1: %\documentstyle[aps,pra,epsf]{revtex}

2: %\documentstyle[preprint,aps,pra]{revtex}

3: \documentstyle[aps,pra]{revtex}

4: %\tighten

5: \draft

6: \begin{document}

7: \tightenlines

8: 

9: \title{Landau-Khalatnikov two-fluid hydrodynamics

10:  of a trapped Bose gas}

11: 

12: \author{Tetsuro Nikuni and Allan Griffin}

13: \address{Department of Physics, University of Toronto,

14: Toronto, Ontario, Canada M5S 1A7} 

15: 

16: \date{\today}

17: \maketitle

18: 

19: \begin{abstract}

20: Starting from the quantum kinetic equation for the non-condensate atoms

21: and the generalized Gross-Pitaevskii equation for the condensate,

22: we derive the two-fluid hydrodynamic equations of a trapped Bose gas

23: at finite temperatures.

24: We follow the standard

25: Chapman-Enskog procedure, starting from a solution of the kinetic

26: equation corresponding to the complete local

27: equilibrium between the condensate and the non-condensate components.

28: Our hydrodynamic equations are shown to reduce to a form identical 

29: to the well-known Landau-Khalatnikov two-fluid equations,

30: with hydrodynamic damping due to the deviation from local equilibrium.

31: The deviation from local equilibrium  within the thermal cloud

32: gives rise to dissipation associated with shear viscosity and thermal conduction.

33: In addition, we show that effects due to the

34: deviation from the diffusive local equilibrium between the

35: condensate and the non-condensate (recently considered by Zaremba, Nikuni

36: and Griffin) can be described by four

37: frequency-dependent second viscosity transport coefficients.

38: We also derive explicit formulas for all the transport coefficients.

39: These results are used to introduce two new characteristic relaxation

40: times associated with hydrodynamic damping.

41: These relaxation times give the rate at which local equilibrium is reached

42: and hence determine whether one is in the two-fluid hydrodynamic region.

43: 

44: \end{abstract}

45: \pacs{PACS numbers: 03.75.Fi, 05.30Jp, 67.40.Db }

46: 

47: \section{introduction}

48: \label{sec:intro}

49: At very low temperatures, the dynamics of a trapped Bose gas is described by

50: the time-dependent Gross-Pitaevskii (GP) equation for the macroscopic

51: wavefunction of the condensate. 

52: As discussed in several

53: recent reviews \cite{RMP,varenna}, there is excellent agreement

54: (within a few percent) between experimental observations of 

55: collective modes for $T\lesssim 0.4T_{\rm BEC}$ 

56: and theoretical calculations based on the $T=0$ GP equation.

57: At elevated temperatures where the condensate is appreciably depleted by

58: thermal excitations, one must consider the coupled motion of the

59: condensate and non-condensate degrees of freedom.

60: In a recent paper, Zaremba, Nikuni and Griffin \cite{ZNG} derived a generalized Gross-Pitaevskii

61: equation for the condensate atoms and a quantum kinetic equation for

62: the non-condensate atoms, which can be used to discuss the coupled

63: dynamics of the two components at finite temperatures. 

64: These two components are coupled through mean-field interactions as well

65: as collisions between the atoms.

66: 

67: Two limiting cases for

68: the dynamics of the gas correspond to the collisionless and hydrodynamic

69: regimes~\cite{varenna,NP}.  

70: Up to the present, most experiments on the collective modes of Bose

71: gases are thought to be in the low-density collisionless limit.

72: In this regime, the main 

73: effect of the non-condensate (thermal cloud) component 

74: is to damp the condensate oscillations.

75: In addition to Landau and Beliaev damping

76: due to the dynamic mean-field interaction between two components,

77: there is damping arising from the lack of diffusive local equilibrium between

78: the condensate and the non-condensate \cite{griffin}.

79: The latter mechanism of damping has recently been worked out in detail by

80: Williams and Griffin \cite{WG}.

81: 

82: In contrast, in the regime where

83: collisions between atoms are rapid enough to establish a state of 

84: dynamic local equilibrium in the non-condensate gas,

85: the dynamics of the system is described by hydrodynamic equations for

86: a few local variables.

87: This regime contains much new physics and it should be accessible taking

88: advantage of the larger densities now available as well as the large 

89: scattering cross sections close to a Feshbach resonance~\cite{85Rb}.

90: We have recently given a detailed derivation and discussion of two-fluid

91: hydrodynamic equations for trapped atomic gases at finite

92: temperatures~\cite{ZNG,griffin,ZGN,NZG,CJP},

93: starting from a generalized GP equation coupled with a kinetic equation.

94: These equations were derived under the assumption that

95: the non-condensate atoms are in local thermodynamic equilibrium 

96: among themselves but are {\it not}

97: in diffusive equilibrium with the condensate atoms.

98: The resulting ZGN$'$ hydrodynamic equations~\cite{NZG,ZNG} involve a

99: new characteristic relaxation time $\tau_{\mu}$, which is the time scale on

100: which local diffusive equilibrium is established. 

101: This equilibration

102: process leads to a novel damping mechanism which is associated with

103: the collisional exchange of atoms between the two components.

104: This ZGN$'$ hydrodynamics is briefly reviewed in Section \ref{zgn'}.

105: In Ref.~\cite{CJP}, we generalized the ZGN$'$ equations to include the effect of

106: deviations from local equilibrium, and worked out hydrodynamic damping

107: associated with the collisions among the non-condensate atoms.

108: At finite temperatures of interest, this deviation from local equilibrium

109: within the thermal cloud gives rise to damping associated with

110: thermal conductivity and the shear viscosity.

111: Such a generalization was first discussed in Section V of Ref.~\cite{NG} starting from

112: the ZGN hydrodynamic equations.

113: We also note that the thermal conductivity and shear viscosity were first derived for a

114: {\it uniform} Bose-condensed gas at finite temperatures in pioneering papers by Kirkpatrick

115: and Dorfman~\cite{KD}.

116: 

117: In the present paper, building on our recent work with Zaremba, we present

118: a more complete derivation of two-fluid hydrodynamic equations,

119: including dissipation.

120: In Section \ref{sec:CE},

121: we solve the kinetic equation by expanding 

122: the distribution function $f({\bf r},{\bf p},t)$ around $f^{(0)}({\bf r},{\bf p},t)$,

123: the latter describing

124: complete local equilibrium between the condensate and the non-condensate.

125: We follow the standard Chapman-Enskog method used to derive hydrodynamic damping

126: in the kinetic theory of classical gases.

127: In this treatment, the lowest order hydrodynamic equations involve no

128: dissipative terms.

129: All hydrodynamic damping effects are included by taking into account deviations

130: from the local equilibrium distribution $f^{(0)}$.

131: In Section \ref{sec:LK}, we prove that,

132: with appropriate definitions of various thermodynamic variables, 

133: our two-fluid hydrodynamic equations with damping have precisely the structure of

134: those first derived by Landau and Khalatnikov~\cite{Landau,Khal}. 

135: In particular, the damping associated with

136: the collisional exchange of atoms between the two components, which has been discussed

137: at length in our previous work \cite{ZNG,NZG,CJP}, is now expressed in terms of

138: {\it frequency-dependent} second viscosity coefficients.

139: This type of damping is in addition to the usual kind of hydrodynamic damping associated

140: with shear viscosity and thermal conductivity \cite{CJP,NG}.

141: 

142: In Section \ref{sec:transport}, we also derive explicit expressions for

143: all the transport coefficients that

144: appear in the dissipative terms in our two-fluid hydrodynamic equations. 

145: For the purpose of illustration, we evaluate the temperature dependence of all

146: the transport coefficients for the case of a {\it uniform} Bose gas.

147: The case of a trapped Bose gas is quite different.

148: Because the condensate density is always much larger than the non-condensate

149: density in the central regions of a trapped Bose gas, we find that the collisions

150: between the condensate and non-condensate atoms are the dominant contribution

151: to all transport coefficients.

152: 

153: The lengthy analysis in Sections \ref{sec:CE}, \ref{sec:LK} and \ref{sec:transport}

154: is, of necessity, very complex and conceptionally quite subtle.

155: Most readers will only be interested in the final conclusions, which we now

156: summarize.

157: We prove that the two-fluid hydrodynamic equations of a trapped Bose gas can be written

158: precisely in the well-known Landau-Khalatnikov form, as summarized in

159: (\ref{landau_eqs}).

160: The new feature which arises in a trapped Bose gas (as compared to 

161: superfluid $^4$He) is that, as noted above, the four second viscosity

162: coefficients can be frequency-dependent, as given by (\ref{zeta_omega}).

163: Finally, explicit expressions for all the transport coefficients are given in

164: (\ref{kappa_final}), (\ref{etafinal}) and (\ref{zeta_final}).

165: An important final result of our analysis (see Section \ref{sec:transport}) is a

166: precise definition of three characteristic relaxation times associated with the

167: various transport coefficients.

168: Moreover we show that the collisions between the condensate and non-condensate

169: atoms always play the dominant role in the hydrodynamic damping of trapped Bose gases.

170: 

171: The present paper brings to a natural conclusion our series of papers (with Zaremba)

172: on the two-fluid hydrodynamics of trapped Bose gases \cite{ZNG,NZG,CJP,NG}.

173: Much remains to be done solving these hydrodynamic equations and experimentally

174: checking the predictions.

175: 

176: \section{ZGN$'$ hydrodynamic equations}

177: \label{zgn'}

178: In this section, we derive the most general form of hydrodynamic equations

179: for the condensate and non-condensate at finite temperatures,

180: and then briefly review the ZGN$'$ two-fluid hydrodynamics.

181: We start with the underlying coupled dynamical equations for the non-condensate

182: and the condensate, as derived in Ref.~\cite{ZNG} and recently

183: reviewed in Ref.~\cite{griffin}.

184: The non-condensate atoms are described by the distribution function

185: $f({\bf r},{\bf p},t)$, which obeys the quantum kinetic equation

186: \begin{eqnarray}

187: {\partial f({\bf r},{\bf p},t) \over \partial t} + {{\bf p} \over m} 

188: \cdot \bbox{\nabla}_{{\bf r}} f({\bf r},{\bf p},t) &-& \bbox{\nabla}_{{\bf r}}

189: U \cdot \bbox{\nabla}_{{\bf p}} f({\bf r} ,{\bf p},t) \cr

190: &=& C_{12}[f,\Phi] + C_{22}[f].

191: \label{eq1}

192: \end{eqnarray}

193: Here the effective potential

194: $U({\bf r},t)\equiv U_{\rm ext}({\bf r})+2g[n_c({\bf r},t)+\tilde 

195: n({\bf r},t)]$ includes the self-consistent Hartree-Fock (HF) mean field, and as usual,

196: we treat the inter-atomic interaction in the $s$-wave approximation with

197: $g=4\pi \hbar^2 a/m$. 

198: The condensate density is $n_c({\bf r},t)\equiv|\Phi({\bf r},t)|^2$ 

199: and the non-condensate density $\tilde n({\bf r},t)$ is given by

200: \begin{equation}

201: \tilde n({\bf r},t)=\int\frac{d{\bf p}}{(2\pi\hbar)^3}f({\bf r},{\bf p},t).

202: \label{eq2}

203: \end{equation}

204: The two collision terms in (\ref{eq1}) are given by

205: \begin{eqnarray}

206: &&C_{22}[f] \equiv \frac{2 g^2}{(2\pi)^5\hbar^7}

207:  \int{d{\bf p}_2}\int{d{\bf p}_3}

208: \int d{\bf p}_4 \cr

209: &&\times\delta ({\bf p}+{\bf p}_2 -{\bf p}_3 -{\bf p}_4)

210: \delta(\tilde\varepsilon_{p}+\tilde\varepsilon_{p_2}

211: -\tilde\varepsilon_{p_3}-\tilde\varepsilon_{p_4}) \cr

212: &&\times\left[(1+f)(1+f_2)f_3f_4-ff_2(1+f_3)(1+f_4)\right]\, ,

213: \label{eq3}

214: \end{eqnarray}

215: \begin{eqnarray}

216: &&C_{12}[f,\Phi]\equiv \frac{2 g^2 n_c}{(2\pi)^2\hbar^4} \int d{\bf p}_1

217: \int d{\bf p}_2 \int d{\bf p}_3 \cr

218: &&\times  \delta(m{\bf v}_c+{\bf p}_1-{\bf p}_2-{\bf p}_3)

219: \delta(\varepsilon_c+\tilde\varepsilon_{p_1}

220: -\tilde\varepsilon_{p_2}-\tilde\varepsilon_{p_3}) \cr

221: &&\times  [\delta({\bf p}-{\bf p}_1)-\delta({\bf p}-{\bf p}_2)

222: -\delta({\bf p}-{\bf p}_3)] \cr

223: &&\times [(1+f_1)f_2f_3-f_1(1+f_2)(1+f_3)],

224: \label{eq4}

225: \end{eqnarray}

226: with $f \equiv f({\bf r, p}, t),\, f_i\equiv f({\bf r, p}_i, t)$.

227: The expression in (\ref{eq4}) takes into account the fact that a

228: condensate atom locally has energy

229: $\varepsilon_c({\bf r},t)=\mu_c({\bf r},t)+\frac{1}{2}mv_c^2({\bf r},t)$

230:  and momentum $m{\bf v}_c$, where the condensate chemical potential 

231: $\mu_c$ and velocity ${\bf v}_c$ will be defined shortly.

232: On the other hand, in our finite-temperature model,

233: a non-condensate atom locally has the HF energy

234: $\tilde\varepsilon_p({\bf r},t)=\frac{p^2}{2m}+U({\bf r},t)$.

235: This particle-like dispersion relation limits our entire analysis 

236: to finite temperatures.

237: 

238: The equation of motion for the condensate

239: is given by a generalized 

240: Gross-Pitaevskii equation \cite{ZNG} for the macroscopic wavefunction 

241: $\Phi({\bf r},t)$ (see also Ref.~\cite{stoof})

242: \begin{equation}

243: i\hbar \frac{\partial\Phi({\bf r},t)}{\partial t}=

244: \left[-\frac{\hbar^2\nabla^2}{2m}+U_{\rm ext}({\bf r})

245: +gn_c({\bf r},t)+2g\tilde n({\bf r},t)-i R({\bf r},t)\right]\Phi({\bf r},t),

246: \end{equation}

247: where

248: \begin{equation}

249: R({\bf r},t)=\frac{\hbar\Gamma_{12}({\bf r},t)}{2n_c({\bf r},t)},

250: \end{equation}

251: with

252: \begin{equation}

253: \Gamma_{12}\equiv\int\frac{d{\bf p}}{(2\pi \hbar )^3}C_{12}[f({\bf r},{\bf p},t),\Phi({\bf r},t)].

254: \label{def_G12}

255: \end{equation}

256: The dissipative term $R$ in (5) is associated with the exchange of atoms

257: between the condensate and non-condensate, as described by the collision

258: integral $C_{12}[f,\Phi]$ in (4).

259: We see that (1) and (5) must be solved self-consistently.

260: It is customary to rewrite the GP equation (5) in terms of the amplitude and phase of

261: $\Phi({\bf r},t)=\sqrt{n_c({\bf r},t)}e^{i\theta({\bf r},t)}$, which leads to

262: (${\bf v}_c\equiv\hbar \bbox{\nabla}\theta({\bf r},t)/m$)

263: \begin{mathletters}

264: \begin{eqnarray}

265: {\partial n_c \over \partial t} + \bbox{\nabla}\cdot(n_c{\bf v}_c)&=& 

266: -\Gamma_{12}[f,\Phi]\,,  

267: \label{eq_nc}

268: \\

269: m\left({\partial\over\partial t}+{\bf v}_c\cdot 

270: \bbox{\nabla}\right) {\bf v}_c&=&-\bbox{\nabla}\mu_c \ ,

271: \label{eq_vc}

272: \end{eqnarray}

273: \label{hydro-C}

274: \end{mathletters}

275: 

276: \noindent

277: where the condensate chemical potential is given by

278: \begin{equation}

279: \mu_c({\bf r}, t) =-\frac{\hbar^2\nabla^2\sqrt{n_c({\bf r},t)} }{2m\sqrt{n_c({\bf r},t)}}

280: +U_{\rm ext}({\bf r})+

281: gn_c({\bf r}, t)+2g\tilde{n}({\bf r}, t)\, .

282: \label{mu_c}

283: \end{equation}

284: One sees that $\Gamma_{12}$ in (8) plays the role of a ``source function" in the 

285: continuity equation for the condensate, arising from the fact that $C_{12}$ collisions

286: do not conserve the number of condensate atoms \cite{ZNG}.

287: 

288: We note that the set of equations (\ref{eq1})-(\ref{def_G12}) has also been derived

289: using the elegant Kadanoff-Baym formulation \cite{milena1,milena2,KB}.

290: More recently, this KB derivation has been extended to cover low temperatures

291: as well \cite{milena3}, by working with a Bogoliubov quasiparticle

292: spectrum instead of the simpler HF spectrum used in the present paper.

293: One could use this extension as the basis for generalizing the present paper.

294: 

295: Following the standard procedure in the classical kinetic theory of gases \cite{huang}, 

296: we take moments of the kinetic equation (\ref{eq1}) with respect to $1,{\bf p}$ and $p^2$

297: to derive the most general form of ``hydrodynamic equations" for the

298: non-condensate. These moment equations take the form ($\mu$ and $\nu$ 

299: are Cartesian components):

300: \begin{mathletters}

301: \begin{eqnarray}

302: &&{\partial{\tilde n}\over\partial t}+\bbox{\nabla}\cdot 

303: (\tilde{n}{\bf v}_n) = \Gamma_{12}[f]\,, \\

304: &&m{\tilde n}\left({\partial\over\partial t}+{\bf v}_n\cdot 

305: \bbox{\nabla}\right) v_{n\mu}=-{\partial P_{\mu\nu}\over\partial x_\nu}

306: -{\tilde n}{\partial U\over\partial x_\mu}

307: -m(v_{n\mu}-v_{c\mu})\Gamma_{12}[f]\,, 

308: \\

309: &&{\partial\tilde\epsilon\over\partial t} +

310: \nabla\cdot(\tilde\epsilon{\bf v}_n) = -\bbox{\nabla}\cdot{\bf Q}

311: -D_{\mu\nu} P_{\mu\nu}+ \left[\frac{1}{2}m({\bf v}_n-{\bf v}_c)^2

312: +\mu_c-U\right]\Gamma_{12}[f]. 

313: \end{eqnarray}

314: \label{hydro_general}

315: \end{mathletters}

316: 

317: \noindent

318: The non-condensate density was defined earlier in (\ref{eq2}), 

319: while the non-condensate local velocity is defined by

320: \begin{equation}

321: {\tilde n}({\bf r},t){\bf v}_n({\bf r}, t)\equiv\int{d{\bf p}

322: \over(2\pi\hbar)^3} {{\bf p}\over m} f({\bf r, p}, t)\,.

323: \label{vn}

324: \end{equation}

325: In addition, we have introduced the following quantities:

326: \begin{mathletters}

327: \begin{eqnarray}

328: P_{\mu\nu}({\bf r}, t)&\equiv& m

329: \int{d{\bf p}\over(2\pi\hbar)^3}\left({p_\mu\over m} - v_{n\mu}\right)

330: \left({p_\nu\over m} - v_{n\nu}\right)

331: f({\bf r, p}, t), \label{eq35a} \\

332: {\bf Q}({\bf r}, t)&\equiv& \int{d{\bf p}\over(2\pi\hbar)^3} {1\over 2m} 

333: ({\bf p}-m{\bf v}_n)^2\left({{\bf p}\over m}-{\bf v}_n\right)

334: f({\bf r, p}, t),\label{eq35b}  \\

335: \tilde \epsilon({\bf r}, t) &\equiv&\int{d{\bf p}\over(2\pi\hbar)^3}

336: {1\over 2m} ({\bf p}-m{\bf v}_n)^2

337: f({\bf r},  {\bf p}, t) \,. 

338: \end{eqnarray}

339: \label{eq35}

340: \end{mathletters}

341: Finally, the symmetric rate-of-strain tensor appearing in (10)

342: is defined as

343: \begin{equation}

344: D_{\mu\nu}({\bf r}, t) \equiv {1 \over 2} \left({\partial v_{n \mu}

345: \over \partial x_\nu} + {\partial v_{n \nu} \over \partial x_\mu}

346: \right).

347: \label{eq36}

348: \end{equation}

349: 

350: The hydrodynamic equations in (\ref{hydro_general}) are formally exact,

351: but obviously are not closed.

352: To proceed, we must specify the conditions under which the dynamics of

353: the system to be determined.

354: In the ZGN$'$ hydrodynamics, we

355: consider the situation in which the $C_{22}$ collisions are

356: sufficiently rapid to establish local equilibrium among the non-condensate atoms.

357: This situation is described by the local equilibrium Bose distribution

358: for the thermal cloud, 

359: \begin{equation}

360: \tilde f({\bf r},{\bf p},t)=

361: \frac{1}{e^{\beta[\frac{1}{2m}({\bf p}-m{\bf v}_n)^2+U-\tilde \mu]}-1}

362: \, .

363: \label{eq8}

364: \end{equation}

365: Here, the temperature parameter $\beta$, normal fluid velocity 

366: ${\bf v}_n$, chemical potential $\tilde \mu$, and mean field $U$ are all

367: functions of ${\bf r}$ and $t$. 

368: One may immediately verify that $\tilde f$ has precisely the form such that

369: it satisfies the condition $C_{22}[\tilde f]=0$,

370: independent of the value of $\tilde \mu$.  In contrast, 

371: one finds that $C_{12}[\tilde f]$ does not vanish in general, namely

372: \begin{eqnarray}

373: C_{12}[\tilde f]&=&\frac{2g^2 n_c}{(2\pi)^2\hbar^4} 

374: [1 - e^{-\beta(\tilde \mu-\frac{1}{2}m({\bf v}_n-{\bf v}_c)^2-\mu_c)}] 

375: \cr && \times \int d{\bf p}_1

376: \int d{\bf p}_2

377: \int d{\bf p}_3

378: \delta(m{\bf v}_c+{\bf p}_1-{\bf p}_2-{\bf p}_3)

379: \delta(\tilde\varepsilon_1+\varepsilon_c

380: -\tilde\varepsilon_2-\tilde\varepsilon_3) \cr

381: && \times [\delta({\bf p}-{\bf p}_1)-\delta({\bf p}-{\bf p}_2)

382: -\delta({\bf p}-{\bf p}_3)]

383: (1+\tilde f_1)\tilde f_2 \tilde f_3.

384: \label{eq40}

385: \end{eqnarray}

386: Using the local distribution function (14) 

387: to evaluate the moments in (2) and (12), we find that 

388: the heat current ${\bf Q}({\bf r},t) = 0$, and that

389: \begin{equation}

390: \tilde n({\bf r},t) = \int{d{\bf p}\over (2\pi\hbar)^3}

391: \tilde f({\bf r, p}, t)

392: = {1\over\Lambda^3} g_{3/2}(z)\,,

393: \label{eq46}

394: \end{equation}

395: \begin{equation}

396: P_{\mu\nu} ({\bf r}, t) = \delta_{\mu\nu}{\tilde P}({\bf r}, t)

397: \equiv\delta_{\mu\nu}\int{d{\bf p}\over (2\pi\hbar)^3} 

398: \frac{({\bf p}^m{\bf v}_n)^2}{3m}

399: \tilde f({\bf r, p}, t)

400: =\delta_{\mu\nu}\frac{1}{\beta\Lambda^3}g_{5/2}(z).

401: \label{eq41}

402: \end{equation}

403: Here $z({\bf r},t)\equiv e^{\beta[\tilde\mu({\bf r},t)-U({\bf r},t)]}$ is the local

404: fugacity, $\Lambda({\bf r},t) \equiv[{2\pi\hbar^2/ mk_B T({\bf r},t)}]^{1/2}$

405: is the local thermal de Broglie wavelength and 

406: $g_n(z)=\sum_{l=1}^{\infty}z^l/l^n$ are the Bose-Einstein functions.

407: The kinetic energy density  is given by

408: $\tilde \epsilon({\bf r},t) ={3\over 2} \tilde P ({\bf r},t)$.

409: 

410: To summarize, using $f \simeq \tilde f$, we obtain the ZGN$'$ 

411: lowest-order hydrodynamic equations for the non-condensate given in 

412: Refs.~\cite{ZNG,griffin,NZG}.

413: In the linearized version of these ZGN$'$ hydrodynamic equations,

414: the condensate is described by

415: \begin{mathletters}

416: \begin{eqnarray}

417: &&\frac{\partial \delta n_c}{\partial t}+\bbox{\nabla}\cdot 

418: (n_{c0}\delta{\bf v_c}) = -\delta \Gamma_{12} \,, \\

419: && m\frac{\partial\delta{\bf v}_c}{\partial t}=-\bbox{\nabla}\delta\mu_c,

420: \end{eqnarray}

421: \label{lin_C}

422: \end{mathletters}

423: and the non-condensate variables satisfy

424: \begin{mathletters}

425: \begin{eqnarray}

426: &&{\partial{\delta \tilde n}\over\partial t}+\bbox{\nabla}\cdot 

427: (\tilde{n}_0\delta{\bf v}_n) = \delta\Gamma_{12}\,, \\

428: &&m{\tilde n_0}\frac{\partial \delta{\bf v}_n}{\partial t}

429: =-\bbox{\nabla} \delta\tilde P

430: -\delta{\tilde n}\bbox{\nabla} U_0-2g\tilde n_0\bbox{\nabla}(\delta \tilde n

431: +\delta n_c) \,, \\

432: &&{\partial\delta\tilde P\over\partial t} =

433: -\frac{5}{3}\nabla\cdot(\tilde P_0\delta{\bf v}_n) +{2\over 3}

434: {\bf v}_n \cdot \bbox{\nabla}\tilde P_0 

435: - {2\over 3}gn_{c0}\delta\Gamma_{12}.

436: \end{eqnarray}

437: \label{linZGN'}

438: \end{mathletters}

439: 

440: \noindent

441: The fluctuation of the condensate chemical potential is given by 

442: \begin{equation}

443: \delta \mu_c=g\delta n_c+2g\delta \tilde n.

444: \end{equation}

445: This assumes the Thomas-Fermi approximation, which means that the

446: first term in (\ref{mu_c}), the quantum pressure term, is neglected.

447: The source function $\delta\Gamma_{12}$ can be usefully expressed \cite{ZNG}

448: in terms of the difference between the condensate and non-condensate chemical potentials,

449: namely

450: \begin{equation}

451: \delta \Gamma_{12}=-\frac{\beta_0n_{c0}}{\tau_{12}}(\delta\tilde\mu

452: -\delta\mu_c)\equiv-\frac{\beta_0n_{c0}}{\tau_{12}}\delta\mu_{\rm diff},

453: \label{G12}

454: \end{equation}

455: where $\mu_{\rm diff}\equiv \tilde\mu-\mu_c$ and 

456: $\tau_{12}$ is the mean collision time \cite{ZNG,NZG} associated with $C_{12}$:

457: \begin{eqnarray}

458: \frac{1}{\tau_{12}}

459: &\equiv&

460:  \frac{2 g^2}{(2\pi)^5\hbar^7} \int d{\bf p}_1

461: \int d{\bf p}_2 \int d{\bf p}_3 \cr

462: &&\times  \delta({\bf p}_1-{\bf p}_2-{\bf p}_3)

463: \delta(\mu_{c0}+\tilde\varepsilon_1

464: -\tilde\varepsilon_2-\tilde\varepsilon_3) \cr

465: &&\times (1+f_{10})f_{20}f_{30},

466: \label{tau12}

467: \end{eqnarray}

468: where $f_{i0}=[e^{\beta_0(\tilde\varepsilon_i-\mu_{c0})}-1]^{-1}$ is the static

469: equilibrium distribution function and $\tilde\varepsilon_i=\frac{p^2}{2m}+U_0$.

470: 

471: Since one can show \cite{ZNG,CJP} that $\delta\mu_{\rm diff}$

472: can be written as a linear combination of $\delta\tilde n$ and $\delta \tilde P$, 

473: the above coupled hydrodynamic equations in (\ref{lin_C}) and  (\ref{linZGN'})

474: are a closed set for the variables

475: $\delta n_c,\delta{\bf v}_c,\delta\tilde n,\delta{\bf v}_n$ and

476: $\delta\tilde P$.

477: However, it will be useful later to have an equation of motion for 

478: $\delta\mu_{\rm diff}$. 

479: This is given by [see Eq.~(86) of Ref.~\cite{ZNG}]

480: \begin{equation}

481: \frac{\partial\delta\mu_{\rm diff}}{\partial t}

482: =-g\bbox{\nabla}\cdot[n_{c0}(\delta{\bf v}_c-\delta{\bf v}_n)]

483: -\frac{1}{3}gn_{c0}\bbox{\nabla}\cdot\delta{\bf v}_n

484: -\frac{\delta\mu_{\rm diff}}{\tau_{\mu}}.

485: \label{eq_mudiff}

486: \end{equation}

487: Here $\tau_{\mu}$ is a new relaxation time governing how

488: $\delta\mu_{\rm diff}$ relaxes to zero, i.e., how fast diffusive equilibrium

489: is reached between the condensate and non-condensate components.

490: It is related to the collision time $\tau_{12}$ in (\ref{tau12}) by the expression

491: \begin{equation}

492: \frac{1}{\tau_{\mu}({\bf r})}=\frac{\beta g n_{c0}}{\sigma_H\tau_{12}},

493: \label{taumu}

494: \end{equation}

495: where the dimensionless hydrodynamic renormalization factor

496: $\sigma_H$ is given by

497: \begin{equation}

498: \sigma_H({\bf r})=\left[\frac{\frac{5}{2}\tilde P_0+2g\tilde n_0n_{c0}

499: +\frac{2}{3}g^2\tilde\gamma_0n_{c0}^2}{\frac{5}{2}\tilde P_0\tilde\gamma_0

500: -\frac{3}{2}g\tilde n_0^2} -1\right]^{-1},

501: \label{sigma_H}

502: \end{equation}

503: where $\tilde\gamma\equiv \frac{g}{k_{\rm B}T\Lambda^3}g_{1/2}(z)$.

504: 

505: We can now look for normal mode solutions of the linearized ZGN$'$

506: equations in (\ref{lin_C}) and (\ref{linZGN'}).

507: Assuming that these fluctuations have a

508: time dependence $e^{-i\omega t}$, one can solve 

509: (\ref{eq_mudiff}) for $\delta\mu_{\rm diff}$ to give

510: \begin{equation}

511: \delta\mu_{\rm diff}=-\frac{\tau_{\mu}}{1-i\omega\tau_{\mu}}

512: \left\{

513: g\bbox{\nabla}\cdot[n_{c0}(\delta{\bf v}_c-\delta{\bf v}_n)]

514: +\frac{1}{3}gn_{c0}\bbox{\nabla}\cdot\delta{\bf v}_n\right\}.

515: \label{eq_mudiff2}

516: \end{equation}

517: In the limit $\omega\tau_{\mu}\to 0$, one sees that $\delta\mu_{\rm diff}\to 0$.

518: This situation corresponds to the complete local equilibrium between the

519: condensate and non-condensate components, with $\tilde\mu({\bf r},t)=\mu_c({\bf r},t)$.

520: In this limit, one can prove that our ZGN$'$ hydrodynamics

521: reduces to the Landau two-fluid hydrodynamics without dissipation terms,

522: as discussed in detail in Ref.~\cite{ZNG}.

523: It is clear that fluctuation of $\delta\mu_{\rm diff}$ described by

524: (\ref{eq_mudiff}) [or equivalently (\ref{eq_mudiff2})] gives rise to

525: a new relaxational mode in addition to usual collective oscillations

526: of the condensate and non-condensate (for a uniform superfluid, these

527: are the first and second sound modes).

528: For a uniform Bose gas, the frequency of this new mode is given by

529: $\omega_{\rm R}=-i/\tau_{\mu}$ \cite{NZG}.

530: Thus, in general, our ZGN$'$ equations predict the existence of a new

531: relaxational mode, associated with the equilibration of the condensate

532: and non-condensate collective variables.

533: 

534: In Ref.~\cite{CJP}, we have extended the theory to include small deviations

535: from the local equilibrium distribution $\tilde f$ in (\ref{eq8}).

536: These give rise to new dissipative terms in the non-condensate

537: equations associated with the shear viscosity ($\eta$) and the thermal conductivity

538: ($\kappa$) of the thermal cloud.

539: The damping of first sound, second sound and 

540: the relaxational mode due to the effect of normal fluid transport coefficients

541: was calculated in Ref.~\cite{CJP}.

542: In particular, we showed there that the relaxational mode was strongly coupled 

543: to (and renormalized by) fluctuations in the local temperature and hence the

544: thermal conductivity.

545: 

546: In summary, the ZGN$'$ hydrodynamics exhibit the physics of the coupled dynamics of the

547: condensate and non-condensate atoms in a clear fashion.

548: However the approach used in the ZGN$'$ theory has a disadvantage in that it is not

549: based on a small expansion parameter, in contrast to the more systematic Chapman-Enskog

550: procedure used here.

551: In Ref.~\cite{CJP}, we only included the effect of $C_{22}$ collisions

552: to discussing the deviation from local equilibrium.

553: This neglect of $C_{12}$ collisions in this connection is only justified when the

554: condensate density is very small compared to the non-condensate density

555: (since the $C_{12}$ term in (\ref{eq4}) is proportional to $n_c$).

556: However in a trapped gas, the $C_{12}$ collision integral is always significant

557: since the condensate is strongly peaked at the trap center, with a density much larger

558: than the non-condensate even at temperatures close to $T_{\rm BEC}$.

559: Thus, we must treat both $C_{12}$ and $C_{22}$ 

560: when considering deviations from local equilibrium.

561: 

562: In the following section, we present a more systematic derivation

563: of the two-fluid hydrodynamics, by following the standard Chapman-Enskog

564: procedure.

565: This derivation is similar to the work by Kirkpatrick

566: and Dorfman \cite{KD} for a {\it uniform} Bose gas.

567: As we discuss in Section \ref{sec:LK}, this new approach allows us to

568: show that the extended ZGN$'$ theory can be written in a form completely equivalent to

569: the Landau-Khalatnikov two-fluid hydrodynamics \cite{Khal} when we include hydrodynamic

570: damping.

571: This set of equations involves the thermal conductivity, shear viscosity and four

572: frequency-dependent second viscosity coefficients.

573: The latter are shown to arise from the fact that the condensate and non-condensate

574: are not in diffusive equilibrium ($\mu_c\neq\tilde\mu$), as discussed by

575: ZGN$'$ \cite{ZNG,NZG}.

576: 

577: \section{Chapman-Enskog expansion for a Bose-condensed gas}

578: \label{sec:CE}

579: \subsection{Lowest-order hydrodynamic equations}

580: 

581: Following the standard procedure of the Chapman-Enskog expansion

582: \cite{ferziger}, we introduce a small expansion parameter $\alpha$ and rewrite

583: the kinetic equation (\ref{eq1}) as

584: \begin{eqnarray}

585: {\partial f({\bf r},{\bf p},t) \over \partial t} + {{\bf p} \over m} 

586: \cdot \bbox{\nabla}_{{\bf r}} f({\bf r},{\bf p},t) &-& \bbox{\nabla}_{{\bf r}} U \cdot

587: \bbox{\nabla}_{{\bf p}} f({\bf r} ,{\bf p},t) \cr

588: &=& \frac{1}{\alpha}\left(C_{12}[f] + C_{22}[f]\right).

589: \label{kineq2}

590: \end{eqnarray}

591: This expansion parameter $\alpha$ will be eventually taken to be 1,

592: but allows one to develop a perturbative solution of (\ref{kineq2}).

593: In order to solve the quantum kinetic equation, 

594: we formally expand the distribution function $f({\bf r},{\bf p},t)$ in powers of $\alpha$:

595: \begin{equation}

596: f=f^{(0)}+\alpha f^{(1)}+ \cdots.

597: \label{f_alpha}

598: \end{equation}

599: Using this expansion (\ref{f_alpha}), we can also expand the various hydrodynamic variables

600: in (\ref{hydro_general})

601: \begin{eqnarray}

602: \tilde n&=&\tilde n^{(0)}+\alpha\tilde n^{(1)}+\cdots, ~~

603: P_{\mu\nu}=P_{\mu\nu}^{(0)}+\alpha P_{\mu\nu}^{(1)}+\cdots, ~~

604: {\bf Q}={\bf Q}^{(0)}+\alpha{\bf Q}^{(1)}+\cdots, \cr

605: \tilde\epsilon&=&\tilde\epsilon^{(0)}+\alpha\tilde\epsilon^{(1)}+\cdots.

606: \label{expand}

607: \end{eqnarray}

608: The superscript $(0)$ denotes the local equilibrium solution (see below) which is

609: determined by the collision integrals (formally when $\alpha\to 0$).

610: We also redefine the source function $\Gamma_{12}$ in (\ref{def_G12}) as

611: \begin{equation}

612: \Gamma_{12}\equiv\frac{1}{\alpha}\int\frac{d{\bf p}}{(2\pi\hbar)^3}

613: C_{12}[f,\Phi]=\frac{1}{\alpha}

614: (\Gamma_{12}^{(0)}+\alpha\Gamma_{12}^{(1)}+\alpha^2\Gamma_{12}\cdots).

615: \label{redef_G12}

616: \end{equation}

617: We also have an expansion for the condensate wavefunction

618: \begin{equation}

619: \Phi=\Phi^{(0)}+\alpha \Phi^{(1)}+\cdots.

620: \end{equation}

621: 

622: In this expansion, however, we assume that the total local density $n(\equiv n_c+\tilde n)$

623: is not altered by the higher order correction terms $f^{(i)}~(i\geq 1)$ in (\ref{f_alpha}).

624: That is, we have

625: \begin{equation}

626: n_c=n_c^{(0)}+\alpha n_c^{(1)}+\cdots,

627: \end{equation}

628: but

629: \begin{equation}

630: n=n_c^{(0)}+\tilde n^{(0)}, ~n_c^{(i)}+\tilde n^{(i)}=0 ~(i\geq 1).

631: \end{equation}

632: We also assume that non-local correction terms $f^{(i)}$ make

633: no contribution to the non-condensate velocity fields ${\bf v}_n$ or to the phase

634: $\theta$ of the condensate wavefunction (and hence to the condensate velocity ${\bf v}_c$).

635: Finally, the condensate chemical potential in (\ref{mu_c})

636: (we work within the Thomas-Fermi approximation) is given in the expansion

637: \begin{equation}

638: \mu_c({\bf r},t)=U_{\rm ext}({\bf r})+g[n({\bf r},t)+\tilde n({\bf r},t)]

639: =\mu_c^{(0)}({\bf r},t)+\alpha\mu_c^{(1)}({\bf r},t)+\cdots,

640: \end{equation}

641: with 

642: \begin{equation}

643: \mu_c^{(0)}\equiv U_{\rm ext}+g(n+\tilde n^{(0)}),~\mu_c^{(1)}=g\tilde n^{(1)}.

644: \label{muc01}

645: \end{equation}

646: 

647: Using the expansion (\ref{f_alpha}) in the kinetic equation (\ref{kineq2}), 

648: we find that the lowest order solution $f^{(0)}$ is determined from

649: \begin{equation}

650: C_{12}[f^{(0)},\Phi^{(0)}]+C_{22}[f^{(0)}]=0.

651: \label{eqforf0}

652: \end{equation}

653: The unique solution of (\ref{eqforf0}) is given by the ``diffusive local equilibrium''

654: Bose distribution function, namely

655: \begin{equation}

656: f^{(0)}({\bf r},{\bf p},t)=

657: \frac{1}{e^{\beta({\bf r},t)

658: [\frac{1}{2m}({\bf p}-m{\bf v}_n({\bf r},t))^2+U({\bf r},t)-

659: \tilde \mu^{(0)}({\bf r},t)]}-1}

660: \, .

661: \label{f0}

662: \end{equation}

663: Here the local equilibrium non-condensate chemical potential

664: $\tilde\mu^{(0)}$ is given by the condition that $C_{12}[f^{(0)},\Phi^{(0)}]=0$,

665: which gives

666: \begin{equation}

667: \tilde \mu^{(0)}=\mu_c^{(0)}+\frac{m}{2}({\bf v}_n-{\bf v}_c)^2. 

668: \label{muc0}

669: \end{equation}

670: Using (\ref{muc01}), this is equivalent to

671: \begin{equation}

672: \tilde \mu^{(0)}=\mu_c^{(0)}+\frac{m}{2}({\bf v}_n-{\bf v}_c)^2 

673: =U_{\rm ext}+gn+g\tilde n^{(0)}+\frac{m}{2}({\bf v}_n-{\bf v}_c)^2 ,

674: \label{muc02}

675: \end{equation}

676: in conjunction with

677: \begin{equation}

678: \tilde n^{(0)}({\bf r},t) = \int{d{\bf p}\over (2\pi\hbar)^3}

679: f^{(0)}({\bf r, p}, t)

680: = {1\over\Lambda^3} g_{3/2}(z^{(0)})\ .

681: \label{ntilde0}

682: \end{equation}

683: Here $z^{(0)}({\bf r},t)\equiv e^{\beta[\tilde\mu^{(0)}({\bf r},t)-U({\bf r},t)]}$ is the local

684: fugacity in diffusive local equilibrium.

685: 

686: It is important to appreciate that diffusive equilibrium

687: is not defined by the distribution function $f^{(0)}$ alone, but is determined

688: self-consistently with the non-condensate chemical potential as given by

689: (\ref{muc0}).

690: One may immediately verify that $f^{(0)}$ satisfies $C_{22}[f^{(0)}]=0$,

691: independent of the value of $\tilde \mu^{(0)}$.  

692: In contrast, $C_{12}[f^{(0)},\Phi^{(0)}]=0$

693: {\it only} if the local chemical potential of the thermal cloud is given by (\ref{muc0})

694: and the condensate and non-condensate densities are determined self-consistently.

695: Of course, it immediately follows that since $C_{12}[f^{(0)},\Phi^{(0)}]=0$,

696: we have $\Gamma_{12}^{(0)}=\Gamma_{12}[f^{(0)},\Phi^{(0)}]=0$ and hence

697: (\ref{redef_G12}) reduces to 

698: \begin{equation}

699: \Gamma_{12}=\Gamma_{12}^{(1)}+\alpha \Gamma_{12}^{(2)}

700: +\cdots.

701: \label{G12_alpha}

702: \end{equation}

703: 

704: 

705: Using the local distribution function (\ref{f0}) 

706: to evaluate the moments in (\ref{eq35b}), we find that 

707: the heat current ${\bf Q}^{(0)}({\bf r},t) = 0$, and 

708: \begin{equation}

709: P_{\mu\nu}^{(0)} ({\bf r}, t) = \delta_{\mu\nu}{\tilde P}^{(0)}({\bf r}, t)

710: \equiv\delta_{\mu\nu}\int{d{\bf p}\over (2\pi)^3} {({\bf p}-m{\bf v}_n)^2\over 3m}

711: f^{(0)}({\bf r, p}, t)

712: =\delta_{\mu\nu}\frac{1}{\beta\Lambda^3}g_{5/2}(z^{(0)}).

713: \label{ptilde0}

714: \end{equation}

715: The local kinetic energy density  is given by

716: $\tilde \epsilon^{(0)}({\bf r},t) ={3\over 2} \tilde P^{(0)} ({\bf r},t)$.

717: 

718: To summarize, the lowest-order hydrodynamic equations for the non-condensate are

719: given by

720: \begin{mathletters}

721: \begin{eqnarray}

722: &&{\partial{\tilde n}\over\partial t}+\bbox{\nabla}\cdot 

723: (\tilde n{\bf v}_n) = \Gamma_{12}^{(1)}

724: \label{eq_ntilde0}

725: \,, \\

726: &&m{\tilde n}\left({\partial\over\partial t}+{\bf v}_n\cdot 

727: \bbox{\nabla}\right) {\bf v}_n=-\bbox{\nabla} \tilde P

728: -{\tilde n}\bbox{\nabla} U

729: -m({\bf v}_n-{\bf v}_c)\Gamma_{12}^{(1)}\,, 

730: \label{eq_vn0}\\

731: &&{\partial\tilde P\over\partial t} +

732: \nabla\cdot(\tilde P{\bf v}_n) = -{2\over 3}\tilde P \bbox{\nabla}

733: \cdot{\bf v}_n + {2\over 3} \left[\frac{1}{2}m({\bf v}_n-{\bf v}_c)^2

734: +\mu_c-U\right]\Gamma_{12}^{(1)},

735: \label{eq_ptilde0}

736: \end{eqnarray}

737: \label{lowest-eqs}

738: \end{mathletters}

739: 

740: \noindent

741: where $\tilde n=\tilde n^{(0)}$, $\tilde P=\tilde P^{(0)}$ and $\mu_c=\mu_c^{(0)}$

742: are given by (\ref{muc0}), (\ref{ntilde0}) and (\ref{ptilde0}).

743: It should be noted that the above equations involve the source term $\Gamma_{12}^{(1)}$.

744: Even though $C_{12}[f^{(0)}]=0$,

745: one sees from (\ref{G12_alpha}) that the lowest order contribution

746: is in fact given by $\Gamma_{12}^{(1)}$, which

747: involves the contribution from the next order correction $f^{(1)}$.

748: Later we will derive an explicit expression for $\Gamma_{12}$ when we include the 

749: effect of deviations from the local equilibrium distribution and transport

750: processes.

751: Here we only display the result for the lowest order contribution which enter

752: into (\ref{lowest-eqs}) [see also (\ref{eq_G12})]:

753: \begin{equation}

754: \Gamma_{12}^{(1)}({\bf r},t)=\sigma_H\left\{

755: \bbox{\nabla}\cdot[n_c({\bf v}_c-{\bf v}_n)]

756: +\frac{1}{3}n_c\bbox{\nabla}\cdot{\bf v}_n\right\},

757: \label{gamma1}

758: \end{equation}

759: where $\sigma_H$ is defined by (\ref{sigma_H}).

760: 

761: It is important to note that even though $\Gamma_{12}^{(1)}$ involves an integral

762: over the collision integral $C_{12}$ (see (\ref{redef_G12})), the expression in (\ref{gamma1})

763: does not involve any collision time.

764: The expression for $\Gamma_{12}^{(1)}$ in (\ref{gamma1}) is consistent with

765: the ZGN$'$ result for $\delta\mu_{\rm diff}$ given in (\ref{eq_mudiff2}) 

766: in the limit $\omega\tau_{\mu}\to 0$ [using (\ref{G12}) and (\ref{sigma_H})].

767: We recall that in this limit, one has $\delta\mu_{\rm diff}\to 0$ and

768: thus $\tilde f$ in (\ref{eq8}) reduces to $f^{(0)}$ in (\ref{f0}).

769: Therefore the hydrodynamic equations given in (\ref{lowest-eqs}) are 

770: equivalent, in the $\omega\tau_{\mu}\to 0$ limit,

771: to those given by the ZGN$'$ theory.

772: As noted in Ref.~\cite{ZNG},

773: these coupled lowest-order hydrodynamic equations

774: in (\ref{lowest-eqs}) and (\ref{hydro-C}) can be combined 

775: and also shown to be precisely equivalent to the Landau two-fluid equations

776: without dissipation due to the transport processes.

777: In Section \ref{sec:LK}, we prove this equivalence in the more general case

778: when dissipation is included.

779: 

780: \subsection{Two-fluid equations with hydrodynamic dissipation}

781: 

782: We next consider the deviation (\ref{f_alpha}) from the local equilibrium

783: distribution function $f^{(0)}$ to first order in the Chapman-Enskog expansion.

784: This deviation $f^{(0)}$ gives rise to additional dissipative

785: terms in the hydrodynamic equations.

786: As usual, in determining the dissipative terms, we restrict ourselves to

787: terms of first order in the velocity fields ${\bf v}_n$ and ${\bf v}_c$.

788: Following Refs.~\cite{KD,NG,CJP}, we write the first correction term in

789: (\ref{f_alpha}) in the form

790: \begin{equation}

791: f^{(1)}=f^{(0)}({\bf r},{\bf p},t) [1+f^{(0)}({\bf r},{\bf p},t)]

792: \psi({\bf r},{\bf p},t),

793: \label{correction}

794: \end{equation}

795: and work with $\psi({\bf r},{\bf p},t)$.

796: To first order in $\alpha$, the $C_{22}$ and $C_{12}$ collision terms in (\ref{eq2})

797: reduce to ($f=f^{(0)}+\alpha f^{(1)}$).

798: \begin{eqnarray}

799: \frac{1}{\alpha}

800: C_{22}[f]&\simeq&C_{22}[f^{(1)}]\simeq \frac{2g^2}{(2\pi)^5\hbar^7}\int d{\bf p}_2

801: \int d{\bf p}_3 \int d{\bf p}_4 \cr

802: &&\times \delta({\bf p}+{\bf p}_2-{\bf p}_3-{\bf p}_4)\delta(\tilde\varepsilon_{p_1}

803: +\tilde\varepsilon_{p_2}-\tilde\varepsilon_{p_3}-\tilde\varepsilon_{p_4}) \cr

804: && \times f^{(0)} f^{(0)}_2\left(1+f^{(0)}_3\right)

805: \left(1+f^{(0)}_4\right)

806: (\psi_3+\psi_4-\psi_2-\psi) \cr

807: && \equiv \hat L_{22}[\psi].

808: \label{L22}

809: \end{eqnarray}

810: \begin{eqnarray}

811: \frac{1}{\alpha}C_{12}[f,\Phi]&\simeq&

812:  \frac{2 g^2 n_c}{(2\pi)^2\hbar^4} \int d{\bf p}_1

813: \int d{\bf p}_2 \int d{\bf p}_3 \cr

814: &&\times  \delta(m{\bf v}_c+{\bf p}_1-{\bf p}_2-{\bf p}_3)

815: \delta(\varepsilon_c^{(0)}+\tilde\varepsilon_{p_1}

816: -\tilde\varepsilon_{p_2}-\tilde\varepsilon_{p_3}) \cr

817: &&\times  [\delta({\bf p}-{\bf p}_1)-\delta({\bf p}-{\bf p}_2)

818: -\delta({\bf p}-{\bf p}_3)] \cr

819: &&\times (1+f^{(0)}_1)f^{(0)}_2f^{(0)}_3

820: (-\beta\mu_c^{(1)}+\psi_2+\psi_3-\psi_1) \cr 

821: && \equiv -\beta g\tilde n^{(1)}\hat L_{12}[1]+\hat L_{12}[\psi],

822: \label{C12_1}

823: \end{eqnarray}

824: where $\varepsilon_c^{(0)}=\mu_c^{(0)}+\frac{1}{2}mv_c^2$ and $\mu_c^{(0)}$

825: is given by (\ref{muc01}).

826: The linearized $\hat L_{12}$ operator is defined by

827: \begin{eqnarray}

828: \hat L_{12}[\psi]

829: &\equiv&

830:  \frac{2 g^2 n_c}{(2\pi)^2\hbar^4} \int d{\bf p}_1

831: \int d{\bf p}_2 \int d{\bf p}_3 \cr

832: &&\times  \delta(m{\bf v}_c+{\bf p}_1-{\bf p}_2-{\bf p}_3)

833: \delta(\varepsilon_c^{(0)}+\tilde\varepsilon_{p_1}

834: -\tilde\varepsilon_{p_2}-\tilde\varepsilon_{p_3}) \cr

835: &&\times  [\delta({\bf p}-{\bf p}_1)-\delta({\bf p}-{\bf p}_2)

836: -\delta({\bf p}-{\bf p}_3)] \cr

837: &&\times (1+f^{(0)}_1)f^{(0)}_2f^{(0)}_3

838: (\psi_2+\psi_3-\psi_1).

839: \label{L12}

840: \end{eqnarray}

841: 

842: Using (\ref{L22})-(\ref{L12}) and expanding the kinetic equation 

843: (\ref{kineq2}) to first order in $\alpha$, we find that 

844: the first non-local correction $f^{(1)}$ is determined by the equation

845: \begin{eqnarray}

846: &&{\partial^0 f^{(0)}({\bf r},{\bf p},t) \over \partial t} + {{\bf p} \over m} 

847: \cdot \bbox{\nabla}_{{\bf r}} f^{(0)}({\bf r},{\bf p},t) - \bbox{\nabla}_{{\bf r}} U \cdot

848: \bbox{\nabla}_{{\bf p}} f^{(0)}({\bf r} ,{\bf p},t) \cr

849: &&= -\beta g\tilde n^{(1)}\hat L_{12}[1]+\hat L_{12}[\psi] + \hat L_{22}[\psi].

850: \label{eq_f1}

851: \end{eqnarray}

852: Here $\partial^0/\partial t$ means that we use the lowest order

853: hydrodynamic equations given by (\ref{lowest-eqs})

854: in evaluating time derivatives of 

855: ${\bf v}_n$, $\tilde\mu$, $T$ and $U$. 

856: The resulting linearized equation which determines the function $\psi$ is 

857: (for details, see Appendix A)

858: \begin{eqnarray}

859: &&\left\{\frac{{\bf u}\cdot\bbox{\nabla}T}{T}\left[\frac{mu^2}{2k_{\rm B}T}-

860: \frac{5g_{5/2}(z)}{2g_{3/2}(z)}\right]+\frac{m}{k_{\rm B}T}D_{\mu\nu}

861: \left(u_{\mu}u_{\nu}-\frac{1}{3}\delta_{\mu\nu}u^2\right) \right. \cr

862: &&

863: +\left.\left(\sigma_2+\frac{mu^2}{3k_{\rm B}T}\sigma_1

864:  \right)\frac{\Gamma_{12}^{(1)}}{\tilde n^{(0)}}\right\}

865: f^{(0)}(1+ f^{(0)})+\beta g\tilde n^{(1)}\hat L_{12}[1]

866: =\hat L_{12}[\psi]+\hat L_{22}[\psi]\equiv \hat L[\psi],

867: \label{linpsi}

868: \end{eqnarray}

869: where the thermal velocity ${\bf u}$ is defined by 

870: $m{\bf u}\equiv {\bf p}-m{\bf v}_n$ and $z=z^{(0)}$.

871: The dimensionless thermodynamic functions $\sigma_1$, $\sigma_2$ in (\ref{linpsi}) are 

872: defined by

873: \begin{eqnarray}

874: \sigma_1({\bf r},t)&\equiv&\frac{\gamma^{(0)}\tilde n^{(0)}

875: \left[\tilde \mu^{(0)}-U\right]-\frac{3}{2}[\tilde n^{(0)}]^2}

876: {\frac{5}{2}\tilde P^{(0)}\gamma^{(0)}-\frac{3}{2}[\tilde n^{(0)}]^2}, \cr

877: \sigma_2({\bf r},t)&\equiv&\beta\frac{\frac{5}{2}\tilde P^{(0)}\tilde n^{(0)}

878: -[\tilde n^{(0)}]^2 \left[\tilde \mu^{(0)}-U\right]}

879: {\frac{5}{2}\tilde P^{(0)}\gamma^{(0)}-\frac{3}{2}[\tilde n^{(0)}]^2},

880: \label{sigma12}

881: \end{eqnarray}

882: where $\gamma^{(0)}({\bf r},t)\equiv \frac{\beta}{\Lambda^3}g_{1/2}(z^{(0)}({\bf r},t))

883: =\tilde\gamma^{(0)}/g$.

884: We note that $C_{12}$ enters in three separate places in (\ref{linpsi}).

885: 

886: The linearized collision operators $\hat L_{12}$ and $\hat L_{22}$

887: satisfy the conditions

888: \begin{eqnarray}

889: &&\hat L_{12}[{\bf p}-m{\bf v}_c]=0, 

890: ~~\hat L_{12}[\tilde\varepsilon_p-\varepsilon_c^{(0)}]=0, \cr

891: &&\hat L_{22}[1]=0, ~~

892: \hat L_{22}[{\bf p}]=0, ~~\hat L_{22}[\tilde\varepsilon_p]=0.

893: \end{eqnarray}

894: In order to have a unique solution of (\ref{linpsi}) for $\psi$, we impose

895: the following additional constraints:

896: \begin{mathletters}

897: \begin{eqnarray}

898: &&\int d{\bf p} ~{\bf u} \ f^{(0)}(1+f^{(0)})\psi=0,  \label{const_vn}\\

899: &&\int d{\bf p}\left(\frac{m}{2}u^2+U-\tilde\mu^{(0)}\right)

900: f^{(0)}(1+f^{(0)})\psi \cr

901: &&~~~~=\frac{1}{\beta}\int d{\bf p} \ln (1+f^{(0)^{-1} })f^{(1)}=0.

902: \label{const_s}

903: \end{eqnarray}

904: \label{const}

905: \end{mathletters}

906: \noindent

907: 

908: \noindent

909: Physically, the constraint (\ref{const_vn}) means that the deviation from local

910: equilibrium make no contribution to the local velocity field ${\bf v}_n$

911: defined in (\ref{vn}).

912: As we discuss in more detail in Section \ref{sec:LK}, the constraint (\ref{const_s}) means

913: that the total energy density and the local entropy density are not altered by

914: the deviation $f^{(1)}$.

915: They have the the same value as given by $f^{(0)}$.

916: 

917: Since (\ref{linpsi}) is a linear integral equation for $\psi$, one may write the most general

918: solution in the following form \cite{ferziger}:

919: \begin{equation}

920: \psi({\bf r},{\bf p},t)=\frac{\bbox{\nabla}T\cdot{\bf u}}{T}A(u)+D_{\mu\nu}

921: \left(u_{\mu}u_{\nu}-\frac{1}{3}u^2\delta_{\mu\nu}\right)B(u)

922: +\Gamma_{12}^{(1)}D(u),

923: \label{solution}

924: \end{equation}

925: where the dependence on $({\bf r},t)$ is left implicit and $u_{\mu}$ is a

926: component of the thermal velocity.

927: Here the functions $A(u),B(u)$ and $D(u)$ are given by the solution to the

928: following three linearized integral equations:

929: \begin{mathletters}

930: \begin{eqnarray}

931: &&{\bf u}\left[\frac{mu^2}{2k_{\rm B}T}-

932: \frac{5g_{5/2}(z)}{2g_{3/2}(z)}\right]f^{(0)}(1+\tilde f^{(0)})=

933: \hat L[{\bf u}A(u)], \label{eqforA}\\

934: &&

935: \frac{m}{k_{\rm B}T}

936: \left(u_{\mu}u_{\nu}-\frac{1}{3}\delta_{\mu\nu}u^2\right)

937: f^{(0)}(1+f^{(0)})

938: =\hat L[\left(u_{\mu}u_{\nu}-\frac{1}{3}\delta_{\mu\nu}u^2\right)B(u)], 

939: \label{eqforB} \\

940: &&\left(\sigma_2+\frac{mu^2}{3k_{\rm B}T}\sigma_1\right)\frac{1}{\tilde n^{(0)}}f^{(0)}(1+f^{(0)})

941: +\frac{\beta g\tilde n^{(1)}}{\Gamma_{12}^{(1)}} \hat L_{12}[1]=\hat L[D(u)].

942: \label{eqforD}

943: \end{eqnarray}

944: \label{eqforABD}

945: \end{mathletters}

946: For the constraints (\ref{const}) to be satisfied, we also need to require that

947: \begin{mathletters}

948: \begin{eqnarray}

949: &&\int \frac{d{\bf p}}{(2\pi\hbar)^2} f^{(0)}(1+f^{(0)})u^2 A(u)=0, 

950: \label{constA}

951: \\

952: &&\int \frac{d{\bf p}}{(2\pi\hbar)^2} f^{(0)}(1+f^{(0)})

953: \left(\frac{mu^2}{2}+U-\tilde \mu^{(0)}\right) D(u)=0.

954: \label{constD}

955: \end{eqnarray}

956: \end{mathletters}

957: 

958: Using the solution for $\psi$ given in (\ref{solution}),

959: one finds that the corrections due to $f^{(1)}$ in (\ref{correction})

960: to the various hydrodynamic variables are given by

961: \begin{eqnarray}

962: \tilde n^{(1)}&=&\int \frac{d{\bf p}}{(2\pi\hbar)^3}

963: f^{(0)}(1+f^{(0)})D(u)\Gamma_{12}^{(1)}({\bf r},t)\equiv -\tau\Gamma_{12}^{(1)},  

964: \label{n1} \\

965: P^{(1)}_{\mu\nu}&=&\delta_{\mu\nu}\tilde P^{(1)}

966: -2\eta\left[D_{\mu\nu}-\frac{1}{3}

967: {\rm Tr}D \delta_{\mu\nu}\right], \\

968: {\bf Q}^{(1)}&=&-\kappa\bbox{\nabla}T,

969: \end{eqnarray}

970: with

971: \begin{eqnarray}

972: \tilde P^{(1)}&=&\tau\frac{2}{3}(U-\tilde \mu^{(0)})\Gamma_{12}^{(1)}

973: \simeq \tau\frac{2}{3}gn_c^{(0)}\Gamma_{12}^{(1)}, \\

974: \tilde\epsilon^{(1)}&=&\frac{3}{2}\tilde P^{(1)}.

975: \end{eqnarray}

976: We note that $\tilde n$ and $\tilde P$ are both altered by an amount proportional

977: to $\Gamma_{12}^{(1)}$.

978: The transport coefficients $\eta$ and $\kappa$ are associated with the functions

979: $A(u)$ and $B(u)$,

980: \begin{mathletters}

981: \begin{eqnarray}

982: \eta&=&-\frac{m}{15}\int\frac{d{\bf p}}{(2\pi\hbar)^3}

983: u^4B(u)f^{(0)}(1+f^{(0)}), 

984: \label{eta}

985: \\

986: \kappa&=&-\frac{m}{6T}\int\frac{d{\bf p}}{(2\pi\hbar)^3}

987: u^4A(u)f^{(0)}(1+f^{(0)}).

988: \label{kappa}

989: \end{eqnarray}

990: \label{eta-kappa}

991: \end{mathletters}

992: 

993: The relaxation time $\tau$ defined in (\ref{n1}), namely

994: \begin{equation}

995: \tau=-\int\frac{d{\bf p}}{(2\pi\hbar)^3}f^{(0)}(1+f^{(0)})D(u),

996: \label{tau}

997: \end{equation}

998: plays a crucial role in the subsequent analysis.

999: Using (\ref{n1}) in (\ref{eqforD}), one can rewrite the integral equation for $D(u)$

1000: in the form

1001: \begin{equation}

1002: \left(\sigma_2+\frac{mu^2}{3k_{\rm B}T}\sigma_1\right)\frac{1}{\tilde n^{(0)}}f^{(0)}(1+f^{(0)})

1003: -\tau\beta g \hat L_{12}[1]=\hat L[D(u)].

1004: \end{equation}

1005: In Section \ref{sec:transport}, we solve the three linearized equations listed in

1006: (\ref{eqforABD}).

1007: It will be shown there that the solution for the function $D(u)$ is

1008: \begin{equation}

1009: D(u)=-\frac{\tau_{\mu}}{\tilde n^{(0)}}\left(\sigma_2+\frac{mu^2}{3k_{\rm B}T}

1010: \sigma_1\right).

1011: \label{Du}

1012: \end{equation}

1013: Using this, one finds that $\tau$ can be identified with the relaxation time $\tau_{\mu}$

1014: defined in (\ref{taumu}).

1015: In the present discussion, the physical meaning of the relaxation time $\tau_{\mu}$

1016: can be clearly seen by writing the source function $\Gamma_{12}^{(1)}$ 

1017: in the form (see (\ref{n1}) and (\ref{expand})) 

1018: \begin{equation} 

1019: \Gamma_{12}^{(1)}=-\frac{\tilde n^{(1)}}{\tau}=

1020: -\frac{\tilde n-\tilde n^{(0)}}{\tau_{\mu}}.

1021: \end{equation}

1022: This kind of relaxation term in the two-fluid hydrodynamic equations such as

1023: (\ref{eq_nc}) and (\ref{eq_ntilde0})

1024: was also discussed in a pioneering paper by Miyake and Yamada~\cite{MY}

1025: in discussing the liquid $^4$He near the superfluid transition

1026: (where a phenomenological relaxation time equivalent to $\tau_{\mu}$ was introduced).

1027: 

1028: In summary, we have obtained the following hydrodynamic equations for 

1029: the non-condensate including the normal fluid transport coefficients

1030: (we now set the expansion parameter $\alpha=1$):

1031: \begin{mathletters}

1032: \label{eqhydro}

1033: \begin{eqnarray}&&\frac{\partial \tilde n}{\partial t}

1034: +\bbox{\nabla}\cdot(\tilde n{\bf v}_n)=\Gamma_{12}

1035: \label{eq_tilden}\\

1036: &&m\tilde n\left(\frac{\partial}{\partial t}+{\bf v}_n\cdot \bbox{\nabla}\right)v_{n\mu}

1037: +\frac{\partial \tilde P}{\partial x_{\mu}}+\tilde n

1038: \frac{\partial U}{\partial

1039: x_{\mu}}=-m(v_{n\mu}-v_{c\mu})\Gamma_{12} \cr

1040: && \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ 

1041: +\frac{\partial}{\partial x_{\nu}}\left\{2\eta\left[D_{\mu\nu}-\frac{1}{3}

1042: ({\rm Tr}D)\delta_{\mu\nu} \right]\right\},

1043: \label{eq_vn}\\

1044: &&\frac{\partial \tilde \epsilon}{\partial t}+\bbox{\nabla}

1045: \cdot(\tilde\epsilon{\bf v}_n)

1046: +(\bbox{\nabla}\cdot{\bf v}_n)\tilde P=\left[\frac{1}{2}m({\bf v}_n-{\bf v}_c)^2

1047: +\mu_c-U\right]\Gamma_{12} \cr

1048: && \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ 

1049: +\bbox{\nabla}\cdot(\kappa\bbox{\nabla} T)

1050: +2\eta\left[ D_{\mu\nu}-\frac{1}{3}({\rm Tr}D)\delta_{\mu\nu}\right]^2,

1051: \label{eq26c}

1052: \end{eqnarray}

1053: \label{hydro-n}

1054: \end{mathletters}

1055: where $\tilde n$ and $\tilde P$ are given by

1056: \begin{eqnarray}

1057: &&\tilde n=\tilde n^{(0)}-\tau_{\mu}\Gamma_{12}^{(1)},  \label{ntildetot}\\

1058: &&\tilde P=\tilde P^{(0)}+\tau_{\mu}\frac{2}{3}gn_c^{(0)}\Gamma_{12}^{(1)}, \label{ptildetot}

1059: \end{eqnarray}

1060: and $\tilde\epsilon=\frac{3}{2}\tilde P$.

1061: Here $\tilde n^{(0)}$ and $\tilde P^{(0)}$ are given by (\ref{ntilde0}) and (\ref{ptilde0}),

1062: respectively.

1063: The equivalent ``quantum"

1064: hydrodynamic equations for the condensate are given in (\ref{hydro-C}),

1065: where the condensate chemical potential is given by

1066: \begin{equation}

1067: \mu_c=\mu_c^{(0)}-g\tau_{\mu}\Gamma_{12}^{(1)}.

1068: \label{muctot}

1069: \end{equation}

1070: 

1071: We now derive an equation for the function $\Gamma_{12}$.

1072: Using (\ref{ntildetot}) and (\ref{eq_tilden}), one obtains

1073: \begin{equation}

1074: \frac{\partial \tilde n}{\partial t}=

1075: \frac{\partial \tilde n^{(0)}}{\partial t}-\tau_{\mu}\frac{\partial \Gamma_{12}}{\partial t}

1076: =-\bbox{\nabla}\cdot(\tilde n{\bf v}_n)+\Gamma_{12}.

1077: \label{eq_totn}

1078: \end{equation}

1079: Using the explicit expression for $\tilde n^{(0)}$ given in (\ref{ntilde0}),

1080: one obtains

1081: \begin{equation}

1082: \frac{\partial \tilde n^{(0)}}{\partial t}

1083: =\left(\frac{3}{2}\tilde n+g \gamma n_c\right)\frac{1}{T}\frac{\partial T}{\partial t}

1084: +g\gamma\left(\frac{\partial \tilde n^{(0)}}{\partial t}-

1085: \frac{\partial n}{\partial t}\right).

1086: \label{eq70}

1087: \end{equation}

1088: Using the continuity equation for the total density $n$, one finds that

1089: (\ref{eq70}) reduces to

1090: \begin{equation}

1091: \frac{\partial \tilde n^{(0)}}{\partial t}

1092: =\frac{1}{1-g\gamma}

1093: \left[\left(\frac{3}{2}\tilde n +g\gamma n_c\right)\frac{1}{T}

1094: \frac{\partial T}{\partial t}+\bbox{\nabla}\cdot(\tilde n{\bf v}_n

1095: +n_c{\bf v}_c) \right].

1096: \label{eq_n0}

1097: \end{equation}

1098: Using (\ref{eq_n0}) in (\ref{eq_totn}), one finds

1099: \begin{equation}

1100: \tau_{\mu}\frac{\partial \Gamma_{12}}{\partial t}+\Gamma_{12}

1101: =\frac{1}{1-g\gamma}

1102: \left[\left(\frac{3}{2}\tilde n +g\gamma n_c\right)\frac{1}{T}

1103: \frac{\partial T}{\partial t}+\bbox{\nabla}\cdot(\tilde n{\bf v}_n

1104: +n_c{\bf v}_c) \right]

1105: +\bbox{\nabla}\cdot(\tilde n{\bf v}_n).

1106: \label{G12_1}

1107: \end{equation} 

1108: We next use (\ref{ptildetot}) in the equation for $\tilde P$

1109: (given by (\ref{eq26c})):

1110: \begin{equation}

1111: \frac{\partial \tilde P}{\partial t}

1112: =\frac{\partial \tilde P^{(0)}}{\partial t}

1113: +\tau_{\mu}\frac{2}{3}gn_c\frac{\partial\Gamma_{12}}{\partial t}

1114: =-\bbox{\nabla}\tilde P\cdot{\bf v}_n-\frac{5}{3}\tilde P \bbox{\nabla}\cdot{\bf v}_n

1115: -\frac{2}{3}gn_c\Gamma_{12}+\frac{2}{3}\bbox{\nabla}\cdot(\kappa\bbox{\nabla}T).

1116: \label{eq73}

1117: \end{equation}

1118: Using the expression for $\tilde P^{(0)}$ given in (\ref{ptilde0}), one finds

1119: \begin{equation}

1120: \frac{\partial \tilde P^{(0)}}{\partial t}=

1121: \left(\frac{5}{2}\tilde P+g\tilde n n_c\right)\frac{1}{T}\frac{\partial T}{\partial t}

1122: +g\tilde n\frac{\partial \tilde n^{(0)}}{\partial t}

1123: +g\tilde n \bbox{\nabla}\cdot(\tilde n{\bf v}_n+n_c{\bf v}_c).

1124: \end{equation}

1125: Substituting this into (\ref{eq73}) in conjunction with (\ref{eq_totn}), we obtain

1126: \begin{eqnarray}

1127: -\left(\frac{2}{3}gn_c+g\tilde n\right)\left(\tau_{\mu}\frac{\partial \Gamma_{12}}

1128: {\partial t}+\Gamma_{12}\right)

1129: &=&\left(\frac{5}{2}\tilde P+g\tilde n n_c\right)\frac{1}{T}\frac{\partial T}{\partial t}

1130: +g\tilde n\bbox{\nabla}\cdot(n_c{\bf v}_c)+\bbox{\nabla}\tilde P\cdot{\bf v}_n  \cr

1131: &&+\frac{5}{3}\tilde P\bbox{\nabla}\cdot{\bf v}_n

1132: -\frac{2}{3}\bbox{\nabla}\cdot(\kappa\bbox{\nabla}T).

1133: \label{G12_2}

1134: \end{eqnarray}

1135: One may now combine (\ref{G12_1}) and (\ref{G12_2})

1136: to eliminate $\partial T/\partial t$ from these two equations.

1137: After a certain amount of rearrangement, we finally obtain our desired equation of

1138: motion for $\Gamma_{12}$:

1139: \begin{equation}

1140: \tau_{\mu}\frac{\partial \Gamma_{12}}{\partial t}+\Gamma_{12}

1141: =\sigma_H\left \{ \bbox{\nabla}\cdot[n_c({\bf v}_c-{\bf v}_n]

1142: +\frac{1}{3}n_c\bbox{\nabla}\cdot{\bf v}_n \right \}

1143: -\frac{2}{3}\frac{\sigma_H\sigma_1}{g}\bbox{\nabla}\cdot

1144: (\kappa\bbox{\nabla}T).

1145: \label{eq_G12}

1146: \end{equation}

1147: If we keep the expansion parameter $\alpha$ and expand $\Gamma_{12}$ as in

1148: (\ref{G12_alpha}), namely 

1149: $\Gamma_{12}=\Gamma_{12}^{(1)}+\alpha\Gamma_{12}^{(2)}$ (we recall that $\Gamma_{12}^{(0)}=0$),

1150: we find $\Gamma_{12}^{(1)}$ is given by (\ref{gamma1}) and 

1151: \begin{equation}

1152: \Gamma_{12}^{(2)}

1153: =-\tau_{\mu}\frac{\partial}{\partial t}\Gamma_{12}^{(1)}

1154: -\frac{2}{3}\frac{\sigma_H\sigma_1}{g\tilde n}\bbox{\nabla}\cdot

1155: (\kappa\bbox{\nabla}T).

1156: \end{equation}

1157: 

1158: In closing this section, we discuss the relation between

1159: the analysis given in this Section and the ZGN$'$ theory~\cite{ZNG}

1160: reviewed in Section~\ref{zgn'}.

1161: In this Section, we started with the complete local equilibrium

1162: distribution given by (\ref{f0}). We then included the deviation from local equilibrium,

1163: as given by (\ref{correction}) with (\ref{solution}).

1164: We showed that the deviation from $f^{(0)}$ associated with $D(u)$ in (\ref{solution})

1165: gives rise to the corrections to the local thermodynamic quantities $\tilde n$, $\tilde P$

1166: and $\tilde\epsilon$.

1167: Such corrections did not arise when we

1168: included the deviation from $\tilde f$ in the ZGN$'$ hydrodynamics~\cite{CJP}.

1169: However, one can show that the type of contribution associated with $D(u)$

1170: is, in fact, already contained in the lowest-order ZGN$'$ distribution function $\tilde f$

1171: given by (\ref{eq8}).

1172: To see this, it is convenient to linearize the distribution function

1173: around {\it static} equilibrium, using $f\simeq f_0+\delta f$.

1174: In the ZGN$'$ theory \cite{ZNG}, one can show that

1175: \begin{equation}

1176: \delta f=\beta_0f_0(1+f_0)\left[\frac{\delta \tilde T}{T_0}

1177: \left(\frac{p^2}{2m}+U_0-\mu_{c0}\right)+{\bf p}\cdot{\bf v}_n-2g\delta n

1178: +\delta \tilde \mu\right].

1179: \label{df_zgn}

1180: \end{equation}

1181: Here we have denoted the temperature fluctuation as $\delta \tilde T$ to make

1182: a distinction from the temperature defined in the diffusive local equilibrium

1183: distribution function (\ref{f0})(we will find that $\delta\tilde T\neq\delta T$).

1184: In the present theory, in contrast, one finds (ignoring the terms in (\ref{solution}) associated 

1185: with the functions $A$ and $B$)

1186: \begin{eqnarray}

1187: \delta f&=&\beta_0f_0(1+f_0)\left[\frac{\delta T}{T_0}

1188: \left(\frac{p^2}{2m}+U_0-\mu_{c0}\right)+{\bf p}\cdot{\bf v}_n-2g\delta n 

1189: +\delta \mu_c^{(0)}\right] \cr ~~ && \cr

1190: &&+f_0(1+f_0)D(u)\delta \Gamma_{12}.

1191: \label{df_ng}

1192: \end{eqnarray}

1193: 

1194: The first term in (\ref{df_ng}) represents the deviation from $f_0$

1195: included in $f^{(0)}$ while the second term is due to $f^{(1)}$.

1196: Using the explicit solution for $D(u)$ given by (\ref{Du})

1197: (derived in Section \ref{sec:transport}), we find that (\ref{df_ng}) can

1198: be written as

1199: \begin{eqnarray}

1200: \delta f&=&\beta_0f_0(1+f_0)\Bigg[\left(\frac{\delta T}{T_0}-\frac{2\sigma_1\tau_{\mu}}

1201: {3\tilde n_0}\delta\Gamma_{12}\right)

1202: \left(\frac{p^2}{2m}+U_0-\mu_{c0}\right)+{\bf p}\cdot{\bf v}_n-2g\delta n  \cr

1203: &&+\delta \mu_c^{(0)}-g(\sigma_H^{-1}+1)\tau_{\mu}\delta\Gamma_{12}\Bigg].

1204: \label{df_ng2}

1205: \end{eqnarray}

1206: We note that this linearized distribution function has the

1207: same form as the ZGN$'$ distribution function in (\ref{df_zgn}), but with a renormalized

1208: local temperature

1209: \begin{equation}

1210: \delta \tilde T\equiv\delta T-\frac{2}{3}T_0\frac{\sigma_1\tau_{\mu}}{\tilde n_0}

1211: \delta\Gamma_{12},

1212: \label{reT}

1213: \end{equation}

1214: and a renormalized local chemical potential

1215: \begin{equation}

1216: \delta\tilde\mu\equiv\delta \mu_c^{(0)}-g(\sigma_H^{-1}+1)\tau_{\mu}\delta\Gamma_{12}.

1217: \label{remu}

1218: \end{equation}

1219: Using $\delta\mu_c=\delta\mu_c^{(0)}-g\tau_{\mu}\delta\Gamma_{12}$

1220: (see (\ref{muctot})) and $\delta \tilde \mu$ from (\ref{remu}), we obtain

1221: \begin{equation}

1222: \delta\mu_{\rm diff}\equiv\delta\tilde\mu-\delta\mu_c

1223: =\delta\tilde \mu-[\delta\mu_c^{(0)}-g\tau_{\mu}

1224: \delta\Gamma_{12}]=g\sigma_H^{-1}\delta\Gamma_{12}.

1225: \end{equation}

1226: This relation between $\delta\mu_{\rm diff}$ and $\delta\Gamma_{12}$ is precisely

1227: that given by (\ref{G12}) and (\ref{taumu}), as derived in the ZGN$'$ theory.

1228: The physical significance of the renormalized thermodynamic quantities, as

1229: given by (\ref{reT}) and (\ref{remu}), will become clear in Section \ref{sec:LK}. 

1230: 

1231: \section{Equivalence to Landau-Khalatnikov Two-Fluid Equations with dissipation}

1232: \label{sec:LK}

1233: 

1234: In this Section, we prove that our hydrodynamic equations in (\ref{hydro-C})

1235: and (\ref{hydro-n}) can be written in the form of the Landau-Khalatnikov two-fluid equations.

1236: We first display the complete Landau-Khalatnikov two-fluid equations involving

1237: dissipative terms~\cite{Khal}:

1238: \begin{mathletters}

1239: \begin{equation}

1240: {\partial n\over\partial t} + \bbox{\nabla}\cdot {\bf j}=0, 

1241: \label{eq_n}

1242: \end{equation}

1243: \begin{eqnarray}

1244: &&m\frac{\partial j_{\mu}}{\partial t}+\frac{\partial}{\partial x_{\nu}}

1245: (\delta_{\mu\nu}P+m\tilde n v_{n\mu}v_{n\nu}+mn_c v_{c\mu}v_{c\nu})

1246: +n\frac{\partial U_{ext}}{\partial x_{\mu}} \cr

1247: && ~~~~~=\frac{\partial}{\partial x_{\nu}}\left\{

1248:  2\eta\left[D_{\mu\nu}-\frac{1}{3}\delta_{\mu\nu}({\rm Tr}D)\right]

1249:  +\delta_{\mu\nu}(\zeta_1\bbox{\nabla}\cdot[mn_c({\bf v}_c-{\bf v}_n)]

1250:  +\zeta_2\bbox{\nabla}\cdot{\bf v}_n)\right\},

1251: \label{eq_j}

1252: \end{eqnarray}

1253: \begin{equation}

1254: {\partial{\bf v}_c\over\partial t} = -\bbox{\nabla}

1255: \left\{\frac{\mu}{m}+\frac{v_c^2}{2}-\zeta_3\bbox{\nabla}\cdot

1256: [mn_c({\bf v}_c-{\bf v}_n)]-\zeta_4\bbox{\nabla}\cdot{\bf v}_n\right\},

1257: \label{eq_vs}

1258: \end{equation}

1259: \begin{equation}

1260: {\partial s\over\partial t} +\bbox{\nabla}\cdot\left(s{\bf v}_n-

1261: \frac{\kappa\bbox{\nabla}T}{T}\right)=\frac{R_s}{T}.

1262: \label{eq_s}

1263: \end{equation}

1264: \label{landau_eqs}

1265: \end{mathletters}

1266: 

1267: \noindent

1268: The total current is given by

1269: ${\bf j}= n_c{\bf v}_c+ \tilde n {\bf v}_n$ and the dissipative function describing the

1270: entropy production rate is given by \cite{Khal}

1271: \begin{eqnarray}

1272: R_s&=&\zeta_2(\bbox{\nabla}\cdot{\bf v}_n)^2+2\zeta_1\bbox{\nabla}\cdot{\bf v}_n

1273: \bbox{\nabla}\cdot[mn_c({\bf v}_c-{\bf v}_n)]+

1274: \zeta_3\left(\bbox{\nabla}\cdot[mn_c({\bf v}_c-{\bf v}_n)]\right)^2 \cr

1275: &&+2\eta\left[D_{\mu\nu}-\frac{1}{3}\delta_{\mu\nu}({\rm Tr}D)\right]^2

1276: +\frac{\kappa}{T}(\bbox{\nabla} T)^2 .

1277: \label{s-rate}

1278: \end{eqnarray}

1279: As we have discussed in Ref~\cite{ZNG}, the normal fluid and the superfluid

1280: densities that appear in the standard Landau two-fluid theory can be identified with

1281: the corresponding non-condensate and condensate densities, within the context

1282: of our finite temperature model based on the HF approximation for single-particle excitations.

1283: We have explicitly made use of this correspondence in writing (\ref{landau_eqs}).

1284: We also note that in (\ref{landau_eqs}) and (\ref{s-rate}),

1285: one can write $n_c({\bf v}_c-{\bf v}_n)$ in the

1286: equivalent form $({\bf j}-n{\bf v}_n)$, which is often used.		

1287: 

1288: The thermodynamic functions that appear in these Landau-Khalatnikov two-fluid equations

1289: satisfy the following superfluid local thermodynamic relations:

1290: \begin{mathletters}

1291: \begin{eqnarray}

1292: &&P+\epsilon=\mu n+sT+m\tilde n({\bf v}_n-{\bf v}_c)^2, 

1293: \label{relation_a} \\

1294: &&dP=nd\mu +sdT-m\tilde n({\bf v}_n-{\bf v}_c)\cdot d({\bf v}_n-{\bf v}_c), 

1295: \label{relation_b} \\

1296: &&d\epsilon=\mu dn +Tds+({\bf v}_n-{\bf v}_c)\cdot d[m\tilde n({\bf v}_n-{\bf v}_c)].

1297: \label{relation_c}

1298: \end{eqnarray}

1299: \label{thermo-relation}

1300: \end{mathletters}

1301: 

1302: \noindent

1303: The various local thermodynamic functions which appear in the LK theory have to be

1304: carefully defined so that they satisfy the relations in (\ref{thermo-relation}).

1305: The local entropy is defined by (as in Ref.~\cite{ZNG})

1306: \begin{equation}

1307: s=\int\frac{d{\bf p}}{(2\pi\hbar)^3}

1308: \left[(1+f)\ln(1+f)-f\ln f\right].

1309: \end{equation}

1310: Using $f=f^{(0)}+f^{(1)}$ and working to first order in $f^{(1)}$, one finds

1311: \begin{equation}

1312: s=\int\frac{d{\bf p}}{(2\pi\hbar)^3}

1313: \left[(1+f^{(0)})\ln(1+f^{(0)})-f^{(0)}\ln f^{(0)}+

1314: \ln(1+f^{(0)^{-1}})f^{(1)}\right].

1315: \end{equation}

1316: From the constraint on $f^{(1)}$ given by (\ref{const_s}), one sees that

1317: the last term, which arises from $f^{(1)}$, makes no contribution to the

1318: local entropy. 

1319: One thus obtains

1320: \begin{equation}

1321: s=

1322: \frac{1}{T}\left[\frac{5}{2}\tilde P^{(0)}-\tilde n^{(0)}

1323: (\tilde \mu^{(0)}-U)\right]

1324: \simeq

1325: \frac{1}{T}\left[\frac{5}{2}\tilde P^{(0)}-\tilde n^{(0)}

1326: (\mu_c^{(0)}-U)-\frac{m\tilde n}{2}({\bf v}_n-{\bf v}_c)^2\right],

1327: \end{equation}

1328: where we have used $\tilde n^{(0)}=\tilde n+O(v_n,v_c)$.

1329: 

1330: The local energy density $\epsilon$ in the Landau-Khalatnikov theory is defined in the

1331: local frame where ${\bf v}_c=0$~\cite{HM}.

1332: In the context of the present theory, this is given by

1333: \begin{equation}

1334: \epsilon=\tilde \epsilon+nU_{\rm ext}+\frac{g}{2}(n^2+2n\tilde n-\tilde n^2)

1335: +\frac{m}{2}\tilde n({\bf v}_n-{\bf v}_c)^2,

1336: \label{totalenergy}

1337: \end{equation}

1338: while the local energy density in the original lab frame is given by

1339: \begin{equation}

1340: \epsilon_{\rm lab}=\epsilon+m\tilde n({\bf v}_n-{\bf v}_c)\cdot{\bf v}_c

1341: +\frac{mn}{2}v_c^2.

1342: \end{equation}

1343: Using (\ref{ntildetot}) and (\ref{ptildetot}) in (\ref{totalenergy}),

1344: one finds that the first order corrections from $\Gamma_{12}$ cancel out, leaving

1345: \begin{equation}

1346: \epsilon=\tilde \epsilon^{(0)}+nU_{\rm ext}+\frac{g}{2}(n^2+2n\tilde n^{(0)}-[\tilde n^{(0)}]^2)

1347: +\frac{1}{2}m\tilde n({\bf v}_n-{\bf v}_c)^2.

1348: \end{equation}

1349: We conclude that {\it both} the local entropy density and the local energy density are

1350: determined by the diffusive local equilibrium distribution function $f^{(0)}$ alone,

1351: and are not altered by the deviation $f^{(1)}$.

1352: 

1353: In contrast, as we now show, the total pressure and the chemical potential must be carefully 

1354: defined so that they satisfy the superfluid thermodynamic relations in (\ref{thermo-relation}).

1355: We first define the {\it non-equilibrium} pressure by

1356: \begin{equation}

1357: P'\equiv\tilde P+\frac{g}{2}(n^2+2n\tilde n-\tilde n^2).

1358: \label{neqP}

1359: \end{equation}

1360: Using (\ref{ntildetot}) and (\ref{ptildetot}) and working to first

1361: order in $\Gamma_{12}^{(1)}$, one obtains

1362: \begin{equation}

1363: P'=P-\tau_{\mu}\frac{gn_c}{3}\Gamma_{12}^{(1)},

1364: \label{neqP2}

1365: \end{equation}

1366: where $P$ is the (diffusive) local equilibrium pressure defined as

1367: \begin{equation}

1368: P\equiv\tilde P^{(0)}+\frac{g}{2}(n^2+2n\tilde n^{(0)}-[\tilde n^{(0)}]^2). \\

1369: \label{eqP}

1370: \end{equation}

1371: We find that the LK thermodynamic relations given in (\ref{thermo-relation}) are not

1372: satisfies if we assume that $P'$ is the pressure ($P$) and $\mu_c$ is the

1373: chemical potential ($\mu$).

1374: Extra terms appear which are associated with $\Gamma_{12}$.

1375: This means that the above identification of the thermodynamic variables is only valid in the

1376: lowest order hydrodynamic equations, where there is no dissipation.

1377:  

1378: We recall that in deriving the Landau equations from the ZGN$'$ equations

1379: in Ref.~\cite{ZNG}, we defined the total pressure by (\ref{neqP}) and $\mu=\mu_c$,

1380: and also found extra terms in the thermodynamic relations proportional to

1381: $\delta\mu_{\rm diff}$ (see Eq.~(71) of Ref.~\cite{ZNG}).

1382: Therefore the precise equivalence between the ZGN$'$ hydrodynamics and the Landau

1383: theory shown in Ref.~\cite{ZNG} was restricted in the limit $\omega\tau_{\mu}\to 0$,

1384: i.e., when $\delta\mu_{\rm diff}\to 0$.

1385: In contrast, if the pressure $P$ is defined to be (\ref{eqP}) and $\mu=\mu_c^{(0)}$, 

1386: we can show that the the superfluid thermodynamic relations in (\ref{thermo-relation})

1387: are satisfied.

1388: Therefore we conclude that the local equilibrium pressure defined in (\ref{eqP})

1389: and the local equilibrium chemical potential $\mu_c^{(0)}$ given by (\ref{muc01})

1390: are, in fact, the correct variables to be use in the Landau-Khalatnikov equations.

1391: We will show later that the corrections to the total pressure and the

1392: chemical potential actually give rise to the additional damping terms associated with

1393: the four second viscosity coefficients $\zeta_i$ in (\ref{landau_eqs}).

1394: 

1395: We now proceed to derive the LK equations from our microscopic theory, one by one.

1396: Our continuity equations for $n_c$ and $\tilde n$ are given by

1397: (\ref{eq_nc}) and (\ref{eq_tilden}).

1398: Adding them, we obtain the continuity equation for the total

1399: density (\ref{eq_n}).

1400: To derive the equation (\ref{eq_j}) for the total current ${\bf j}$,

1401: we combine our two continuity equations and the two velocity

1402: equations (\ref{eq_vc}) and (\ref{eq_vn}) to give

1403: \begin{eqnarray}

1404: &&m\frac{\partial j_{\mu}}{\partial t}+\frac{\partial}{\partial x_{\nu}}

1405: (\delta_{\mu\nu}P'+m\tilde n v_{n\mu}v_{n\nu}+mn_c v_{c\mu}v_{c\nu})

1406: +n\frac{\partial U_{\rm ext}}{\partial x_{\mu}} \cr

1407: && =\frac{\partial}{\partial x_{\nu}}\left\{

1408:  2\eta\left[D_{\mu\nu}-\frac{1}{3}\delta_{\mu\nu}({\rm Tr}D)\right]\right\}.

1409: \label{eq_j2}

1410: \end{eqnarray}

1411: Using (\ref{neqP2}), we find

1412: \begin{eqnarray}

1413: &&m\frac{\partial j_{\mu}}{\partial t}+\frac{\partial}{\partial x_{\nu}}

1414: (\delta_{\mu\nu}P+m\tilde n v_{n\mu}v_{n\nu}+mn_c v_{c\mu}v_{c\nu})

1415: +n\frac{\partial U_{\rm ext}}{\partial x_{\mu}} \cr

1416: && =\frac{\partial}{\partial x_{\nu}}\left\{

1417:  2\eta\left[D_{\mu\nu}-\frac{1}{3}\delta_{\mu\nu}({\rm Tr}D)\right]

1418: +\delta_{\mu\nu}\tau\frac{gn_c}{3}\Gamma_{12}^{(1)}\right\}.

1419: \label{eq_j3}

1420: \end{eqnarray}

1421: To consistently include damping due to the first order correction

1422: term in the Chapman-Enskog expansion, we use

1423: $\Gamma_{12}=\Gamma_{12}^{(1)}$ as given in (\ref{gamma1}).

1424: We then find that (\ref{eq_j3}) is identical with the LK equation (\ref{eq_j})

1425: with the second viscosity coefficients $\zeta_1$ and $\zeta_2$ 

1426: given by

1427: \begin{equation}

1428: \zeta_1=\frac{gn_c}{3m}\tau_{\mu}\sigma_H, ~~ 

1429: \zeta_2=\frac{gn_c^2}{9}\tau_{\mu}\sigma_H.

1430: \label{zeta12}

1431: \end{equation}

1432: Using $\mu_c=\mu_c^{(0)}+g\tilde n^{(1)}=\mu-g\tau\Gamma_{12}^{(1)}$ (see (\ref{muc01}))

1433: and the expression for $\Gamma_{12}^{(1)}$ in our equation for the

1434: condensate velocity given in (\ref{eq_vc}), we find the latter can be written precisely

1435: in the LK form (\ref{eq_vs}).

1436: Comparison between the two equations shows that the second viscosity coefficients

1437: $\zeta_3$ and $\zeta_4$ are

1438: given by

1439: \begin{equation}

1440: \zeta_3=\frac{g}{m^2}\tau_{\mu}\sigma_H, ~~\zeta_4=\frac{gn_c}{3m}\tau_{\mu}\sigma_H.

1441: \label{zeta34}

1442: \end{equation}

1443: We note that our results for the second viscosities satisfy the Onsager reciprocal

1444: relation $\zeta_1=\zeta_4$ (this equality follows quite generally, as shown

1445: by Eq.~(4.28) of Ref.~\cite{HM}).

1446: 

1447: Finally, we derive the equation for the local entropy.

1448: Using (\ref{relation_c}), we have

1449: \begin{equation}

1450: T\frac{\partial s}{\partial t}=\frac{\partial \epsilon}{\partial t}

1451: -\mu\frac{\partial n}{\partial t}

1452: -m({\bf v}_n-{\bf v}_c)\frac{\partial}{\partial t}

1453: [\tilde n({\bf v}_n-{\bf v}_c)].

1454: \label{entropy_1}

1455: \end{equation}

1456: With the expression for the local energy density $\epsilon$ given in (\ref{totalenergy}),

1457: we find (\ref{entropy_1}) reduces to

1458: \begin{eqnarray}

1459: \frac{\partial s}{\partial t}&=&\frac{\partial\tilde\epsilon}{\partial t}

1460: +[U_{\rm ext}-\mu+g(n+\tilde n)]\frac{\partial n}{\partial t}

1461: +gn_c\frac{\partial \tilde n}{\partial t}-\frac{\partial n}{\partial t}

1462: \frac{m}{2}({\bf v}_n-{\bf v}_c)^2 \cr

1463: && \cr

1464: &=&\frac{\partial\tilde\epsilon}{\partial t}+g\tilde n^{(1)}\frac{\partial n}{\partial t}

1465: +gn_c\frac{\partial \tilde n}{\partial t}.

1466: \label{entropy_2}

1467: \end{eqnarray}

1468: Here we have neglected the last term in the first line, since it is

1469: of third order in the local velocities.

1470: Using our hydrodynamic equations (\ref{eqhydro}), 

1471: we find (\ref{entropy_2}) reduces to the form (\ref{eq_s}),

1472: assuming the entropy production rate $R_s$ is given by

1473: \begin{eqnarray}

1474: R_s&=&\tau_{\mu} g\Gamma_{12}

1475: \left\{g\bbox{\nabla}\cdot[n_c({\bf v}_c-{\bf v}_n)]+\frac{1}{3}n_c\bbox{\nabla}\cdot{\bf v}_n

1476: \right\} \cr

1477: &&+2\eta\left[D_{\mu\nu}-\frac{1}{3}\delta_{\mu\nu}({\rm Tr}D)\right]^2

1478: +\frac{\kappa}{T}(\bbox{\nabla} T)^2 .

1479: \label{s-rate2}

1480: \end{eqnarray}

1481: Using (\ref{gamma1}) and the expression in (\ref{zeta12}) and (\ref{zeta34}),

1482: we see that (\ref{s-rate2}) is equivalent to the Landau-Khalatnikov~\cite{Khal}

1483: expression given in (\ref{s-rate}).

1484: 

1485: We have thus shown that our equations based on a microscopic theory built on Bose

1486: condensation can be written in a form precisely identical

1487: to the phenomenological Landau-Khalatnikov two-fluid equations including

1488: the damping associated with the shear viscosity, 

1489: thermal conductivity and four second viscosity coefficients.

1490: An analogous derivation of the Landau-Khalatnikov equations

1491: for a uniform Bose gas was first given by Kirkpatrick and Dorfman \cite{KD}.

1492: However at finite temperatures, where the dominant excitations are particle-like

1493: Hartree-Fock excitations, KD did not obtain the second viscosities since they neglected

1494: the source term

1495: $\Gamma_{12}$ associated with deviation from local equilibrium produced by the 

1496: $C_{12}$ collisions.

1497: We have shown that the second viscosity coefficients are directly related to the 

1498: $\Gamma_{12}$ term first discussed in Ref.~\cite{ZNG},

1499: which represents the collisional exchange of atoms between

1500: the condensate and non-condensate.

1501: 

1502: In the above derivation of the second viscosity terms,

1503: we used $\Gamma_{12}=\Gamma_{12}^{(1)}$.

1504: This restricts the validity of the results to the case $\omega\tau_{\mu}\ll 1$

1505: when we consider collective fluctuations with frequency $\omega$.

1506: However, our discussion can be easily extended to the situation when $\omega\tau_{\mu}$ is

1507: not small, by using (see (\ref{eq_mudiff2}))

1508: \begin{equation}

1509: \Gamma_{12}(\omega)=\frac{\sigma_H}{1-i\omega\tau_{\mu}}

1510: \left\{\bbox{\nabla}\cdot[n_c({\bf v}_c-{\bf v}_n)]+\frac{1}{3}n_c\bbox{\nabla}\cdot

1511: {\bf v}_n\right\}.

1512: \end{equation}

1513: Using this expression, we can still write our equations in the Landau-Khalatnikov form,

1514: but now with the {\it frequency-dependent} second viscosity coefficients

1515: \begin{equation}

1516: \zeta_i(\omega)=\frac{\zeta_i}{1-i\omega\tau_{\mu}}.

1517: \label{zeta_omega}

1518: \end{equation}

1519: Everything else in our derivation goes through.

1520: 

1521: The expression (\ref{zeta_omega}) for the frequency-dependent second viscosity coefficients has

1522: in fact the expected form, as derived from general considerations \cite{LL}.

1523: The second viscosity , such as associated with compression and expansion, 

1524: arises when a gas is coupled

1525: to an internal relaxation process (for example, the transfer of energy from

1526: the translational degrees of freedom of a molecule to the vibrational degrees of

1527: freedom).

1528: If the relaxation time of the internal process is denoted by $\tau_{\rm R}$,

1529: the frequency-dependent second viscosity coefficient is given by

1530: $\zeta(\omega)=\zeta_0/(1-i\omega\tau_{\rm R})$, 

1531: where $\zeta_0\propto \tau_{\rm R}$.

1532: In a Bose-condensed gas at finite temperatures, the non-condensate atoms are

1533: coupled to the condensate degree of freedom, and we have shown that 

1534: the characteristic relaxation

1535: time for the equilibration between the two components is given by $\tau_{\mu}$.

1536: In this connection, we might recall that

1537: in the superfluid $^4$He, the second viscosities are due to the fact that

1538: the total number of phonons and rotons is not conserved~\cite{Khal}.

1539: Above $T_{\rm BEC}$ (when $n_c=0$), all the second viscosity coefficients $\zeta_i$

1540: in (\ref{zeta12}) and (\ref{zeta34}) vanish, as expected in a normal dilute

1541: single-component gas.

1542: 

1543: We finally note that the Landau-Khalatnikov equations could have been

1544: derived from the ZGN$'$ hydrodynamic equations if we simply identified the

1545: total pressure $P$ with that by (\ref{neqP}) and (\ref{neqP2}),

1546: and took the chemical potential to be $\mu=\mu_c+g\tau_{\mu}\delta\Gamma_{12}$

1547: (see Eq.~(\ref{muctot})).

1548: This leads more naturally to frequency-dependent second viscosities.

1549: On the other hand, the physical meaning of this choice of

1550: the local equilibrium pressure and  chemical potential is not made clear.

1551: 

1552: \section{Calculation of transport coefficients}

1553: \label{sec:transport}

1554: In this section, we solve the linearized equation for functions $A$, $B$

1555: and $D$ in (\ref{eqforABD}) which determine the deviation from local equilibrium

1556: as described by $\psi$ in (\ref{correction}) and (\ref{solution}).

1557: We can then calculate the transport coefficients $\eta$ and $\kappa$ as given

1558: in (\ref{eta-kappa}).

1559: We follow the standard procedure in the Chapman-Enskog method, as reviewed in \cite{ferziger}.

1560: In this approach, one solves the linearized equation

1561: by expanding $\psi$ in a basis set of polynomial functions.

1562: These polynomial functions are chosen to satisfy the constraints such as (\ref{const}) which

1563: the solution $\psi$ must satisfy.

1564: In a classical gas, one uses Sonine polynomials \cite{ferziger}.

1565: One can also define analogous polynomials for a degenerate Bose gas~\cite{UU}.

1566: As usual, we calculate the transport coefficients using

1567: the lowest order polynomial approximation, which 

1568: usually gives very accurate results for

1569: the transport coefficients.

1570: For a more detailed mathematical discussion which is easily generalized to

1571: Bose condensed gases, we refer to \cite{ferziger,UU}.

1572: 

1573: \subsection{The thermal conductivity}

1574: In evaluating the thermal conductivity, it is convenient to rewrite

1575: (\ref{kappa}) as

1576: \begin{eqnarray}

1577: \kappa&=&-\frac{1}{3}k_{\rm B}\int\frac{d{\bf p}}{(2\pi\hbar)^3}{\bf u}A(u)

1578: \cdot{\bf u}\left[\frac{mu^2}{2k_{\rm B}T_0}-\frac{5g_{5/2}(z_0)}{2g_{3/2}(z_0)}\right]

1579: f_0(1+f_0) \cr

1580: &=&-\frac{1}{3}k_{\rm B}\int\frac{d{\bf p}}{(2\pi\hbar)^3}{\bf u}A(u)

1581: \cdot\hat L[{\bf u}A(u)].

1582: \label{kappa2}

1583: \end{eqnarray}

1584: Here we explicitly use the static equilibrium distribution function $f_0$

1585: to evaluate the transport coefficients.

1586: To solve the linear integral equation (\ref{eqforA}) for $A(u)$, we introduce a

1587: simple ansatz of the form \cite{NG,UU}:

1588: \begin{equation}

1589: A(u)=A\left[\frac{mu^2}{2k_{\rm B}T_0}-\frac{5g_{5/2}(z_0)}{2g_{3/2}(z_0)}\right].

1590: \label{ansatzA}

1591: \end{equation} 

1592: This is the lowest order polynomial function that satisfies the constraint

1593: given by (\ref{constA}).

1594: The constant $A$ in (\ref{ansatzA}) is determined by multiplying (\ref{eqforA}) by 

1595: ${\bf u}[mu^2/2k_{\rm B}T-5g_{5/2}(z_0)/2g_{3/2}(z_0)]$ and integrating

1596: over ${\bf p}$, giving

1597: \begin{eqnarray}

1598: A&=&

1599: \int \frac{d{\bf p}}{(2\pi\hbar)^3} u^2

1600: \left[\frac{mu^2}{2k_{\rm B}T_0}-\frac{5g_{5/2}(z_0)}{2g_{3/2}(z_0)}\right]^2

1601: f_0(1+f_0) \cr

1602: & &

1603: \times

1604: \left\{\int \frac{d{\bf p}}{(2\pi\hbar)^3}

1605: \left[\frac{mu^2}{2k_{\rm B}T_0}-\frac{5g_{5/2}(z_0)}{2g_{3/2}(z_0)}

1606: \right]{\bf u}

1607: \cdot \hat L\left[

1608: \left\{\frac{mu^2}{2k_{\rm B}T_0}-\frac{5g_{5/2}(z_0)}{2g_{3/2}(z_0)}

1609: \right\}{\bf u}\right]\right\}^{-1}

1610: \label{eqAconst}

1611: \end{eqnarray}

1612: Using (\ref{ansatzA}) and (\ref{eqAconst}) in (\ref{kappa2}), we find

1613: \begin{eqnarray}

1614: \kappa&=&-\frac{k_{\rm B}}{3}

1615: \left\{\int \frac{d{\bf p}}{(2\pi\hbar)^3} u^2

1616: \left[\frac{mu^2}{2k_{\rm B}T_0}-\frac{5g_{5/2}(z_0)}{2g_{3/2}(z_0)}\right]^2

1617: f_0(1+f_0) \right\}^2\cr

1618: & &

1619: \times

1620: \left\{\int \frac{d{\bf p}}{(2\pi\hbar)^3}

1621: \left[\frac{mu^2}{2k_{\rm B}T_0}-\frac{5g_{5/2}(z_0)}{2g_{3/2}(z_0)}

1622: \right]{\bf u}

1623: \cdot \hat L\left[

1624: \left\{\frac{mu^2}{2k_{\rm B}T_0}-\frac{5g_{5/2}(z_0)}{2g_{3/2}(z_0)}

1625: \right\}{\bf u}\right]\right\}^{-1}. 

1626: \label{kappa3}

1627: \end{eqnarray}

1628: 

1629: To evaluate the ${\bf p}$ integrals in (\ref{kappa3}), it is convenient to introduce

1630: the dimensionless velocity variable by

1631: \begin{equation}

1632: {\bf x}\equiv\left(\frac{m}{2k_{\rm B}T_0}\right)^{1/2}{\bf u}=

1633: \left(\frac{1}{2mk_{\rm B}T_0}\right)^{1/2}{\bf p}.

1634: \label{u2x}

1635: \end{equation}

1636: With this new variable, we can rewrite the linearized collision operator as

1637: \begin{equation}

1638: \hat L[\psi]=\frac{8m(k_{\rm B}T_0)^2a^2}{\pi^3\hbar^3}

1639: \left(\hat L'_{22}[\psi]+\pi^{3/2}n_{c0}\Lambda^3_0\hat L_{12}'[\psi]\right),

1640: \end{equation}

1641: where the dimensionless collision operators $\hat L_{22}'$ and $\hat L_{12}'$ 

1642: are defined by

1643: \begin{eqnarray}

1644: \hat L_{22}'[\psi]&\equiv&\int d{\bf x}_2\int d{\bf x}_3\int d{\bf x}_4

1645: \delta({\bf x}+{\bf x}_2-{\bf x}_3-{\bf x}_4)

1646: \delta(x^2+x_2^2-x_3^2-x_4^2) \cr

1647: &&\times f_{10}f_{20}(1+f_{30})(1+f_{40})

1648: (\psi_3+\psi_4-\psi_2-\psi),

1649: \end{eqnarray}

1650: \begin{eqnarray}

1651: \hat L_{12}'[\psi]&\equiv&\int d{\bf x}_1\int d{\bf x}_2\int d{\bf x}_3

1652: \delta({\bf x}-{\bf x}_2-{\bf x}_3)

1653: \delta(x_1^2-\beta_0 gn_{c0}-x_2^2-x_3^2) \cr

1654: &&\times [\delta({\bf x}-{\bf x}_1)-\delta({\bf x}-{\bf x}_2)-\delta({\bf x}-{\bf x}_3)]

1655: (1+f_{10})f_{20}f_{30}(\psi_2+\psi_3-\psi),

1656: \end{eqnarray}

1657: where $f_{i0}=(z_0^{-1}e^{x_i^2}-1)^{-1}$.

1658: Carrying out the ${\bf p}$ (or ${\bf u}$) integrals in (\ref{kappa3}), one finds

1659: \begin{equation}

1660: \kappa=\frac{75k_{\rm B}}{64a^2m}\left(\frac{mk_{\rm B}T}{\pi}\right)^{1/2}

1661: \frac{\pi^{1/2}}{I^{\kappa}_{22}(z_0)+\Lambda^3_0n_{c0}I_{12}^{\kappa}(z_0)}

1662: \left[\frac{7}{2}g_{7/2}(z_0)-\frac{5g^2_{5/2}(z_0)}{2g_{3/2}(z_0)} \right]^2,

1663: \label{kappa_final}

1664: \end{equation}

1665: where the functions $I_{22}^{\kappa}(z_0)$ and $I_{12}^{\kappa}(z_0)$ are defined by

1666: \begin{equation}

1667: I_{22}^{\kappa}\equiv-\int d{\bf x}~{\bf x}x^2\cdot\hat L_{22}'[{\bf x}x^2],

1668: \end{equation}

1669: \begin{equation}

1670: I_{12}^{\kappa}\equiv-\pi^{3/2}\int d{\bf x}~{\bf x}x^2\cdot\hat L_{22}'[{\bf x}x^2].

1671: \end{equation}

1672: 

1673: In Ref.~\cite{NG}, we derived a convenient

1674: formula for the integral $I_{22}^{\kappa}$, namely

1675: \begin{eqnarray}

1676: I_{22}^{\kappa}(z_0)&=&\sqrt{2}\pi^3=\int_0^{\infty}dx_0\int_0^{\infty}dx_r

1677: \int_{-1}^1dy\int_{-1}^1 dy' F_{22}(x_0,x_r,y,y';z_0) \cr

1678: &&\times x_0^4x_r^7(y^2+y'^2-2y^2y'^2),

1679: \end{eqnarray}

1680: where

1681: \begin{equation}

1682: F_{22}(x_0,x_r,y,y';z_0)

1683: =\frac{z^2_0e^{-x_0^2-x_r^2}}

1684: {(1-z_0e^{-x_1^2})(1-z_0e^{-x_2^2})(1-z_0e^{-x_3^2})(1-z_0e^{-x_4^2})},

1685: \label{F22}

1686: \end{equation}

1687: with

1688: \begin{eqnarray}

1689: &&x_1^2=\frac{1}{2}(x_0^2+2x_0x_ry+x_r^2), ~x_2^2=\frac{1}{2}(x_0^2-2x_0x_ry+x_r^2) \cr

1690: &&x_3^2=\frac{1}{2}(x_0^2+2x_0x_ry'+x_r^2), ~x_4^2=\frac{1}{2}(x_0^2-2x_0x_ry'+x_r^2).

1691: \end{eqnarray}

1692: We note that $I_{22}^{\kappa}(z_0)$ is a universal function of the equilibrium fugacity

1693: $z_0$, where $z_0=e^{-\beta gn_{c0}({\bf r})}$.

1694: To derive a similar expression for $I_{12}^{\kappa}$, we introduce the

1695: transformation 

1696: \begin{equation}

1697: {\bf x}_2=\frac{1}{2}({\bf x}_0+{\bf x}_r), ~~

1698: {\bf x}_3=\frac{1}{2}({\bf x}_0-{\bf x}_r).

1699: \end{equation}

1700: We then express ${\bf x}_r$ in the polar coordinate $(x_r,\theta,\phi)$

1701: where $\theta$ is the azimuthal angle with respect to the vector ${\bf x}_0$,

1702: ie, ${\bf x}_r\cdot{\bf x}_0=x_rx_0\cos\theta$.

1703: With these new variables, one obtains the following expression for $I_{12}^{\kappa}$:

1704: \begin{eqnarray}

1705: I_{12}^{\kappa}(z_0)&=&8\pi^{7/2}\int_0^{\infty}dx_r\int_{-1}^1 dy F_{12}(x_r,y;z_0)

1706: x_r^2(x_r^2+\beta_0 gn_{c0})^{3/2} \cr

1707: &&\times\left[x_r^2(x_r^2+3\beta_0 gn_{c0})(1-y^2)+\frac{9}{4}(\beta_0 gn_{c0})^2\right],

1708: \end{eqnarray}

1709: where $y=\cos\theta$ and

1710: \begin{equation}

1711: F_{12}(x_r,y;z_0)=\frac{ z_0e^{-x_1^2} }

1712: { (1-z_0e^{-x_1^2})(1-z_0e^{-x_2^2})(1-z_0e^{-x_3^2}) },

1713: \label{F12}

1714: \end{equation}

1715: with

1716: \begin{eqnarray}

1717: &&x_1^2=2(x_r^2+\beta_0 gn_{c0}), \cr

1718: &&x_2^2=x_r^2+x_ry\sqrt{x_r^2+\beta_0 gn_{c0}}+\frac{1}{2}\beta_0 gn_{c0} \cr

1719: &&x_3^2=x_r^2-x_ry\sqrt{x_r^2+\beta_0 gn_{c0}}+\frac{1}{2}\beta_0 gn_{c0}.

1720: \end{eqnarray}

1721: The formula in (\ref{kappa_final}) gives the thermal conductivity $\kappa$

1722: as a universal function of $gn_{c0}({\bf r})$ or equivalently in terms of

1723: the local fugacity $z_0({\bf r})=e^{-\beta_0 gn_{c0}({\bf r})}$.

1724: If we ignore the contribution from $C_{12}$ collisions, i.e.,

1725: set $I_{12}^{\kappa}$ to zero, (\ref{kappa_final}) reduces to the

1726: expression for $\kappa$ derived in our earlier work~\cite{NG,CJP}.

1727: 

1728: One can also write the expression for $\kappa$ in (\ref{kappa_final}) in the

1729: following useful form:

1730: \begin{equation}

1731: \kappa=\frac{5\sqrt{2}}{\pi^3}

1732: \left(\tau_{\kappa}\frac{\tilde n_0 k_{\rm B}^2T_0}{m}\right)

1733: \left\{\frac{7g_{7/2}(z_0)}{2g_{5/2}(z_0)}-

1734: \left[\frac{5g_{5/2}(z_0)}{2g_{3/2}(z_0)}\right]^2 \right\},

1735: \label{kappa_final2}

1736: \end{equation}

1737: where $\tau_{\kappa}$ is the ``thermal relaxation time''

1738: associated with the thermal conductivity, as defined in Appendix \ref{relax_times}.

1739: In turn, one can also write the reciprocal of this relaxation time $\tau_{\kappa}$ as

1740: the sum of contributions from $C_{12}$ and $C_{22}$ collisions,

1741: \begin{equation}

1742: \frac{1}{\tau_{\kappa}}=\frac{1}{\tau_{\kappa,12}}+\frac{1}{\tau_{\kappa,22}},

1743: \label{tau_k2}

1744: \end{equation}

1745: where these relaxation times are given explicitly in Appendix~\ref{relax_times}.

1746: The physical meaning of this $\tau_{\kappa}$ relaxation time is discussed in

1747: Appendix \ref{relax_times}, using a simple relaxation time approximation

1748: for the collision integrals in (\ref{eq1}).

1749: 

1750: \subsection{The shear viscosity}

1751: In evaluating the shear viscosity $\eta$, it is convenient to rewrite (\ref{eta}) as

1752: \begin{eqnarray}

1753: \eta&=&

1754: -\frac{m}{10}\int \frac{d{\bf p}}{(2\pi\hbar)^3}

1755: \left(u_{\mu}u_{\nu}-\frac{1}{3}\delta_{\mu\nu}u^2\right)

1756: B(u)\left(u_{\mu}u_{\nu}-\frac{1}{3}\delta_{\mu\nu}u^2\right)

1757: f_0(1+f_0) \cr

1758: &=&

1759: -\frac{k_{\rm B}T_0}{10}\int \frac{d{\bf p}}{(2\pi\hbar)^3}

1760: B(u)\left(u_{\mu}u_{\nu}-\frac{1}{3}\delta_{\mu\nu}u^2\right)

1761: \hat L\left[B(u)\left(u_{\mu}u_{\nu}-\frac{1}{3}\delta_{\mu\nu}u^2\right)\right].

1762: \label{eta1}

1763: \end{eqnarray}

1764: To solve (\ref{eta1}), the simplest consistent ansatz \cite{NG,UU} is to use $B(u)\equiv B$.

1765: The constant $B$ can be determined by multiplying (\ref{eqforB}) by

1766: $(u_{\mu}u_{\nu}-\delta_{\mu\nu}u^2/3)$ and integrating over ${\bf p}$, to give

1767: \begin{eqnarray}

1768: B&=&\frac{m}{k_{\rm B}T_0}

1769: \left\{\int \frac{d{\bf p}}{(2\pi\hbar)^3}

1770: \left(u_{\mu}u_{\nu}

1771: -\frac{1}{3}\delta_{\mu\nu}u^2\right)^2

1772: f_0(1+f_0)\right\} \cr

1773: &&\times\left\{\int \frac{d{\bf p}}{(2\pi\hbar)^3}

1774: \left(u_{\mu}u_{\nu}-\frac{1}{3}\delta_{\mu\nu}u^2\right)

1775: \hat L\left[u_{\mu}u_{\nu}

1776: -\frac{1}{3}\delta_{\mu\nu}u^2 \right]\right\}^{-1}.

1777: \label{eqBconst}

1778: \end{eqnarray}

1779: Using this in (\ref{eta1}), we obtain

1780: \begin{eqnarray}

1781: \eta&=&-\frac{m^2}{10k_{\rm B}T_0}

1782: \left\{\int \frac{d{\bf p}}{(2\pi\hbar)^3}

1783: \left(u_{\mu}u_{\nu}

1784: -\frac{1}{3}\delta_{\mu\nu}u^2\right)^2

1785: f_0(1+f_0)\right\}^2 \cr

1786: &&\left\{\int \frac{d{\bf p}}{(2\pi\hbar)^3}

1787: \left(u_{\mu}u_{\nu}-\frac{1}{3}\delta_{\mu\nu}u^2\right)

1788: \hat L\left[u_{\mu}u_{\nu}

1789: -\frac{1}{3}\delta_{\mu\nu}u^2 \right]\right\}^{-1}.

1790: \label{eta2}

1791: \end{eqnarray}

1792: With the dimensionless variable defined in (\ref{u2x}), this expression for

1793: the shear viscosity $\eta$ can be rewritten as

1794: \begin{equation}

1795: \eta=\frac{5\pi^3}{32\sqrt{2}a^2}

1796: (mk_{\rm B}T_0)^{1/2}

1797: \frac{g^2_{5/2}(z_0) }{I_{22}^{\eta}(z_0)+\Lambda^3_0 n_{c0} I_{12}^\eta(z_0) },

1798: \label{etafinal}

1799: \end{equation}

1800: where

1801: \begin{eqnarray}

1802: I_{22}^{\eta}&\equiv& -\int d{\bf x}

1803: \left(x_{\mu}x_{\nu}-\frac{1}{3}\delta_{\mu\nu}x^2\right)

1804: \hat L'_{22}\left[x_{\mu}x_{\nu}

1805: -\frac{1}{3}\delta_{\mu\nu}x^2 \right] \cr

1806: &=&\frac{\pi^3}{\sqrt{2}}\int_0^{\infty} dx_0 \int_0^{\infty}

1807: dx_r\int_{-1}^1 dy \int_{-1}^1 dy' F_{12}(x_0,x_r,y,y';z_0) \cr

1808: && \times x_0^2x_r^7(1+y^2+y'^2-3y^2y'^2),

1809: \end{eqnarray}

1810: and

1811: \begin{eqnarray}

1812: I_{12}^{\eta}&\equiv& -\pi^{3/2}\int d{\bf x}

1813: \left(x_{\mu}x_{\nu}-\frac{1}{3}\delta_{\mu\nu}x^2\right)

1814: \hat L'_{12}\left[x_{\mu}x_{\nu}

1815: -\frac{1}{3}\delta_{\mu\nu}x^2 \right] \cr

1816: &=&8\pi^{7/2}\int_0^{\infty}dx_r\int_{-1}^{1}dy

1817: F_{12}(x_r,y;z_0) \cr

1818: &&\times x_r^2\sqrt{x_r^2+\beta_0 gn_{c0}}

1819: \left[x_r^2(x_r^2+\beta_0 gn_{c0})(1-y^2)+\frac{1}{3}(\beta_0 gn_{c0})^2\right].

1820: \end{eqnarray}

1821: These expressions involve the same functions $F_{22}$ and $F_{12}$ defined 

1822: earlier in (\ref{F22}) and (\ref{F12}).

1823: 

1824: Analogous to (\ref{kappa_final2}) for the thermal conductivity,

1825: can also write the expression for $\eta$ in (\ref{etafinal}) in the

1826: following form:

1827: \begin{equation}

1828: \eta=\tau_{\eta}\tilde n_0k_{\rm B}T_0 \left[\frac{g_{5/2}(z_0)}{g_{3/2}(z_0)}\right],

1829: \label{etafinal2}

1830: \end{equation}

1831: where the viscous relaxation time $\tau_{\eta}$ is defined in Appendix~\ref{relax_times}.

1832: Again one can write the reciprocal of the relaxation time $\tau_{\eta}$ as

1833: \begin{equation}

1834: \frac{1}{\tau_{\eta}}=\frac{1}{\tau_{\eta,12}}+\frac{1}{\tau_{\eta,22}},

1835: \label{tau_e2}

1836: \end{equation}

1837: where these $C_{12}$ and $C_{22}$

1838: relaxation times are given in Appendix~\ref{relax_times}.

1839: As with $\tau_{\kappa}$, the relaxation time $\tau_{\eta}$ can be understood in

1840: terms of a simple relaxation time approximation for the collision term in the

1841: kinetic equation.

1842: 

1843: \subsection{The second viscosity coefficients}

1844: To find the expression for

1845: $\tau$ as defined in (\ref{tau}), we use the simple

1846: ansatz for the form of the solution for $D(u)$ of (\ref{eqforD}),

1847: \begin{equation}

1848: D(u)=D\left(\sigma_2+\frac{mu^2}{3k_{\rm B}T_0}\sigma_1\right),

1849: \label{approxD}

1850: \end{equation}

1851: where $\sigma_1$ and $\sigma_2$ are defined in (\ref{sigma12}).

1852: As usual, we leave the dependence on $({\bf r},t)$ implicit.

1853: One easily verifies that (\ref{approxD}) satisfies the constraint (\ref{constD}).

1854: The constant $D$ can be determined by integrating (\ref{eqforD}) over ${\bf p}$:

1855: \begin{eqnarray}

1856: D\int \frac{d{\bf p}}{(2\pi\hbar)^3}\hat L_{12}[\sigma_2+\frac{mu^2}{3k_{\rm B}T_0}\sigma_1]

1857: &=&\frac{1}{\tilde n^{(0)}}\int \frac{d{\bf p}}{(2\pi\hbar)^3}\left(\sigma_2

1858: +\frac{mu^2}{3k_{\rm B}T_0}\right)f_0(1+f_0) \cr

1859: &&-\beta_0 g\tau \int \frac{d{\bf p}}{(2\pi\hbar)^3}\hat L_{12}[1].

1860: \label{eqVC2}

1861: \end{eqnarray}

1862: Using $\hat L_{12}[\frac{mu^2}{2}]=\beta_0 gn_{c0}\hat L_{12}[1]$ and

1863: $\int \frac{d{\bf p}}{(2\pi\hbar)^3}\hat L_{12}[1]=-n_{c0}/\tau_{12}$,

1864: where $\tau_{12}$ is defined in (\ref{tau12}), (\ref{eqVC2}) gives finally

1865: \begin{equation}

1866: D=-\frac{(\tau_{12}+\beta_0 gn_{c0}\tau)}{n_{c0}[\sigma_2+\frac{2}{3}\sigma_1

1867: \beta(\mu_{c0}-U_0)]}.

1868: \label{solutionD}

1869: \end{equation}

1870: Using (\ref{solutionD}) and (\ref{approxD}) in the expression for $\tau$ given by

1871: (\ref{tau}), we can solve to give an explicit expression for $\tau$, namely

1872: \begin{equation}

1873: \frac{1}{\tau}=\frac{1}{\tau_{12}}\left\{\frac{n_{c0}} {\tilde n_0}

1874: \left[\sigma_2+\frac{2}{3}\sigma_1\beta_0(\mu_{c0}-U_0)\right]-\beta_0 g n_{c0}

1875: \right\}

1876: =\frac{\beta_0 gn_{c0}}{\sigma_H \tau_{12}}=\frac{1}{\tau_{\mu}}.

1877: \end{equation}

1878: We thus see that $\tau$ is precisely the relaxation time $\tau_{\mu}$ first

1879: introduced in the ZGN$'$ two-fluid hydrodynamics.

1880: We can now express the four second viscosity coefficients in

1881: (\ref{zeta12}) and (\ref{zeta34}) in terms of the

1882: $\tau_{12}$ collision time defined in (\ref{tau12}):

1883: \begin{equation}

1884: \zeta_1=\frac{k_{\rm B}T}{3m}\sigma_H^2\tau_{12}, ~

1885: \zeta_2=\frac{n_ck_{\rm B}T}{9}\sigma_H^2\tau_{12}, ~

1886: \zeta_3=\frac{k_{\rm B}T}{m^2n_c}\sigma_H^2\tau_{12}, ~

1887: \zeta_4=\zeta_1.

1888: \label{zeta_final}

1889: \end{equation}

1890: 

1891: \subsection{Numerical results for a uniform Bose gas}

1892: For illustration, we calculate the transport coefficients in (\ref{kappa_final})

1893: and (\ref{etafinal}) for a {\it uniform} Bose gas  ($U_{\rm ext}=0$).

1894: As in Ref.~\cite{CJP}, we choose $gn/k_{\rm B}T_{\rm BEC}=0.2$.

1895: In Fig.1 and Fig.2, we plot the temperature dependence of the dimensionless

1896: transport coefficients $\bar\kappa$ and $\bar\eta$, defined by

1897: \begin{equation}

1898: \bar\kappa\equiv \kappa/nv_{\rm cl}^2\tau_0k_{\rm B},~~

1899: \bar\eta\equiv \eta/nv_{\rm cl}m\tau_0.

1900: \end{equation}

1901: Here $v_{\rm cl}\equiv (5k_{\rm B}T_{\rm BEC}/3m)^{1/2}$ is the hydrodynamic

1902: sound velocity of a classical gas at $T=T_{\rm BEC}$

1903: and $\tau_0^{-1}\equiv \sqrt{2}(8\pi a^2) n (8k_{\rm B}T_{\rm BEC}/\pi m)^{1/2}$

1904: is the classical mean collision time evaluated at $T=T_{\rm BEC}$.

1905: To see the separate effects of the $C_{22}$ collisions and $C_{12}$ collisions,

1906: we also plot the results obtained by taking either $I_{12}^{\kappa,\eta}=0$

1907: or $I_{22}^{\kappa,\eta}=0$.

1908: In Ref.~\cite{CJP}, we neglected $C_{12}$.

1909: We see that the both $\kappa$ and $\eta$ are reduced when we include the $C_{12}$

1910: collisions.

1911: At lower temperatures $T\lesssim 0.5T_{\rm BEC}$, both $\kappa$ and $\eta$ are

1912: dominated by the $C_{12}$ collision integral.

1913: 

1914: In Fig.~3, we plot the four second viscosity coefficients given in (\ref{zeta_final})

1915: for a uniform Bose gas.

1916: We recall that in Fig.1 of Ref.~\cite{NZG}, we gave the temperature dependence

1917: of $\tau_{\mu}$ and $\tau_{12}$ for $gn/k_{\rm B}T_{\rm BEC}=0.1$.

1918: In Fig.~3, we use the dimensionless second viscosity coefficients, defined by

1919: \begin{equation}

1920: \bar \zeta_1 \equiv \zeta_1/v_{\rm cl}^2\tau_0, ~

1921: \bar \zeta_2 \equiv \zeta_2m/nv_{\rm cl}^2\tau_0, ~

1922: \bar \zeta_3 \equiv \zeta_2n/mv_{\rm cl}^2\tau_0.

1923: \end{equation}

1924: 

1925: The transport coefficients in a {\it trapped} Bose gas behave quite differently from 

1926: those of a uniform Bose gas.

1927: In particular, since $n_c\gg \tilde n$ holds in the central region of the trap at all

1928: temperatures, the contribution of $C_{12}$ collisions dominates over the 

1929: contribution of the $C_{22}$ collisions at all temperatures below $T_{\rm BEC}$.

1930: We will discuss the implications of this at the end of Section \ref{sec:conclusions}.

1931: 

1932: \section{conclusions}

1933: \label{sec:conclusions}

1934: In this paper, we have derived two-fluid hydrodynamic equations starting

1935: from the quantum kinetic equation and the generalized GP equation derived in

1936: \cite{ZNG,NZG,milena2}.

1937: However, to complement and extend our earlier work~\cite{ZNG,NZG},

1938: we started from the complete local equilibrium single-particle distribution $f^{(0)}$

1939: as given by (\ref{f0}) and (\ref{muc0}).

1940: Using the Chapman-Enskog approach,

1941: we then included the effects of a small deviation from this local equilibrium form.

1942: This deviation from local equilibrium within the thermal cloud

1943: brings in the usual kind of hydrodynamic

1944: damping due to the thermal conductivity and shear viscosity of the  thermal cloud. 

1945: A summary of our major results is given in the final paragraph of Section \ref{sec:intro}.

1946: 

1947: In addition,

1948: we showed that the there is additional dissipation associated with the collisional

1949: exchange of atoms between the condensate and non-condensate components.

1950: When we write our hydrodynamic equations in the Landau-Khalatnikov ~\cite{Khal,HM}

1951: form given by (\ref{landau_eqs}), this damping is described in terms of the usual

1952: four second viscosity coefficients for a Bose superfluid.

1953: The appearance of the second viscosity coefficients in the equations for

1954: the total current ${\bf j}$ in (\ref{eq_j}) and for the superfluid velocity ${\bf v}_c$ 

1955: in (\ref{eq_vs}) is due to the deviation of the total pressure and the chemical potential from 

1956: their local equilibrium values.

1957: We might also recall that

1958: Khalatnikov \cite{Khal} discusses a specific model for 

1959: the second viscosity coefficients in superfluid $^4$He by introducing 

1960: ``local chemical potentials" for the phonons ($\mu_{\rm ph}$) and rotons ($\mu_{\rm r}$).

1961: These describe a situation where such excitations (describing the normal fluid)

1962: are out of local equilibrium with the superfluid component.

1963: Clearly this discussion has connections with our calculations based on the condensate

1964: and non-condensate not being in diffusive equilibrium.

1965: 

1966: The frequency-dependence of the second viscosity coefficients is a result of the

1967: fact that our two-fluid hydrodynamics deals with the dynamics of the condensate

1968: and non-condensate components as {\it separate} degrees of freedom.

1969: This feature is made more explicit in our recent papers \cite{ZNG,NZG,CJP}.

1970: In particular, it gives rise to a new relaxational zero frequency mode.

1971: As mentioned at the end of Section~\ref{zgn'}, and more explicitly in Ref.~\cite{CJP},

1972: this mode may be viewed as the (renormalized) version of the zero-frequency thermal

1973: diffusion mode \cite{huang} above $T_{\rm BEC}$.

1974: The presence of this new mode below $T_{\rm BEC}$ is somewhat hidden in the formulation 

1975: in terms of the LK two-fluid equations given in (\ref{landau_eqs}).

1976: 

1977: In Section \ref{sec:transport},

1978: we derived explicit formulas for all the transport coefficients within our model.

1979: In Ref.~\cite{CJP}, we only took

1980: into account the deviations from local equilibrium due to the 

1981: $C_{22}$ collision integral 

1982: in calculating the shear viscosity and the thermal conductivity coefficients.

1983: In the present paper, we have also included the contribution to these quantities 

1984: from the $C_{12}$ collision integral.

1985: From (\ref{kappa_final2}) and (\ref{etafinal2}), we see that both

1986: $\kappa$ and $\eta$ are given in a form proportional to characteristic relaxation times

1987: $\tau_{\kappa}=(\tau_{\kappa,12}^{-1}+\tau_{\kappa,22}^{-1})^{-1}$ and

1988: $\tau_{\eta}=(\tau_{\eta,12}^{-1}+\tau_{\eta,22}^{-1})^{-1}$, respectively,

1989: which are defined and motivated in Appendix \ref{relax_times}.

1990: 

1991: In a rough estimate, we find $\tau_{\kappa,22},\tau_{\eta,22} \sim \tau_{\rm cl}

1992: \sim 1/\tilde n$

1993: and $\tau_{\kappa,12},\tau_{\eta,12}\sim (\tilde n/ n_c) \tau_{\rm cl}

1994: \sim 1/n_c$,

1995: where $\tau_{\rm cl}$ is the classical collision time defined in

1996: (\ref{taucl}).

1997: We therefore observe that the effect of $C_{12}$ collisions reduces the magnitude

1998: of both $\kappa$ and $\eta$ by a factor $\sim 1/(1+n_c/\tilde n)$,  a result also

1999: noted in Ref.~\cite{KD} for a {\it uniform} gas.

2000: The contribution of the $C_{12}$ collisions is always important in a trapped gas,

2001: since the condensate density at the central region of 

2002: a trap is much larger than the density of the thermal cloud even at temperatures

2003: very close to $T_{\rm BEC}$.

2004: In a trapped Bose gas, we find the effect of the $C_{12}$ collisions is enhanced by

2005: a large factor $n_c/\tilde n \gg 1$.

2006: This means that $\kappa$ and $\eta$ are always dominated by the contribution of the

2007: $C_{12}$ collisions.

2008: Since the  effective relaxation times $\tau_{\kappa,12}$ and

2009: $\tau_{\eta,12}$ are smaller than the classical collision time $\tau_{\rm cl}$

2010: by a factor $\tilde n/n_c\ll 1$, this

2011: implies that in a trapped Bose gas, the hydrodynamic region may be {\it much easier}

2012: to reach at finite temperatures than expected

2013: from naive considerations based on using the classical collision time

2014: (i.e., $\omega\tau_{\rm cl}\ll 1$).

2015: That is to say, one might easily have $\omega\tau_{\kappa,12}\ll 1$ and

2016: $\omega\tau_{\eta,12}\ll 1$, even though $\omega\tau_{\rm cl}\gg 1$.

2017: This has very important implications in deciding if one is in the

2018: collisionless or the hydrodynamic region.

2019: 

2020: One problem not dealt with in this paper is the fact that in a trapped Bose gas,

2021: the decreasing density in the tail of the thermal cloud means that the hydrodynamic

2022: description breakdowns eventually.

2023: This problem enters the evaluation of the expressions for the

2024: $\eta$ and $\kappa$ transport coefficients given in Section \ref{sec:transport}.

2025: In recent papers dealing with the case above $T_{\rm BEC}$ \cite{KPS,NG},

2026: this problem was handled in an ad-hoc manner by introducing a spatial cutoff in the 

2027: thermal cloud.

2028: In a future paper, we given an alternative approach based on starting with an

2029: improved solution of the kinetic equation, which naturally includes the cross-over to the

2030: non-interacting gas in the thermal gas tail.

2031: 

2032: \begin{center}

2033: {\bf ACKNOWLEDGMENTS}

2034: \end{center}

2035: We thank E. Zaremba for his interest and comments and G. Shlyapnikov for a

2036: useful discussion.

2037: T.N. was supported by JSPS of Japan, while A.G. was supported by NSERC

2038: of Canada.

2039: 

2040: \begin{appendix}

2041: \section{}

2042: 

2043: We briefly sketch the derivation of the first-order kinetic equation given in (\ref{linpsi}).		Using (\ref{f0}) in (\ref{eq_f1}), one has

2044: \begin{eqnarray}

2045: &&\left[ \frac{\partial^0}{\partial t} +\frac{{\bf p}}{m}\cdot

2046: \bbox{\nabla}_{{\bf r}} -\bbox{\nabla}_{{\bf r}} U({\bf r},t)\cdot\bbox{\nabla}_{\bf p}\right]

2047: f^{(0)}({\bf r},{\bf p},t) \nonumber \\

2048: &&=\left[\frac{1}{z}\left(\frac{\partial^0}{\partial t}+

2049: \frac{{\bf p}}{m}\cdot\bbox{\nabla} \right)z

2050: +\frac{mu^2}{2k_{\rm B}T^2}

2051: \left(\frac{\partial^0}{\partial t}+

2052: \frac{{\bf p}}{m}\cdot\bbox{\nabla} \right)T \right. \nonumber \\

2053: &&\ \ \ \  \left. +\frac{m{\bf u}}{k_{\rm B}T}\cdot

2054: \left(\frac{\partial^0}{\partial t}+

2055: \frac{{\bf p}}{m}\cdot\bbox{\nabla} \right){\bf v}_n

2056: +\frac{\bbox{\nabla}U({\bf r},t)}{k_{\rm B}T}\cdot {\bf u}

2057: \right] f^{(0)}(1+f^{(0)}).

2058: \label{A1}

2059: \end{eqnarray}

2060: The notation $\partial^0/\partial t$ is explained below (\ref{eq_f1}).

2061: Using the expressions for the density $\tilde n^{(0)}$

2062: given by (\ref{ntilde0})

2063: and the pressure $\tilde P^{(0)}$ in (\ref{ptilde0}), one finds

2064: \begin{eqnarray}

2065: \frac{\partial^0\tilde n^{(0)}}{\partial t}

2066: &=&\frac{3\tilde n^{(0)}}{2 T}\frac{\partial^0 T}{\partial t}

2067: +\frac{\gamma k_{\rm B}T}{z}\frac{\partial^0 z}{\partial t} \,, \cr

2068: && \cr

2069: \frac{\partial^0 \tilde P^{(0)}}{\partial t}

2070: &=&\frac{5\tilde P^{(0)}}{2T}\frac{\partial^0 T}{\partial t}

2071: +\frac{\tilde n^{(0)}k_{\rm B}T}{z}\frac{\partial^0 z}{\partial t}\,,

2072: \end{eqnarray}

2073: where $\gamma$ is the variable introduced after (\ref{sigma12})

2074: and $z=z^{(0)}$ as defined below (\ref{ntilde0}).

2075: One may combine these equations with (\ref{eq_ntilde0}) and (\ref{eq_ptilde0})

2076: to show that the equations in (A2) reduce to

2077: \begin{eqnarray}

2078: \frac{\partial^0 T}{\partial t}&=&-\frac{2}{3}T(\bbox{\nabla}\cdot{\bf v}_n)

2079: -{\bf v}_n\cdot\bbox{\nabla}T+\frac{2T}{3\tilde n^{(0)}}\sigma_1 \Gamma_{12}, \cr

2080: && \cr

2081: \frac{\partial^0 z}{\partial t}&=&-{\bf v}_n\cdot\bbox{\nabla}z+

2082: \sigma_2z\frac{\Gamma_{12}}{\tilde n^{(0)}}.

2083: \end{eqnarray}

2084: The analogous equation for $\partial^0{\bf v}_n/\partial t$ is given directly

2085: by (\ref{eq_vn0}).

2086: Using these results in (A1), one finds that it reduces to

2087: \begin{eqnarray}

2088: &&\left(\frac{\partial^0}{\partial t}

2089: +\frac{{\bf p}}{m}\cdot\bbox{\nabla}_{{\bf r}}-\bbox{\nabla}_{{\bf r}}U\cdot\bbox{\nabla}_{\bf p}

2090: \right)f^{(0)} \cr

2091: &&=\left\{

2092: \frac{1}{T}{\bf u}\cdot\bbox{\nabla}T\left(\frac{mu^2}{2k_{\rm B}T}

2093: -\frac{5\tilde P^{(0)}}{2\tilde n^{(0)}k_{\rm B}T} \right)

2094: +\frac{m}{k_{\rm B}T}\left[{\bf u}\cdot({\bf u}\cdot\bbox{\nabla}){\bf v}_n

2095: -\frac{u^2}{3}\bbox{\nabla}\cdot{\bf v}_n\right]\right. \cr

2096: &&+\left.\left[\sigma_2+\frac{mu^2}{3k_{\rm B}T}\sigma_1+\frac{m}{k_{\rm B}T}

2097: {\bf u}\cdot({\bf v}_c-{\bf v}_n) \right]

2098: \frac{\Gamma_{12}}{\tilde n^{(0)}}\right\}f^{(0)}

2099: (1+f^{(0)}),

2100: \label{lhs_f1}

2101: \end{eqnarray}

2102: where we recall ${\bf u}\equiv {\bf p}/m-{\bf v}_n$.

2103: In calculating the dissipative terms, we only consider terms to

2104: first order in the velocity fields ${\bf v}_n$ and ${\bf v}_c$.

2105: Since $\Gamma_{12}$ is proportional to ${\bf v}_n$ and ${\bf v}_c$

2106: (see (\ref{gamma1})),

2107: we can neglect the last term (proportional to ${\bf v}_c-{\bf v}_n$) in (\ref{lhs_f1}).

2108: This linearized version of (\ref{lhs_f1}) 

2109: can be rewritten in the form shown on the left hand side of (\ref{linpsi}).

2110: 

2111: \section{}

2112: \label{relax_times}

2113: The relaxation time $\tau_{\kappa}$ in (\ref{kappa_final2}) is defined by

2114: \begin{eqnarray}

2115: \tau_{\kappa}&\equiv& -

2116: \left\{\int \frac{d{\bf p}}{(2\pi\hbar)^3}

2117: \left[\frac{mu^2}{2k_{\rm B}T_0}

2118: -\frac{5g_{5/2}(z_0)}{2g_{3/2}(z_0)}\right]^2u^2

2119: f_0(1+f_0) \right\} \cr

2120: && \times \left\{\int \frac{d{\bf p}}{(2\pi\hbar)^3} 

2121: \left[\frac{mu^2}{2k_{\rm B}T_0}

2122: -\frac{5g_{5/2}(z_0)}{2g_{3/2}(z_0)}\right]{\bf u} \cdot

2123: \hat L \left[\left\{\frac{mu^2}{2k_{\rm B}T_0}

2124: -\frac{5g_{5/2}(z_0)}{2g_{3/2}(z_0)} \right\}{\bf u}\right]\right\}^{-1} \cr

2125: ~~&& \cr 

2126: &=& \frac{ 15\pi^{9/2}\hbar^3 }{ 8m(k_{\rm B}T_0)^2a^2 }

2127: \left[\frac{ \frac{7}{2}g_{7/2}(z_0)-\frac{5}{2}g_{5/2}^2(z_0)/g_{3/2}(z_0) }

2128: {I_{22}^{\kappa}(z_0)+n_{c0}\Lambda_0^3 I_{12}^{\kappa}(z_0) } \right]\cr

2129: ~~&& \cr

2130: &=&

2131: \frac{15\sqrt{2}\pi^{7/2}}{4}\tau_{\rm cl}

2132: \left[\frac{ \frac{7}{2}g_{7/2}(z_0)g_{3/2}(z_0)-\frac{5}{2}g_{5/2}^2(z_0) }

2133: {I_{22}^{\kappa}(z_0)+n_{c0}\Lambda_0^3 I_{12}^{\kappa}(z_0) }\right].

2134: \label{tau_k}

2135: \end{eqnarray}

2136: Here 

2137: \begin{equation}

2138: \tau_{\rm cl}^{-1}\equiv \sqrt{2}(8\pi a^2) \tilde n_0({\bf r})

2139:  (8k_{\rm B}T_0/\pi m)^{1/2},

2140: \label{taucl}

2141: \end{equation}

2142: is the collision time of a classical gas with density $\tilde n_0$ and quantum 

2143: cross section $\sigma=8\pi a^2$.

2144: In turn, the relaxation times in (\ref{tau_k2}) are defined by

2145: \begin{eqnarray}

2146: \tau_{\kappa,22}&\equiv& - \left\{\int \frac{d{\bf p}}{(2\pi\hbar)^3}

2147: \left[\frac{mu^2}{2k_{\rm B}T_0}-\frac{5g_{5/2}(z_0)}{2g_{3/2}(z_0)}\right]^2u^2

2148: f_0(1+f_0) \right\}\cr

2149: && \times \left\{\int \frac{d{\bf p}}{(2\pi\hbar)^3} 

2150: \left[\frac{mu^2}{2k_{\rm B}T_0}

2151: -\frac{5g_{5/2}(z_0)}{2g_{3/2}(z_0)}\right]{\bf u} \cdot

2152: \hat L_{22} \left[\left\{\frac{mu^2}{2k_{\rm B}T_0}

2153: -\frac{5g_{5/2}(z_0)}{2g_{3/2}(z_0)} \right\}{\bf u}\right] \right\}^{-1} \cr

2154: ~~&& \cr 

2155: &=& \frac{ 15\pi^{9/2}\hbar^3 }{ 8m(k_{\rm B}T_0)^2a^2 }

2156: \left[\frac{ \frac{7}{2}g_{7/2}(z_0)-

2157: \frac{5}{2}g_{5/2}^2(z_0)/

2158: g_{3/2}(z_0) }{I_{22}^{\kappa}(z_0)} \right]\cr

2159: ~~&& \cr

2160: &=&\frac{15\sqrt{2}\pi^{7/2}}{4}\tau_{\rm cl}

2161: \left[\frac{ \frac{7}{2}g_{7/2}(z_0)g_{3/2}(z_0)-

2162: \frac{5}{2}g_{5/2}^2(z_0)}

2163: {I_{22}^{\kappa}(z_0)}\right],

2164: \end{eqnarray}

2165: and

2166: \begin{eqnarray}

2167: \tau_{\kappa,12}&\equiv&

2168:  - \left\{\int \frac{d{\bf p}}{(2\pi\hbar)^3}

2169: \left[\frac{mu^2}{2k_{\rm B}T_0}-\frac{5g_{5/2}(z_0)}{2g_{3/2}(z_0)}\right]^2u^2

2170: f_0(1+f_0) \right\}

2171: \cr &&

2172:  \times \left\{ \int \frac{d{\bf p}}{(2\pi\hbar)^3} 

2173: \left[\frac{mu^2}{2k_{\rm B}T_0}

2174: -\frac{5g_{5/2}(z_0)}{2g_{3/2}(z_0)}\right]{\bf u} \cdot

2175: \hat L_{12} \left[\left\{\frac{mu^2}{2k_{\rm B}T_0}

2176: -\frac{5g_{5/2}(z_0)}{2g_{3/2}(z_0)} \right\}{\bf u}\right] \right\}^{-1} \cr

2177: ~~&&

2178:  \cr 

2179: &=& \frac{ 15\pi^{9/2}\hbar^3 }{ 8m(k_{\rm B}T_0)^2a^2 }

2180: \left[\frac{ \frac{7}{2}g_{7/2}(z_0)-

2181: \frac{5}{2}g_{5/2}^2(z_0)/

2182: g_{3/2}(z_0) }

2183: {n_{c0}\Lambda_0^3 I_{12}^{\kappa}(z_0) } \right]\cr

2184: ~~&& \cr

2185: &=&

2186: \frac{15\sqrt{2}\pi^{7/2}} {4} \tau_{\rm cl}

2187: \left[ \frac{ \frac{7}{2}g_{7/2}(z_0)g_{3/2}(z_0)-\frac{5}{2}g_{5/2}^2(z_0)}

2188: {n_{c0}\Lambda_0^3 I_{12}^{\kappa}(z_0) } \right].

2189: \end{eqnarray}

2190: 

2191: In an analogous way, the

2192: relaxation time $\tau_{\eta}$ in (\ref{etafinal2}) is defined by

2193: \begin{eqnarray}

2194: \tau_{\eta}&\equiv&

2195: -\left[\int \frac{d{\bf p}}{(2\pi\hbar)^3}

2196: \left(u_{\mu}u_{\nu}-\frac{1}{3}\delta_{\mu\nu}u^2\right)^2

2197: f_0(1+f_0) \right]\cr

2198: &&\times\left\{

2199: \int \frac{d{\bf p}}{(2\pi\hbar)^3}\left(

2200: u_{\mu}u_{\nu}-\frac{1}{3}\delta_{\mu\nu}u^2

2201: \right) \hat L\left[u_{\mu}u_{\nu}-\frac{1}{3}\delta_{\mu\nu}u^2\right]

2202: \right\}^{-1} \cr

2203: ~~&& \cr

2204: &=&\frac{5\pi^{9/2}\hbar^3}{16m(k_{\rm B}T_0)^2a^2}

2205: \left[\frac{ g_{5/2}(z_0) }{I_{22}^{\eta}(z_0)+n_{c0}\Lambda_0^3 I_{12}^{\eta}(z_0)}\right]

2206: \cr ~~&& \cr

2207: &=&\frac{5\sqrt{2}\pi^{7/2}}{2}\tau_{\rm cl}

2208: \left[\frac{g_{5/2}(z_0)g_{3/2}(z_0)}

2209: {I_{22}^{\eta}(z_0)+n_{c0}\Lambda_0^3I_{12}^{\eta}(z_0)} \right].

2210: \label{tau_e}

2211: \end{eqnarray}

2212: The relaxation times in (\ref{tau_e2}) are defined by

2213: \begin{eqnarray}

2214: \tau_{\eta,22}&\equiv&

2215: -\left[\int \frac{d{\bf p}}{(2\pi\hbar)^3}

2216: \left(u_{\mu}u_{\nu}-\frac{1}{3}\delta_{\mu\nu}u^2\right)^2

2217: f_0(1+f_0) \right]\cr

2218: &&\times\left\{

2219: \int \frac{d{\bf p}}{(2\pi\hbar)^3}\left(

2220: u_{\mu}u_{\nu}-\frac{1}{3}\delta_{\mu\nu}u^3

2221: \right) \hat L_{22}\left[u_{\mu}u_{\nu}-\frac{1}{3}\delta_{\mu\nu}u^3\right]

2222: \right\}^{-1} \cr ~~ && \cr

2223: &=&\frac{5\pi^{9/2}\hbar^3}{16m(k_{\rm B}T_0)^2a^2}

2224: \left[\frac{ g_{5/2}(z_0) }{I_{22}^{\eta}(z_0)}\right] \cr ~~&& \cr

2225: &=&\frac{5\sqrt{2}\pi^{7/2}}{2}\tau_{\rm cl}

2226: \left[\frac{g_{5/2}(z_0)g_{3/2}(z_0)}

2227: {I_{22}^{\eta}(z_0)} \right],

2228: \end{eqnarray}

2229: and

2230: \begin{eqnarray}

2231: \tau_{\eta,12}&\equiv&

2232: -\left[\int \frac{d{\bf p}}{(2\pi\hbar)^3}

2233: \left(u_{\mu}u_{\nu}-\frac{1}{3}\delta_{\mu\nu}u^2\right)^2

2234: f_0(1+f_0) \right]\cr

2235: &&\times\left\{

2236: \int \frac{d{\bf p}}{(2\pi\hbar)^3}\left(

2237: u_{\mu}u_{\nu}-\frac{1}{3}\delta_{\mu\nu}u^2

2238: \right) \hat L_{12}\left[u_{\mu}u_{\nu}-\frac{1}{3}\delta_{\mu\nu}u^2\right]

2239: \right\}^{-1} \cr ~~ && \cr

2240: &=&\frac{5\pi^{9/2}\hbar^3}{16m(k_{\rm B}T_0)^2a^2}

2241: \left[\frac{ g_{5/2}(z_0) }{n_{c0}\Lambda_0^3 I_{12}^{\eta}(z_0)}\right] \cr 

2242: ~~&& \cr

2243: &=&\frac{5\sqrt{2}\pi^{7/2}}{2}\tau_{\rm cl}\left[

2244: \frac{g_{5/2}(z_0)g_{3/2}(z_0)}{n_{c0}\Lambda_0^3I_{12}^{\eta}(z_0)}

2245: \right].

2246: \end{eqnarray}

2247: 

2248: We note that in a non-degenerate gas, these expressions simplify and we find

2249: $\tau_{\kappa}=\tau_{\kappa,22}=\frac{15}{8}\tau_{\rm cl}(n_0({\bf r}))$

2250: and

2251: $\tau_{\eta}=\tau_{\eta,22}=\frac{5}{4}\tau_{\rm cl}(n_0({\bf r}))$.

2252: The latter expression for $\tau_{\eta}$ agree with the result for the

2253: shear viscosity given in Eq.~(11) by Kavoulakis et al. \cite{KPS2}.

2254: 

2255: The physical meaning of these new relaxation times becomes clear when we compare

2256: our Chapman-Enskog analysis in the text with a simple relaxation time approximation \cite{huang}.

2257: In the relaxation time approximation, one simply replaces the collision term in (\ref{eq1}) with

2258: $-[f-f^{(0)}]/\tau_{\rm rel}$, where $\tau_{\rm rel}$ is a phenomenological

2259: relaxation time characterizing how fast the system relaxes to local equilibrium.

2260: With this approximation, the solutions of the linearized equations in

2261: (\ref{eqforA}) and (\ref{eqforB}) for the functions $A(u)$ and $B(u)$

2262: are found to be simply given by

2263: \begin{equation}

2264: A(u)=-\tau_{\rm rel}\left[\frac{mu^2}{2k_{\rm B}T}-\frac{5g_{5/2}(z)}

2265: {2g_{3/2}(z)}\right], ~~

2266: B(u)=-\tau_{\rm rel}\frac{m}{k_{\rm B}T}.

2267: \label{relax_approx}

2268: \end{equation}

2269: In contrast, our Chapman-Enskog solution for $A(u)$ is given by (\ref{ansatzA})

2270: with the coefficient $A$ given by (\ref{eqAconst}), while one has

2271: $B(u)=B$ with the constant $B$ given by (\ref{eqBconst}).

2272: In terms the relaxation times $\tau_{\kappa}$ and $\tau_{\eta}$ defined above, 

2273: we find (\ref{ansatzA}) and (\ref{eqBconst}) can be written as

2274: \begin{equation}

2275: A(u)=-\tau_{\kappa}\left[\frac{mu^2}{2k_{\rm B}T}-\frac{5g_{5/2}(z)}

2276: {2g_{3/2}(z)}\right], ~~

2277: B(u)=-\tau_{\eta}\frac{m}{k_{\rm B}T}.

2278: \label{relax_approx2}

2279: \end{equation}

2280: Comparing (\ref{relax_approx2}) with (\ref{relax_approx}), we see that

2281: both $\tau_{\kappa}$ and $\tau_{\eta}$ can be identified with the relaxation time

2282: $\tau_{\rm rel}$.

2283: That is, in the simple relaxation time approximation, $\kappa$ and $\eta$

2284: are still given by the formulas (\ref{kappa_final2}) and (\ref{etafinal2}),

2285: but with $\tau_{\kappa}=\tau_{\eta}=\tau_{\rm rel}$.

2286: We also note that Eq.~(7) of Ref.~\cite{KPS2} gives a general expression for various 

2287: collisional relaxation times above $T_{\rm BEC}$, which is given by 

2288: a formula analogous to (\ref{tau_k}) and (\ref{tau_e}).

2289: 

2290: It is these effective relaxation times that determine whether one is in the hydrodynamic

2291: two-fluid domain, $\omega\tau_{\kappa}\lesssim 1$, $\omega\tau_{\eta}\lesssim 1$

2292: (see discussion in Section \ref{sec:conclusions}).

2293: They should be used in place of the classical collision time $\tau_{\rm cl}$ in

2294: (\ref{taucl}). 

2295: 

2296: \section{}

2297: 

2298: Hohenberg and Martin~\cite{HM} worked out the dispersion relation of the

2299: hydrodynamic modes in a

2300: a {\it uniform} Bose superfluid using the Landau-Khalatnikov two-fluid equations.

2301: The frequencies of the first and second sound modes are 

2302: given by

2303: \begin{equation}

2304: \omega^2=u_i^2k^2-iD_ik^2\omega,

2305: \end{equation}

2306: where the sound velocities $u_i$ are determined by the coupled equations

2307: \begin{eqnarray}

2308: u_1^2+u_2^2&=&\frac{T\rho_s {\bar s}^2}{\rho_n {\bar c_v}}+\left.\frac{\partial P}

2309: {\partial {\bar s}}\right|_{\bar{s}}, \\

2310: u_1^2u_2^2&=&\frac{T\rho_s{\bar s}^2}{\rho_n\bar{c_v}}\left.

2311: \frac{\partial P}{\partial \rho}\right|_{T}.

2312: \end{eqnarray}

2313: Here $\bar s\equiv s/mn$ is the entropy per unit mass and $\bar c_v$ is the

2314: specific heat per unit mass.

2315: The damping coefficients $D_i$ are determined by the coupled equations

2316: \begin{eqnarray}

2317: D_1+D_2&=&\frac{4\eta}{3\rho_n}+\frac{\zeta_2}{\rho_n}-\frac{\rho_s}{\rho_n}

2318: (\zeta_1+\zeta_4)+\frac{\zeta_3\rho_s}{\rho_n}\rho+\frac{\kappa}{\rho {\bar c_v}}, \\

2319: u_1^2D_1+u_2^2D_2&=&\frac{\partial P}{\partial \rho}+\frac{\zeta_2+4\eta /3 }{\rho}

2320: \frac{\rho_s}{\rho_n}\left(\frac{T{\bar s}^2}{\bar c_v}-\frac{2T{\bar s}}

2321: {\rho {\bar c_v}} \left.\frac{\partial P}{\partial T}\right|_{\rho}

2322: +\left.\frac{\partial P}{\partial \rho} \right|_{\bar s}\right)

2323: +\zeta_3\frac{\rho_s}{\rho_n}\left. \frac{\partial P}{\partial \rho}\right|_{\bar s}\rho 

2324: \\

2325: &&-(\zeta_1+\zeta_4)\frac{\rho_s}{\rho_n}\left(-\frac{T{\bar s}}{{\bar c_v} \rho}

2326: \left.\frac{\partial P}{\partial T}\right|_{\rho}+\left.\frac{\partial P}{\partial\rho}

2327: \right|_{\bar s}\right).

2328: \end{eqnarray}

2329: 

2330: We note that the above general expressions are valid for both liquid $^4$He and

2331: Bose gases.

2332: For the liquid $^4$He, these formulas can be simplified by using ${\bar c_v}\approx 

2333: {\bar c_p}$. 

2334: One has

2335: \begin{eqnarray}

2336: u_1^2&\approx&\frac{\partial P}{\partial \rho}  \\

2337: u_2^2&\approx&-\frac{\rho_s}{\rho_n}\frac{\partial T}{\partial (1/{\bar s})}, \\

2338: D_1&\approx&\frac{\zeta_2+4\eta/3}{\rho}, \\

2339: D_2&\approx&\frac{1}{\rho}

2340: \left\{ \frac{\kappa}{ T(\partial{\bar s}/\partial T)}+\frac{\rho_s}{\rho_n}

2341: \left[\zeta_3\rho^2-\rho(\zeta_1+\zeta_4)+\zeta_2+\frac{4}{3}\eta\right]\right\}.

2342: \end{eqnarray}

2343: However, these are not valid for a dilute Bose gas.

2344: Calculating the various thermodynamic derivatives with the

2345: Hartree-Fock single-particle spectrum used in this paper~\cite{GZ},

2346: the sound velocities and damping coefficients for a uniform Bose gas are given by

2347: \begin{eqnarray}

2348: u_1^2&\approx& \frac{5\tilde P_0}{3m\tilde n_0}+\frac{2g\tilde n_0}{m}

2349: \left(1-\frac{2n_{c0}^2}{9\tilde n_0^2} \sigma_H \right) \\

2350: u_2^2&\approx& \frac{gn_{c0}}{m}(1-\sigma_H)\\

2351: D_1&\approx& \frac{4\eta}{3\rho_n}+\frac{\zeta_2}{\rho_n}

2352: -\frac{\rho_s}{\rho_n}

2353: (\zeta_1+\zeta_4)+\frac{\zeta_3\rho_s^2}{\rho_n}

2354: +\frac{4}{9}\frac{\kappa T_0}{\rho_n u_1^2}\left(1+\frac{2n_{c0}}{3\tilde n_0}

2355: \sigma_1\sigma_H \right)^2, 

2356: \\

2357: D_2&\approx&\rho_s\zeta_3+\frac{4}{9}\frac{\rho_s}{\rho_n^2 u_2^2}(\sigma_1\sigma_H)^2

2358: \kappa T_0 .

2359: \end{eqnarray} 

2360: Here $\sigma_H$ is defined in (\ref{sigma_H}), while $\rho_n=m\tilde n_0$ and $\rho_s=mn_{c0}$.

2361: 

2362: \end{appendix}

2363: 

2364: 

2365: \begin{references}

2366: 

2367: \bibitem{RMP} F. Dalfovo, S. Giorgini, L.P. Pitaevskii and S. Stringari,

2368: Rev. Mod. Phys. {\bf 71}, 463 (1999).

2369: 

2370: \bibitem{varenna} {\it Bose-Einstein Condensation in Atomic Gases},

2371: Proceedings of the International School of Physics ``Enrico Fermi'' ,

2372: ed. by M. Inguscio, S. Stringari and C.E. Wieman (IOS Press, Amsterdam, 1999).

2373: 

2374: \bibitem{ZNG}E. Zaremba, T. Nikuni and A. Griffin,

2375: J. Low Temp. Phys. {\bf 116}, 277 (1999).

2376: 

2377: \bibitem{NP} P. Nozi\`eres and D. Pines, {\it The theory of Quantum Liquids}

2378: (Addison-Wesley, Redwood City, California, 1990), Vol. II.

2379: 

2380: \bibitem{griffin} A. Griffin, in ``{\it The 13th Australian Physics Summer

2381: School: Bose-Einstein Condensation}" ed. by C.M. Savage and M. Das

2382: (World Scientific, Singapore, 2000), in press; cond-mat/0006382.

2383: 

2384: \bibitem{WG} J.E. Williams and A. Griffin, Phys. Rev. A, in press (2000);

2385: cond-mat/003481. 

2386: 

2387: \bibitem{ZGN}E. Zaremba, A. Griffin and T. Nikuni,

2388: Phys. Rev. {\bf A57}, 4695 (1998).

2389: 

2390: \bibitem{85Rb} S.L. Cornish, N.R. Claussen, J.L. Roberts, E.A. Cornell

2391: and C.E. Wieman, Phys. Rev. Lett. {\bf 85} 1795 (2000).

2392: 

2393: \bibitem{NZG}T. Nikuni, E. Zaremba and A. Griffin,

2394: Phys. Rev. Lett. {\bf 83}, 10 (1999).

2395: 

2396: \bibitem{CJP}T. Nikuni, A. Griffin and E. Zaremba,

2397: Can. J. Phys. {\bf 78}, 415 (2000).

2398: 

2399: \bibitem{Landau} L.D. Landau, J. Phys. (U.S.S.R.) {\bf 5}, 71 (1941).

2400: 

2401: \bibitem{Khal} I.K. Khalatnikov, {\it An Introduction to the Theory of Superfluidity}

2402: (W. A. Benjamin, New York, 1965). For a useful summary of the Landau-Khalatnikov

2403: two-fluid hydrodynamics, see also J. Wilks, {\it The Properties of Liquid and Solid Helium}

2404: (Clarendon Press, Oxford, 1967), esp. Chapters 7 and 8. 

2405: 

2406: \bibitem{KD} T.R. Kirkpatrick and J.R. Dorfman, J. Low Temp. Phys.

2407: {\bf 58}, 304 (1985); {\bf 58}, 399 (1985). 

2408: 

2409: \bibitem{NG} T. Nikuni and A. Griffin, J. Low Temp. Phys.

2410: {\bf 111}, 793 (1998).

2411: 

2412: \bibitem{stoof} H.T.C. Stoof, J. Low Temp. Phys. {\bf 114}, 11 (1999).

2413: 

2414: \bibitem{milena1} M. Imamovic-Tomasovic and A. Griffin,

2415: Phys. Rev. A {\bf 60} 494 (1999)

2416: 

2417: \bibitem{milena2} M. Imamovic-Tomasovic and A. Griffin,

2418: {\it Progress in Nonequilibrium Green's Functions},

2419: Proceedings of the Conference on ``Kadanoff-Baym Equations:

2420: Progress and Perspectives for Many-body Physics'',

2421: ed. by Michael Bonitz (World Scientific, Singapore, 2000), p.404; cond-mat/9911402.

2422: 

2423: \bibitem{KB} L.P. Kadanoff and G. Baym, {\it Quantum Statistical Mechanics}

2424: (Addison-Wesley, 1989).

2425: 

2426: \bibitem{milena3} M. Imamovic-Tomasovic and A. Griffin, J. Low Temp. Phys.,

2427: submitted for publication.

2428: 

2429: \bibitem{huang} K. Huang, {\it Statistical Mechanics} (J. Wiley, New York, 1987),

2430: 2nd ed.

2431: 

2432: \bibitem{UU} E. A. Uehling and G.E. Uhlenbeck, Phys. Rev. {\bf 43}, 552 (1933).

2433: 

2434: \bibitem{MY} K. Miyake and K. Yamada, Prog. Theor. Phys. {\bf 56},

2435: 1689 (1976).

2436: 

2437: \bibitem{HM} P.C. Hohenberg and P.C. Martin, Ann. Phys. {\bf 34}, 291 (1965).

2438: 

2439: \bibitem{ferziger} For a detailed review of the Chapman-Enskog method as

2440: applied to classical gases, see J.H. Ferziger and H.G. Kaper,

2441: {\it Mathematical Theory of Transport Processes in Gases},

2442: North-Holland, London (1972).

2443: 

2444: \bibitem{LL} L.D. Landau and M.D. Lifshitz, {\it Fluid Mechanics},

2445: Pergamon, Oxford (1959), p.384

2446: 

2447: \bibitem{GZ} A. Griffin and E. Zaremba, Phys. Rev. A {\bf 56}, 4839 (1997).

2448: 

2449: \bibitem{KPS} G.M. Kavoulakis, C.J. Pethick and H. Smith, Phys. Rev. A {\bf 57}, 2938 (1998).

2450: 

2451: \bibitem{KPS2} G.M. Kavoulakis, C.J. Pethick and H. Smith, Phys. Rev. Lett. {\bf 81}, 4036 (1998).

2452: 

2453: \end{references}

2454: 

2455: 

2456: %\clearpage

2457: 

2458: \centerline{\bf FIGURE CAPTIONS}

2459: \begin{itemize}

2460: \item[FIG.1:] 

2461: Plot of the thermal conductivity $\kappa$ in a uniform gas for 

2462: $gn=0.2k_{\rm B}T_{\rm BEC}$ as a function of temperature.

2463: We also plot the results by taking $I_{12}^{\kappa}=0$ (dashed line)

2464: and $I_{22}^{\kappa}=0$ (dot-dashed line).

2465: 

2466: \item[FIG.2:] 

2467: Plot of the shear viscosity coefficient in a uniform gas for 

2468: $gn=0.2k_{\rm B}T_{\rm BEC}$ as a function of temperature.

2469: We also plot the results by taking $I_{12}^{\eta}=0$ (dashed line)

2470: and $I_{22}^{\eta}=0$ (dot-dashed line).

2471: 

2472: 

2473: \item[FIG.3:] 

2474: Plot of the second viscosity coefficients $\zeta_i$

2475: in a uniform gas for $gn=0.2k_{\rm B}T_{\rm BEC}$

2476: as a function of temperature.

2477: 

2478: 

2479: \end{itemize}

2480: \end{document} 

2481: