{$\scriptstyle k_3$}

1169: \includegraphics[width=0.5in, height=0.4in]{Figures/3to1.eps}

1170: \end{flushright}

1171: \end{minipage}

1172: %

1173:   +

1174: %

1175: \begin{minipage}{0.65in}

1176: \begin{flushright}

1177: \psfrag{i1}[cr]{$\scriptstyle k_3$}

1178: \psfrag{i2}[cr]{$\scriptstyle k_1$}

1179: \psfrag{i3}[cr]{$\scriptstyle k_2$}

1180: \includegraphics[width=0.5in, height=0.4in]{Figures/3to1.eps}

1181: \end{flushright}

1182: \end{minipage}

1183: %

1184: \,\right).

1185: %

1186: \label{a-times-b}

1187: \eeq\\[1ex]

1188: %

1189: %

1190: From the resulting nine graphs, three correspond to the inequivalent ways of

1191: naming the cut lines in diagram Fig.~\ref{fig:phi3}(a), while the remaining six

1192: graphs belong to diagram Fig.~\ref{fig:phi3}(b).

1193: 

1194: 

1195: 

1196: 

1197: 

1198: 

1199: \subsubsection*{No unique cut}

1200: 

1201: Next we have to deal with those contributions to the self-energy with more than

1202: one complete cut (type 2). In this case we decide to use the cut which makes

1203: the left portion $G^+$ of the diagram as small as possible. This means to cut

1204: all lines that leave the \vp-base, except when they are not efficient in

1205: separating the two parts. In the example of Fig.~\ref{fig:type2} this is the

1206: left cut. This prescription leads to an unambiguously defined cut and we can

1207: repeat all the arguments already used above. The only difference concerns the

1208: nature of the right hand portion of the diagrams:

1209: %

1210: the contributions to ${\cal M}^-_{(2)}$ must have the property that at least

1211: one of the external $k_i$-lines is connected to the outgoing $p$-line only via

1212: paths which contain at least one \vp-vertex.

1213: %

1214: Note that ${\cal M}_{(1)}$ and ${\cal M}_{(2)}$ don't have any graphs in

1215: common.

1216: 

1217: 

1218: 

1219: 

1220: 

1221: 

1222: 

1223: 

1224: 

1225: \subsection{Collision term}

1226: \label{subsec:collision-term}

1227: 

1228: For real fields the $n$-point functions $\Delta^+$ and $\Delta^-$ are related

1229: by complex conjugation:

1230: %

1231: \beq

1232:    \Delta^-(k_1, \ldots, k_n) = {\Delta^+}^*(-k_1, \ldots, -k_n)

1233: .

1234: \eeq

1235: %

1236: By referring to the corresponding diagrammatic representation we can check that

1237: this holds for the restricted functions $\Delta^\pm_{(1)}$ and

1238: $\Delta^\pm_{(2)}$, too.

1239: %

1240: After sending $k_i \to -k_i$ the full collision term can thus be written

1241: %

1242: \beqa

1243: &&\hphantom{}

1244:     C(p)

1245:   = \frac {1}{2} \sum_{a=1,2} \sum_{n=2}^\infty \frac{1}{n!}

1246:     \int \frac{d^4k_1}{(2\pi)^4} \ldots \frac{d^4k_n}{(2\pi)^4} \,

1247:     (2\pi)^4\delta^4(p+k_1+\ldots+k_n)

1248: \label{Coll:general}

1249: \\&&\hphantom{XX}

1250:     \times \Big\{

1251:                     {\cal M}^+_{(1)}(p, k_1, \ldots, k_n)

1252:                     {{\cal M}^+_{(a)}}^*(p, k_1, \dots, k_n)

1253:                     i\Delta^>_0(k_1) \ldots i\Delta^>_0(k_n) i\Delta^>(p)

1254: \nonumber\\&&\hphantom{XXXX}

1255:                   - {{\cal M}^+_{(1)}}^*(-p, -k_1, \ldots, -k_n)

1256:                     {\cal M}^+_{(a)}(-p, -k_1, \dots, -k_n)

1257:                     i\Delta^<_0(k_1) \ldots i\Delta^<_0(k_n) i\Delta^<(p)

1258:            \Big\}

1259: \nonumber

1260: .

1261: \eeqa

1262: %

1263: The sum over $a=1,2$ represents the contributions of self-energy diagrams of

1264: type 1 and 2, respectively. We emphasize that this factorization is exact. The

1265: form~(\ref{Green_free_<>}) for the Wigner functions suggests

1266: that~(\ref{Coll:general}) is an expansion in the particle phase space

1267: density. But this is not the case, since the distribution function is also

1268: contained in the scattering matrix. A true expansion in the particle densities

1269: has been done in~\cite{Jeon}, which shares some technical similarities with our

1270: analysis.

1271: %

1272: 

1273: 

1274: 

1275: In order to proceed towards the standard collision term of a Boltzmann

1276: equation, we still have to overcome three difficulties.

1277: %

1278: First, a matrix element squared can obviously only be obtained for type 1

1279: diagrams, where $a=1$.

1280: %

1281: Second, the propagators corresponding to the cut lines are free ones, while the

1282: propagator for the $p$-line is a full one. In the Boltzmann equation, however,

1283: all distribution functions are expected to be of the same type.

1284: %

1285: And third, if we want to combine the scattering matrices of the last two lines,

1286: they have to be symmetric with respect to inverting all momenta.

1287: %

1288: The last point is satisfied if we assume that our system is symmetric under

1289: CP. Note that this refers to both the interaction and the initial conditions,

1290: since if the evolution starts with initial particle distributions $f$ that are

1291: not CP-even, then the matrix elements won't be CP-even either, even if the

1292: interaction conserves CP.

1293: %

1294: %

1295: The first two points force us to restrict ourselves to the tree

1296: level,

1297: %

1298: since then the difference between free and full propagators vanishes. In

1299: addition, the scattering diagrams include no internal \vm-vertices: diagrams

1300: with \vm-vertices have several ``$+-$''-lines which cannot all be on-shell in a

1301: tree diagram, and according to~(\ref{Green_free_<>}) these diagrams vanish.

1302: %

1303: 

1304: The fact that we have to exclude quantum corrections is not unexpected.  After

1305: all, the Boltzmann equation is a classical equation, so our derivation

1306: definitely has to break down at some level when adding quantum effects. As we

1307: have seen, this breakdown takes place immediately beyond the classical level:

1308: we can recover a standard Boltzmann collision term of the

1309: form~(\ref{Boltzmann}) only by the restriction to classical processes.

1310: %

1311: Now we can simplify the collision term to

1312: %

1313: \beqa

1314:     C(p)

1315:   &=&

1316:     \frac{1}{2}

1317:     \sum_{n=2}^\infty \frac{1}{n!}

1318:     \int \frac{d^4k_1}{(2\pi)^4} \ldots \frac{d^4k_n}{(2\pi)^4}

1319:     (2\pi)^4\delta^4(p+k_1+\ldots+k_n)

1320: %

1321: \label{Coll:type1}

1322: \\&&\hphantom{X}

1323: %

1324:     \times \big| {\cal M}^+_{(1)}(p, k_1, \ldots, k_n) \big|^2

1325: %

1326: \nonumber\\&&\hphantom{XXX}

1327: %

1328:   \times

1329:            \Big(

1330:                   i\Delta^>(k_1) \ldots i\Delta^>(k_n) i\Delta^>(p)

1331:                  -i\Delta^<(k_1) \ldots i\Delta^<(k_n) i\Delta^<(p)

1332:            \Big)

1333: \nonumber

1334: .

1335: \eeqa

1336: %

1337: The final step is to use the on-shell ansatz~(\ref{final:ansatz:Wigner}) for

1338: the propagators and to perform the integrals over the zero-components of the

1339: momenta. The ansatz for $i\Delta^<(k_i)$ contains two $\delta$-functions, one

1340: that corresponds to positive energies and one that corresponds to negative

1341: energies. In the case of a negative energy, we additionally invert the

1342: corresponding spatial momentum.

1343: %

1344: This way we obtain contributions with all possible combinations of particles

1345: $k_1$ to $k_n$ either going into or coming out of the scattering.

1346: %

1347: Since the scattering amplitude is totally symmetric with respect to the order

1348: of its arguments, all particles going in are exchangeable, and so are all the

1349: particles that come out of the scattering. Thus we can group those terms that

1350: have the same number of ingoing and outgoing particles. For $j$ outgoing

1351: particles (besides $p$), there are

1352: %

1353: \beq

1354:   N(j) = \frac{n!}{j!(n-j)!} = \binom{n}{j}

1355: \eeq

1356: %

1357: identical terms. The complete collision term is

1358: %

1359: \beqa

1360:   C(p) &=&

1361:    \frac 12 \frac{\pi}{\omega_p} \delta(p_0-\omega_p) \,

1362:    \sum_{n=2}^\infty \frac{1}{n!} \,

1363:    \int \frac{d^3k_1}{(2\pi)^32\omega_1}

1364:          \ldots

1365:          \frac{d^3k_n}{(2\pi)^32\omega_n} \,\,

1366:    \sum_{j=0}^n \binom{n}{j} 

1367: %

1368: \label{collision-term:final}

1369: \\ &&\hphantom{XX}

1370: %

1371:    \times (2\pi)^4\delta^4(p+k_1+\ldots+k_j-k_{j+1}-\ldots -k_n)

1372: %

1373: \nonumber\\ &&\hphantom{XXXX}

1374: %

1375:   \times

1376:   \big| {\cal M}^+_{(1)}(p, k_1, \ldots, k_j, -k_{j+1}, \ldots, -k_n)

1377:   \big|^2

1378: %

1379: \nonumber\\[.1ex]&&\hphantom{XXXXXX}

1380: %

1381:   \times \Big[

1382:               \big(1+f_1\big) \ldots \big(1+f_j\big)

1383:               f_{j+1}         \ldots f_n

1384:               \big(1+f_p\big)

1385: %

1386: \nonumber\\&&\hphantom{XXXXXXXXXXX}

1387: %

1388:       -f_1 \ldots f_j

1389:        \big(1+f_{j+1}\big) \ldots \big(1+f_n\big)

1390:        f_p

1391:   \Big]

1392: ,

1393: \nonumber

1394: \eeqa

1395: %

1396: where $f_i \equiv f(\vec{k}_i)$, $\omega_i=(\vec{k}_i^2+m^2)^{1/2}$, and all

1397: four momenta are on-shell: $k_i=(\omega_i, \vec{k}_i)$.

1398: %

1399: The collision term consists of two parts, according to the two terms in square

1400: brackets, referred to as gain and loss term, respectively. They describe the

1401: increase or decrease of the density of particles with momentum $p$ in the

1402: plasma due to the scattering.

1403: %

1404: Typically, several contributions vanish because of kinematical restrictions,

1405: for example the one with all particles going in.  How this works in detail

1406: depends on the type of the interaction and the particle masses, see also the

1407: example in the next section.

1408: %

1409: Since in the kinetic equation~(\ref{kinetic-equation:final}) we only integrate

1410: over positive $p_0$, terms proportional to the negative energy $p_0=-\omega_p$

1411: were dropped here.

1412: 

1413: 

1414: 

1415: 

1416: 

1417: 

1418: 

1419: 

1420: \subsection{Example: $\phi^3$-theory at the 2-loop level}

1421: \label{subsec:example}

1422: 

1423: As an example consider a real, massive scalar theory with a $\lambda\phi^3/3!$

1424: self-interaction. The relevant 2-loop self-energy diagrams are shown in

1425: Fig.~\ref{fig:phi3}.

1426: %

1427: The cuts in these diagrams comprise three lines, so the corresponding

1428: scattering processes in the Boltzmann collision term will have four external

1429: lines, and at tree level are shown in Fig.~\ref{fig:2to2}. The matrix element

1430: for these diagrams is

1431: %

1432: \beq

1433:     {\cal M}^+(p, k_1, k_2, k_3)

1434:   = \lambda^2 \,

1435:     \left( \,

1436:            \frac{1}{(p+k_2)^2} + \frac{1}{(p+k_3)^2} + \frac{1}{(p+k_1)^2} \,

1437:     \right)

1438: .

1439: \eeq

1440: %

1441: The collision term can then be read off from~(\ref{collision-term:final}). In

1442: principle there are 4 contributions, corresponding to 0, 1, 2 or 3 particles

1443: coming out together with $p$, but due to kinematic restrictions only 2-to-2

1444: scattering can occur:

1445: %

1446: \beqa

1447:   C(p) &=&

1448:    \frac{\pi}{4\omega_p} \delta(p_0-\omega_p) \,

1449:    \int \frac{d^3k_1}{(2\pi)^32\omega_1}

1450:         \frac{d^3k_2}{(2\pi)^32\omega_2}

1451:         \frac{d^3k_3}{(2\pi)^32\omega_3}

1452: %

1453: \\ &&\hphantom{XX}

1454: %

1455:    \times (2\pi)^4\delta^4(p+k_1-k_2-k_3) \,

1456:   \big|{\cal M}^+(p, k_1, -k_2, -k_3)\big|^2

1457: %

1458: \nonumber\\&&\hphantom{XXXXXX}

1459: %

1460:   \times

1461:   \Big[

1462:       (1+f_p) (1+f_1) f_2 f_3  -  f_p f_1 (1+f_2) (1+f_3)

1463:   \Big]

1464: .

1465: \nonumber

1466: \eeqa

1467: %

1468: One could have obtained this result directly from the self-energy computed with

1469: the CTP Feynman rules,

1470: %

1471: \beqa

1472:      -i\Pi^{+-}(p)

1473:  &=& \int \frac{d^4q}{(2\pi)^4} \frac{d^4k}{(2\pi)^4}

1474: \label{self-energy:2-loop}

1475: \\&&\hphantom{XX}

1476:      \Big[

1477:           \frac 12 i\Delta^{-+}(q)     i\Delta^{++}(p+q)

1478:                    i\Delta^{+-}(p+q+k) i\Delta^{-+}(k)

1479:                    i\Delta^{--}(p+q)

1480: \nonumber\\&&\hphantom{XXX}

1481:           +  i\Delta^{++}(p+q) i\Delta^{+-}(p+q+k)

1482:              i\Delta^{--}(q+k) i\Delta^{-+}(q)

1483:              i\Delta^{-+}(k)

1484:      \Big]

1485: ,

1486: \nonumber

1487: \eeqa

1488: %

1489: but already for these comparatively simple diagrams this is not trivial, and it

1490: becomes rather involved for self-energies with more

1491: loops~\cite{Carrington:2004tm}.

1492: %

1493: \begin{figure}[!tb]

1494: %

1495: \begin{center}

1496: %

1497: \begin{minipage}[t]{4in}

1498: %

1499: \psfrag{i1}[cc]{$\scriptstyle k_2$}

1500: \psfrag{i2}[cc]{$\scriptstyle k_3$}

1501: \psfrag{o1}[cc]{$\scriptstyle p$}

1502: \psfrag{o2}[cc]{$\scriptstyle k_1$}

1503: \psfrag{v1}[cc]{$\scriptstyle +$}

1504: \psfrag{v2}[cc]{$\scriptstyle +$}

1505: \includegraphics[width=1in]{Figures/2to2_track.eps}

1506: %

1507: \hspace*{0.4in}

1508: %

1509: \psfrag{i1}[cc]{$\scriptstyle k_3$}

1510: \psfrag{i2}[cc]{$\scriptstyle k_2$}

1511: \psfrag{o1}[cc]{$\scriptstyle p$}

1512: \psfrag{o2}[cc]{$\scriptstyle k_1$}

1513: \psfrag{v1}[cc]{$\scriptstyle +$}

1514: \psfrag{v2}[cc]{$\scriptstyle +$}

1515: \includegraphics[width=1in]{Figures/2to2_track.eps}

1516: %

1517: \hspace*{0.4in}

1518: %

1519: \psfrag{i1}[cc]{$\scriptstyle k_2$}

1520: \psfrag{i2}[cc]{$\scriptstyle k_3$}

1521: \psfrag{o1}[cc]{$\scriptstyle p$}

1522: \psfrag{o2}[cc]{$\scriptstyle k_1$}

1523: \psfrag{v1}[cc]{$\scriptstyle +$}

1524: \psfrag{v2}[cc]{$\scriptstyle +$}

1525: \includegraphics[width=1in]{Figures/2to2_annihil.eps}

1526: %

1527: \end{minipage}

1528: %

1529: \caption{\small

1530: The tree diagrams contributing to the scattering matrix for four external

1531: particles.}

1532: \label{fig:2to2}

1533: %

1534: \end{center}

1535: %

1536: \end{figure}

1537: %

1538: 

1539: 

1540: 

1541: 

1542: 

1543: 

1544: 

1545: 

1546: 

1547: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1548: %

1549: %

1550: %  Complex

1551: %

1552: %

1553: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1554: 

1555: \section{Complex scalar field}

1556: \label{sec:complex}

1557: 

1558: In the case of a complex scalar field, when there are particles and

1559: antiparticles, we can adopt the general line of argumentation from the real

1560: case, but we have to make adjustments at a few points.

1561: 

1562: First of all, charge conservation constrains the cut: the net charge flow

1563: across the cut must be equal to the net charge flow through the whole

1564: self-energy diagram. Thus the cut must comprise $2n+1$ lines, $n+1$ leading

1565: from the left to the right, and $n$ leading back from the right to the left.

1566: %

1567: Then the analog to the basic claim~(\ref{basic-claim}) for complex fields is

1568: %

1569: \beqa

1570:       -i\Pi^{+-}_n(p)

1571:   &=& \frac{1}{(n+1)!n!}

1572:       \int \frac{d^4k_1}{(2\pi)^4} \, \ldots \, \frac{d^4k_{n+1}}{(2\pi)^4}

1573:       \int \frac{d^4q_1}{(2\pi)^4} \, \ldots \, \frac{d^4q_n}{(2\pi)^4}

1574: %

1575: \label{basic-claim:complex}

1576: \\[1ex]

1577: %

1578:   &&\hphantom{X}

1579:   \times

1580:       (2\pi)^4 \delta^4(p + q_1+\ldots+q_n - k_1-\ldots-k_{n+1})

1581: \nonumber\\[1ex]

1582:   &&\hphantom{XX}

1583:   \times

1584:        {\cal M}^+_{(1)}(-k_1, \ldots , -k_{n+1} ; p, q_1, \ldots, q_n)

1585: \nonumber\\[1ex]

1586:   &&\hphantom{XXX}

1587:   \times

1588:        i\Delta^{+-}_0(k_1) \ldots i\Delta^{+-}_0(k_{n+1})

1589:        i\Delta^{-+}_0(q_1) \ldots i\Delta^{-+}_0(q_n)

1590: \nonumber\\[1ex]

1591:   &&\hphantom{XXXX}

1592:   \times

1593:        {\cal M}^-_{(1)}(-p, -q_1, \ldots, -q_n ; k_1, \ldots, k_{n+1})

1594: .

1595: \nonumber

1596: \eeqa

1597: %

1598: The proof runs almost exactly like in the real case, the only difference is

1599: that we have to distinguish the two sets of cut lines. Lines running from the

1600: left to the right cannot be interchanged with lines running from the right to

1601: the left, because they transport charge in different directions.  This leads to

1602: the factor $1/(n+1)!n!$ that represents one group of $n+1$ lines carrying

1603: charge to the right and another group of $n$ lines carrying charge to the left.

1604: 

1605: 

1606: The final result is similar to~(\ref{collision-term:final}), but both gain and

1607: loss term now consist of two parts. In one part the distribution functions

1608: represent a charge coming out of the reaction, where we have all combinations

1609: of these functions being either $(1+f_+)$ for a particle coming out, or $f_-$

1610: for an antiparticle going in. The other part contains distribution functions

1611: that represent a charge going into the reaction, that is all combinations with

1612: either $f_+$ for a particle going in or a $(1+f_-)$ for an antiparticle coming

1613: out.

1614: 

1615: It is straightforward to generalize to the case of several scalar particle

1616: species in a similar way.  Each species corresponds to a group of lines or

1617: distribution functions, respectively, where members of different groups cannot

1618: be interchanged with each other.

1619: 

1620: 

1621: 

1622: 

1623: 

1624: 

1625: 

1626: 

1627: 

1628: 

1629: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1630: %

1631: %

1632: %  Discussion

1633: %

1634: %

1635: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1636: 

1637: \section{Discussion}

1638: \label{sec:discussion}

1639: 

1640: 

1641: We start with the equation of motion for the out-of-equilibrium Green function

1642: for a real scalar field.

1643: %

1644: Using gradient expansion, on-shell approximation and small coupling expansion

1645: the left hand side leads to the flow term of a Boltzmann equation in a well

1646: known way.

1647: %

1648: We make a perturbative expansion of the self-energies on the right hand side

1649: and give a diagrammatic proof that

1650: %

1651: the relevant self-energies have a factorization property, and that

1652: %

1653: for a certain type of contributions to the self-energy the right hand side can

1654: be rewritten as the matrix element squared of scattering processes times the

1655: distribution functions of the particles involved in the reactions. This works

1656: for those self-energy contributions that correspond to tree level scattering

1657: diagrams. The scattering processes obtained include any number of external

1658: particles and thus correspond to self-energy diagrams with any number of loops.

1659: %

1660: 

1661: 

1662: 

1663: If effects beyond the classical level are required, it is in general not

1664: correct to simply compute the scattering probabilities including quantum

1665: corrections and plug them into a standard Boltzmann collision term.

1666: %

1667: In specific situations it is possible to fit quantum corrections into the

1668: picture, however at some cost.

1669: %

1670: In~\cite{Isert:2001kq} for example, a model of scalar quarks and gluons was

1671: examined and a standard Boltzmann collision term was constructed from the

1672: self-energy up to two loops.

1673: %

1674: In order to be able to do this, the scattering matrices were not computed

1675: simply by following the CTP rules, but for some diagrams Feynman propagators

1676: had to be used for some internal lines.

1677: %

1678: Furthermore, kinematical arguments had to be used to get rid of some diagrams

1679: that didn't match into the picture.

1680: %

1681: The approach presented there required an explicit study of each diagram and can

1682: not be generalized to arbitrary diagrams or theories.

1683: 

1684: 

1685: 

1686: 

1687: The simple structure~(\ref{Boltzmann}) cannot be obtained beyond the classical

1688: level for two reasons.

1689: %

1690: First, there are the self-energy diagrams that allow several complete

1691: cuts. This is precisely the type of diagrams that causes problems in the

1692: extension of the vacuum Cutkosky rules to finite

1693: temperature~\cite{Kobes,Gelis:1997zv}.

1694: %

1695: Second, there is the difference between the free propagators for the cut

1696: $k_i$-lines and the full propagator for the $p$-line.

1697: %

1698: Since we know that the scattering diagrams we obtain must not contain

1699: corrections on the $p$-lines but do have corrections on the $k_i$-lines, it

1700: seems tempting to collect these corrections and attribute them to the free

1701: propagators, thus promoting them to full ones.

1702: %

1703: In effect this means that we would have to do the perturbative analysis of the

1704: self-energy in terms of resummed and thus full propagators.  As a consequence,

1705: all diagrams contributing to the self-energy must not have corrections to

1706: already full lines.

1707: %

1708: Then the available self-energy diagrams are not sufficient to be rewritten as

1709: the square of a matrix element, however: certain diagrams are missing. For

1710: instance, the 2-loop diagram in Fig.~\ref{fig:phi3}(b) provides only the

1711: interference terms in~(\ref{a-times-b}), the squares of the individual

1712: amplitudes can only come from Fig.~\ref{fig:phi3}(a), which has a forbidden

1713: correction.

1714: 

1715: 

1716: 

1717: This shows that there is no consistent and systematic way to obtain a standard

1718: Boltzmann collision term beyond the tree level.  At the tree level, a standard

1719: Boltzmann collision term is found that includes scattering processes with any

1720: number of external particles.

1721: 

1722: 

1723: 

1724: 

1725: 

1726: 

1727: 

1728: 

1729: 

1730: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1731: %

1732: %

1733: %  Literature

1734: %

1735: %

1736: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1737: 

1738: \begin{thebibliography}{99}

1739: 

1740: 

1741: \bibitem{Kad62} L.P. Kadanoff and G. Baym, 

1742: {\it Quantum Statistical Mechanics} (Benjamin, New York, 1962).

1743: 

1744: 

1745: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1746: 

1747: 

1748: \bibitem{Li:1982gk}

1749:   S.~P.~Li and L.~D.~McLerran,

1750:   %``Toward A Transport Theory Of The Quark - Gluon Plasma In Ultrarelativistic

1751:   %Nuclear Collisions: An Example From Scalar Meson Theory,''

1752:   Nucl.\ Phys.\ B {\bf 214}, 417 (1983).

1753:   %%CITATION = NUPHA,B214,417;%%

1754: 

1755: \bibitem{Calzetta:1986cq}

1756:   E.~Calzetta and B.~L.~Hu,

1757:   %``Nonequilibrium Quantum Fields: Closed Time Path Effective Action, Wigner

1758:   %Function And Boltzmann Equation,''

1759:   Phys.\ Rev.\ D {\bf 37}, 2878 (1988).

1760:   %%CITATION = PHRVA,D37,2878;%%

1761: 

1762: \bibitem{Mrowczynski:1992hq}

1763:   S.~Mrowczynski and U.~W.~Heinz,

1764:   %``Towards a relativistic transport theory of nuclear matter,''

1765:   Annals Phys.\  {\bf 229}, 1 (1994).

1766:   %%CITATION = APNYA,229,1;%%

1767: 

1768: \bibitem{Klevansky:1997wm}

1769:   S.~P.~Klevansky, A.~Ogura and J.~Hufner,

1770:   %``Derivation of transport equations for a strongly interacting Lagrangian  in

1771:   %powers of h-bar and 1/N(c),''

1772:   Annals Phys.\  {\bf 261}, 37 (1997).

1773:   %[arXiv:hep-ph/9708263].

1774:   %%CITATION = HEP-PH 9708263;%%

1775: 

1776: \bibitem{PSW}

1777:   T.~Prokopec, M.~G.~Schmidt and S.~Weinstock,

1778:   %``Transport equations for chiral fermions to order h-bar and electroweak

1779:   %baryogenesis,''

1780:   Annals Phys.\  {\bf 314}, 208 (2004); {\bf 314}, 267 (2004).

1781:   %%CITATION = HEP-PH 0312110;%%

1782:   %%CITATION = HEP-PH 0406140;%%

1783: 

1784: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1785: 

1786: 

1787: \bibitem{Carrington:2004tm}

1788:   M.~E.~Carrington and S.~Mrowczynski,

1789:   %``Transport theory beyond binary collisions,''

1790:   Phys.\ Rev.\ D {\bf 71}, 065007 (2005).

1791:   %[arXiv:hep-ph/0406097].

1792:   %%CITATION = HEP-PH 0406097;%%

1793: 

1794: 

1795: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1796: 

1797: 

1798: \bibitem{Kobes}

1799:   R.~L.~Kobes and G.~W.~Semenoff,

1800:   %``Discontinuities Of Green Functions In Field Theory At Finite Temperature

1801:   %And Density,''

1802:   Nucl.\ Phys.\ B {\bf 260}, 714 (1985); {\bf 272}, 329 (1986).

1803:   %%CITATION = NUPHA,B260,714;%%

1804:   %%CITATION = NUPHA,B272,329;%%

1805: 

1806: \bibitem{Gelis:1997zv}

1807:   F.~Gelis,

1808:   %``Cutting rules in the real-time formalisms at finite temperature,''

1809:   Nucl.\ Phys.\ B {\bf 508}, 483 (1997).

1810:   %[arXiv:hep-ph/9701410].

1811:   %%CITATION = HEP-PH 9701410;%%

1812: 

1813: 

1814: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1815: 

1816: 

1817: \bibitem{Schwinger:1960qe}

1818:   J.~S.~Schwinger,

1819:   %``Brownian Motion Of A Quantum Oscillator,''

1820:   J.\ Math.\ Phys.\  {\bf 2}, 407 (1961).

1821:   %%CITATION = JMAPA,2,407;%%

1822: 

1823: \bibitem{Keldysh:1964ud}

1824:   L.~V.~Keldysh,

1825:   %``Diagram Technique For Nonequilibrium Processes,''

1826:   Zh.\ Eksp.\ Teor.\ Fiz.\  {\bf 47}, 1515 (1964)

1827:   [Sov.\ Phys.\ JETP {\bf 20}, 1018 (1965)].

1828:   %%CITATION = ZETFA,47,1515;%%

1829: 

1830: \bibitem{Chou:1984es}

1831:   K.~C.~Chou, Z.~B.~Su, B.~L.~Hao and L.~Yu,

1832:   %``Equilibrium And Nonequilibrium Formalisms Made Unified,''

1833:   Phys.\ Rept.\  {\bf 118}, 1 (1985).

1834:   %%CITATION = PRPLC,118,1;%%

1835: 

1836: 

1837: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1838: 

1839: 

1840: \bibitem{GroupTheory}

1841: This is a standard theorem of group theory, see e.g.

1842: W.~Ledermann and A.~J.~Weir,

1843: {\it Introduction to Group Theory} (Addison Wesley, 1997).

1844: 

1845: 

1846: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1847: 

1848: 

1849: \bibitem{Jeon}

1850:   S.~Jeon and P.~J.~Ellis,

1851:   %``Multiple scattering expansion of the self-energy at finite temperature,''

1852:   Phys.\ Rev.\ D {\bf 58}, 045013 (1998).

1853:   %[arXiv:hep-ph/9802246].

1854:   %%CITATION = HEP-PH 9802246;%%

1855: 

1856: 

1857: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1858: 

1859: 

1860: % the 2-loop massacre

1861: \bibitem{Isert:2001kq}

1862:   D.~S.~Isert and S.~P.~Klevansky,

1863:   %``Transport theory for a scalar quark \& gluon model,''

1864:   Eur.\ Phys.\ J.\ A {\bf 12}, 453 (2001).

1865:   %%CITATION = EPHJA,A12,453;%%

1866: 

1867: 

1868: 

1869: 

1870: \end{thebibliography}

1871: 

1872: \end{document}

1873: 

1874: 

1875: 

1876: