1: %\documentstyle[twocolumnsaps,prl,preprint,epsf,floats,psfig]{revtex}

2: \documentstyle[prl,aps,multicol,psfig,array]{revtex}

3: %\tightenlines

4: 

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

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

7: \newcommand{\bqa}{\begin{eqnarray}}

8: \newcommand{\eqa}{\end{eqnarray}}

9: 

10: 

11: \parindent=20pt

12: \parskip=2pt

13: \pagestyle{plain}

14: \font\tenrm=cmr10

15: \def\square{\vcenter{\vbox{\hrule height.4pt

16:           \hbox{\vrule width.4pt height8pt

17:           \kern8pt\vrule width.4pt}\hrule height.4pt}}}

18: \def\boxx{\square}

19: 

20: 

21: \begin{document}

22: 

23: \preprint{

24: \vbox{\halign{&##\hfil\cr

25: &ITF-UU-01/29\cr        & cond-mat/yymmnn \cr

26: %&\today\cr 

27: }}}

28: 

29: 

30: \title{Ground State Pressure and Energy Density of a %n Interacting

31:  Homogeneous Bose Gas in Two Dimensions}

32: 

33: \author{Jens O. Andersen}

34: \address{Institute for Theoretical Physics, University of Utrecht,\\

35:        Leuvenlaan 4, 3584 CE Utrecht, The Netherlands\\

36: (\today)}

37: 

38: 

39: \maketitle

40: 

41: \begin{abstract}

42: {

43: We consider an interacting homogeneous 

44: Bose gas at zero temperature in two spatial dimensions.  

45: The properties of the system can be calculated as an expansion in powers of

46: $g$, where $g$ is the coupling constant. 

47: We calculate the ground state

48: pressure and the ground state

49: energy density to second order in the quantum loop expansion.

50: The renormalization group

51: is used to sum up leading and subleading logarithms from all orders

52: in perturbation theory.

53: In the dilute limit, the renormalization group improved 

54: pressure and energy density are expansions in powers

55: of the $T$-matrix.

56: }

57: \end{abstract}

58: 

59: \begin{multicols}{2}

60: %\newpage

61: \section{Introduction}

62: The remarkable achievement of Bose-Einstein condensation (BEC) of alkali

63: atoms in harmonic traps~\cite{bec1,bec2,bec3} 

64: has created an enormous interest in the properties of dilute Bose gases

65: (For a review, see e.g. Ref.~\cite{string} and references therein).

66: 

67: The homogeneous Bose gas 

68: in three dimensions has been studied in

69: great detail over the past 50 years (For a review, see e.g.~\cite{grif}). 

70: At zero temperature,

71: the quantum loop expansion is essentially an expansion in powers

72: of $\sqrt{\rho a^3}$, where $a$ is the two-body $S$-wave scattering length

73: and $\rho$ is the density.

74: Lee and Yang~\cite{leeyang} 

75: were the first to calculate the leading

76: quantum correction to the energy density. 

77: Part of the second quantum correction to the energy density

78: was obtained by Wu, by 

79: Hugenholz and Pines, and by Sawada~\cite{to}.

80: Only recently a complete two-loop result has been 

81: obtained by Braaten and Nieto~\cite{eric}.

82: The result depends, in addition to the scattering length, also

83: on the scattering amplitude for $3\rightarrow 3$ scattering.

84: 

85: The homogeneous Bose gas in two dimensions has also been studied extensively.

86: The chemical potential and 

87: ground state energy density of a two-dimensional homogeneous Bose gas

88: were first calculated by Schick~\cite{schiff}.

89: By summing up the ladder diagrams contributing to the

90: chemical potential, he was able to show that, in the dilute limit, the 

91: relevant expansion parameter is not the coupling constant $g$, but

92: rather the effective interaction

93: $[\log(\rho a^2)]^{-1}$, where $\rho$ is the density and $a$

94: is the range of the interaction. The expansion parameter is the 

95: two-body $T$-matrix, and 

96: Schick determined the leading-order results in this expansion.

97: Corrections to these results have been considered by several 

98: authors~\cite{popov1,popov2,fis,rg2,fried,hines}.

99: Very recently, a formal proof of the result by Schick was given by

100: Lieb and Yngvason~\cite{lieb1,lieb2}. 

101: 

102: In the present paper, we reconsider the homogeneous Bose gas in two

103: dimensions at zero temperature. We calculate the

104: pressure and 

105: energy density of the ground state to second order in the quantum

106: loop expansion. We also apply the renormalization group to sum up leading

107: and subleading logarithms from all orders of perturbation theory.

108: In the dilute limit, the renormalization group improved 

109: pressure and energy density are essentially expansions in powers

110: of the $T$-matrix.

111: 

112: The outline of the paper is as follows. In Sec.~II, we 

113: briefly discuss the perturbative framework developed in Ref.~\cite{eric}

114: to calculate the ground state properties of a homogeneous Bose gas.

115: In Sec.~III, we calculate the ground state pressure

116: to two loop order.

117: In Sec.~IV, we 

118: calculate the ground state

119: energy density through two loops. % and we derive the scale anomaly. 

120: In Sec.~V, we apply the renormalization group to sum up leading

121: and subleading logarithms from all orders in perturbation theory.

122: Finally, we summarize our results in Sec.~VI.

123: Calculational details are included in an appendix.

124: 

125: \section{Perturbative Framework}

126: In this section, we discuss the perturbative framework set up in 

127: Ref.~\cite{eric} to calculate the

128: effects on the ground state

129: from quantum fluctuations around the mean field.

130: 

131: The action is

132: \bqa\nonumber

133: S&=&\int dt\;

134: \Bigg\{

135: \int d^2x\;

136: \psi^*({\bf x},t)

137: \!\!\left[

138: i\hbar{\partial\over\partial t}

139: +{\hbar^2\over2m}\nabla^2+\mu

140: \right]

141: \psi({\bf x},t)

142: \\ \nonumber

143: &&

144: \hspace{-.2cm}

145: -{1\over2}\int d^2x\int d^2x^{\prime}\;

146: \Bigg[

147: \psi^*({\bf x},t)\psi^*({\bf x}^{\prime},t) V_0({\bf x}-{\bf x}^{\prime})

148: %\right. 

149: \times

150: \\

151: &&

152: %\left.

153: \psi({\bf x},t)\psi({\bf x}^{\prime},t)\Bigg]

154: \Bigg\}

155: \;.

156: \label{ac}

157: \eqa

158: $\psi^*({\bf x},t)$ is a complex field operator that creates a boson at the

159: position ${\bf x}$, $\mu$ is the chemical potential, and $V_0({\bf x})$ 

160: is the two-body potential. In the following, we set $\hbar=2m=1$.

161: Factors of $\hbar$ and $2m$ can be reinserted using dimensional analysis.

162: 

163: The action Eq.~(\ref{ac}) is symmetric under a phase transformation

164: \bqa

165: \psi({\bf x},t)\rightarrow e^{i\alpha}\psi({\bf x},t)\;.

166: \eqa

167: 

168: The $U(1)$-symmetry ensures that the density $\rho$ and current 

169: density ${\bf j}$ satisfy the continuity equation

170: \bqa

171: \dot{\rho}+\nabla\cdot{\bf j}=0\;.

172: \eqa

173: In the ground state, the current density ${\bf j}$ vanishes identically and

174: the condensate has a constant phase. The $U(1)$-symmetry can then be used to

175: make the condensate real everywhere.

176: 

177: If the interatomic potential $V_0({\bf x})$ is short range, 

178: it can be mimicked by local interactions. If the 

179: energies are low enough, the scattering amplitude 

180: can be approximated by $s$-wave scattering and the action Eq.~(\ref{act}) can

181: be approximated by~\cite{e2} 

182: \bqa%\nonumber

183: S&=&\int dt

184: \int d^2x\;

185: \psi^*

186: \left[

187: i%\hbar

188: {\partial\over\partial t}

189: +\nabla^2+\mu

190: \right]\psi

191: %\\&&

192: -{1\over2}

193: g\Big(\psi^*\psi\Big)^2

194: \;.

195: \label{act}

196: \eqa

197: Here, $g$ is a coupling constant that must be tuned to reproduce 

198: some low-energy observable of the true potential $V_0({\bf x})$.

199: 

200: The quantum field theory defined by the action Eq.~(\ref{act})

201: has ultraviolet divergences that must be removed by renormalization of

202: $\mu$ and $g$. %, and the field $\psi$. 

203: There is also an ultraviolet divergence in the expression for the density

204: $\rho$. This divergence can be removed by adding a counterterm $\delta\rho$.

205: Alternatively, one can eliminate the divergences associated with

206: $\mu$ and $\rho$ by a normal-ordering prescription of the fields 

207: in Eq.~(\ref{act}). The coupling constant is renormalized in the usual way

208: by replacing the bare coupling with the physical one.

209: 

210: If we use a simple momentum cutoff $M$ to cut off the

211: ultraviolet 

212: divergences in the loop integrals, there will be terms proportional to

213: $M^p$, where $p$ is a positive integer.

214: There are also terms that are proportional to $\log M$.

215: The coefficients of the power divergences depend on the regularization

216: method and are therefore artifacts of the regulator.

217: On the other hand, the coefficients of $\log(M)$ are independent of the

218: regulator and they therefore represent real physics.

219: In this paper, we use dimensional regularization 

220: to regulate both infrared and ultraviolet divergences.

221: In dimensional regularization, one calculates the loop

222: integrals in $d=2-2\epsilon$ dimensions for values of $\epsilon$ where

223: the integrals converge. One then analytically continues back to $d=2$

224: dimensions.

225: With dimensional regularization, an arbitrary renormalization scale $M$

226: is introduced. This scale can be identified with the simple momentum

227: cutoff mentioned above.

228: An advantage of dimensional regularization is that it 

229: automatically sets power 

230: divergences to zero, while logarithmic divergences

231: show up as poles in $\epsilon$.

232: In two dimensions, the one-loop 

233: counterterms for the chemical potential $\mu$ and the density $\rho$

234: are quadratic ultraviolet

235: divergences, while the one-loop counterterm for the coupling constant

236: $g$ is a logarithmic ultraviolet divergence.

237: At the two-loop level, the counterterms for the chemical potential and 

238: the density are also quadratic divergences. The counterterm for the 

239: coupling constant is a double logarithmic divergence.

240: 

241: 

242: We next parameterize the quantum field $\psi$

243: in terms of a time-independent

244: condensate $v$ and a quantum fluctuating field $\tilde{\psi}$:

245: \bqa

246: \psi=v+\tilde{\psi}\;.

247: \eqa

248: The fluctuating field can be written in terms of two real fields:

249: \bqa

250: \label{split}

251: \tilde{\psi}={1\over\sqrt{2}}\left(\psi_1+i\psi_2\right)\;.

252: \eqa

253: Substituting Eq.~(\ref{split}) into Eq.~(\ref{act}), the action can 

254: be decomposed into three terms

255: \bqa

256: \label{terms}

257: S[\psi]=S[v]+S_{\rm free}[\psi_1,\psi_2]

258: +S_{\rm int}[v,\psi_1,\psi_2]\;.

259: \eqa

260: $S[v]$ is the classical action

261: \bqa

262: S[v]=\int dt\int d^2x

263: \left[\mu v^2-{1\over2}gv^4%+v\nabla^2v

264: \right]\;,

265: \eqa

266: while the free part of the action is

267: \bqa\nonumber

268: S_{\rm free}[\psi_1,\psi_2]&=&\int dt\int d^2x

269: \left[

270: {1\over2}\left(\dot{\psi}_1\psi_2-\psi_1\dot{\psi_2}\right)

271: \right.\\

272: &&

273: \hspace{-1cm}

274: \left.

275: +{1\over2}

276: \psi_1\left(\nabla^2+X\right)\psi_1+{1\over2}

277: \psi_2\left(

278: \nabla^2+Y\right)\psi_2

279: \right]\;.

280: \label{free}

281: \eqa

282: The interaction part of the action is

283: \bqa\nonumber

284: S_{\rm int}[v,\psi_1,\psi_2]&=&

285: \int dt\int d^2x\left[

286: \sqrt{2}T\psi_1

287: \right. \\

288: &&

289: \hspace{-1cm}

290: \left.

291: %+{1\over2}

292: %X\psi_1^2+{1\over2}Y\psi_2^2

293: +{1\over\sqrt{2}}Z\psi_1\left(\psi_1^2+\psi_2^2\right)

294: %&&\left.

295: -{1\over8}g\left(\psi_1^2+\psi_2^2\right)^2

296: \right]\;.

297: \label{inter}

298: \eqa

299: The sources in Eq.~(\ref{inter}) are

300: \bqa

301: T&=&\left[\mu-gv^2\right]v%+{1\over2m}\nabla^2v

302: \;,\\

303: X&=&%\Lambda^2+%2m

304: \left[\mu-3gv^2\right]\;,\\

305: Y&=&%2m

306: \left[\mu-gv^2\right]\;,\\

307: Z&=&-gv\;.

308: \eqa

309: The propagator that corresponds to the free action $S_{\rm free}

310: [\psi_1,\psi_2]$

311: in Eq.~(\ref{free}) is

312: \bqa

313: \label{prop}

314: D(\omega,p)&=&\frac{i}{\omega^2

315: -\epsilon^2(p)+i\epsilon}\left(\begin{array}{cc}

316: p^2-Y&-i\omega \\

317: i\omega&p^2-X

318: \end{array}\right)\;.

319: \eqa

320: Here ${\bf p}$ is the wavevector, $\omega$ is the frequency, and

321: $\epsilon(p)$ is the dispersion relation:

322: \bqa

323: \label{disp}

324: \epsilon(p)=\sqrt{(p^2-X)(p^2-Y)}\;.

325: \eqa

326: The value of the condensate $v_0$ that minimizes the classical action is given

327: by the equation $T=0$.

328: Both the propagator Eq.~(\ref{prop}) and the dispersion relation 

329: Eq.~(\ref{disp}) greatly simplify at $v_0$, since

330: $Y=0$ here.

331: The dispersion relation then becomes 

332: gapless, which 

333: reflects the spontaneous breakdown of the $U(1)$-symmetry

334: (Goldstone's theorem). 

335: The dispersion relation is

336: linear for small wavevectors and 

337: is quadratic for large wavevectors, which is that of a 

338: free nonrelativistic 

339: particle. The propagator is defined with an $i\epsilon$ prescription

340: in the usual way. The diagonal parts of the propagator are denoted

341: by a solid and a dashed line, respectively. The off-diagonal parts

342: are denoted by a solid-dashed and dashed-solid line, respectively.

343: 

344: The partition function ${\cal Z}$ can be expressed as a path integral

345: \bqa

346: {\cal Z}=\int{\cal D}\psi_1{\cal D}\psi_2\,e^{iS[\psi_1,\psi_2]}\;.

347: \eqa

348: All the thermodynamic observables can be derived from ${\cal Z}$.

349: For instance, the free energy density ${\cal F}$ is given by

350: \bqa

351: {\cal F}(\mu)=i{\log{\cal Z}\over VT}\;,

352: \eqa

353: where $VT$ is the spacetime volume of the system.

354: 

355: The density $\rho$ is given by the expectation value 

356: $\langle\psi^{\dagger}\psi\rangle$ in the ground state. 

357: It can therefore be expressed

358: as 

359: \bqa

360: \label{rhodef}

361: \rho(\mu)=-{\partial {\cal F}(\mu)\over \partial \mu}\;.

362: \eqa

363: The energy density ${\cal E}$ is given by the Legendre transform of the

364: the free energy density

365: \bqa

366: \label{rela}

367: {\cal E}(\rho)={\cal F}(\mu)+\rho\mu\;.

368: \eqa

369: 

370: At this point it is convenient to introduce the 

371: thermodynamic potential $\Omega(\mu,v)$.  

372: The thermodynamic potential is given by all one-particle 

373: irreducible vacuum graphs and 

374: can be expanded in the number of loops

375: \bqa

376: \label{loopp}

377: \Omega(\mu,v)&=&\Omega_0(\mu,v)+\Omega_1(\mu,v)+\Omega_2(\mu,v)+...\;,

378: \eqa

379: where the subscript $n$ indicates the contribution from the

380: nth order in the loop expansion.

381: The free energy ${\cal F}$ is given by 

382: all connected vacuum graphs and is independent of the 

383: condensate $v$. If we evaluate $\Omega$ at a value of the condensate that

384: satisfies the condition

385: \bqa

386: \label{vc}

387: \bar{v}=\langle\psi\rangle\;,

388: \eqa

389: it can be shown that all the one-particle reducible graphs (those that can

390: be disconnected by cutting a single line) vanish.

391: We then have

392: \bqa

393: \label{loop}

394: {\cal F}(\mu)&=&

395: \Omega_0(\mu,\bar{v})+\Omega_1(\mu,\bar{v})+\Omega_2(\mu,\bar{v})+...\;.

396: \eqa

397: The condition Eq.~(\ref{vc}) is equivalent to 

398: \bqa

399: \label{tad1}

400: {\partial\Omega\over \partial v}=0\;.

401: \eqa

402: The loop expansion Eq.~(\ref{loop}) does not coincide with the expansion

403: in powers quantum corrections. To obtain the expansion in powers of

404: quantum correction, we must

405: expand the condensate $\bar{v}$ about the classical minimum $v_0$:

406: \bqa

407: \bar{v}=v_0+v_1+v_2+...\;.

408: \eqa

409: By substituting Eq.~(\ref{loopp}) into Eq.~(\ref{tad1}), we obtain the

410: first quantum correction $v_1$ to the classical minimum: 

411: \bqa

412: \label{v1}

413: v_1=-{\partial\Omega_1(\mu,v)\over\partial v}\Bigg|_{v=v_0}\Bigg/

414: {\partial^2\Omega_0(\mu,v)\over\partial v^2}\Bigg|_{v=v_0}

415: \;.

416: \eqa

417: This first quantum correction to the free energy density is

418: \bqa

419: {\cal F}_1(\mu)=\Omega_1(\mu,v_0)\;,

420: \eqa

421: and the second quantum correction to the free energy density

422: is

423: \bqa\nonumber

424: {\cal F}_2(\mu)&=&\Omega_2(\mu,v_0)+

425: v_1{\partial \Omega_1(\mu,v)\over\partial v}\Bigg|_{v=v_0}

426: \\

427: &&

428: +{1\over2}v^2_1{\partial^2\Omega_0(\mu,v)\over\partial v^2}\Bigg|_{v=v_0}\;.

429: \eqa

430: 

431: 

432: 

433: \section{Pressure to two loops}

434: In this section, we calculate the pressure as a function of the

435: chemical potential $\mu$ and the renormalized coupling $g$ to two loops.

436: \subsection{Mean-field free energy}

437: The thermodynamic potential in the mean-field approximation is

438: \bqa

439: \Omega_0(\mu,v)=-

440: \mu v^2+{1\over2}gv^4\;.

441: \eqa

442: The mean-field free energy is given by 

443: the classical thermodynamic potential evaluated at the classical

444: minimum $v_0$:

445: \bqa

446: \label{f0}

447: {\cal F}_0(\mu)=-{\mu^2\over2g}\;.

448: \eqa

449: \subsection{One-loop free energy}

450: The one-loop contribution to the free energy is 

451: \bqa

452: \label{f111}

453: {\cal F}_{1}(\mu)=

454: {\cal F}_{1a}(\mu)+\Delta_1{\cal F}\;,

455: \eqa

456: where

457: \bqa

458: \label{f1a}

459: {\cal F}_{1a}(\mu)=

460: {i\over2}\int{d\omega\over 2\pi}\int{d^dp\over(2\pi)^d}\log\det D^{-1}\;,

461: \eqa

462: and $\Delta_1{\cal F}$ is the one-loop counterterm

463: \bqa

464: \label{1c}

465: \Delta_1{\cal F}&=&-{\mu\over g}\Delta_1\mu+{\mu^2\over2g^2}\Delta_1g\;.

466: \eqa

467: The propagator $D(\omega,p)$ is evaluated at the classical minimum, where

468: $Y=0$.

469: By integrating over $\omega$, Eq.~(\ref{f1a}) becomes

470: \bqa\nonumber

471: {\cal F}_{1a}(\mu)&=&{1\over2}\int{d^dp\over(2\pi)^d}\;\epsilon(p)\;\\

472: \label{f1div}

473: &=&{1\over2}I_{0,-1}(2\mu)\;,

474: \eqa

475: where $I_{m,n}$ is defined in the appendix.

476: Using Eq.~(\ref{i1}) in the appendix, we obtain 

477: \bqa

478: \label{divv}

479: {\cal F}_{1a}(\mu)=-{\mu^2\over16\pi}\left[{1\over\epsilon}

480: -L-{1\over2}

481: +{\cal O}(\epsilon)

482: \right]\;,

483: \eqa

484: where $L=\log\left(\mu/2M^2\right)$.

485: The counterterms at one-loop are~\cite{berg}:

486: \bqa

487: \label{dm1}

488: \Delta_1\mu&=&0\;, \\

489: \label{dg1}

490: \Delta_1g&=&{g^2\over8\pi\epsilon}\;.

491: \eqa

492: Adding Eqs.~(\ref{1c}) and~(\ref{divv}), we obtain

493: the one-loop contribution to the free energy

494: \bqa

495: {\cal F}_{1}(\mu)=

496: {\mu^2\over16\pi}\left[L+{1\over2}\right]\;.

497: \eqa 

498: Note that ${\cal F}_1(\mu)$ is independent of the coupling

499: constant $g$.

500: Adding Eqs.~(\ref{f0}) and~(\ref{f1}), we obtain the one-loop approximation to

501: the free energy:

502: \bqa

503: \label{f1}

504: {\cal F}_{0+1}(\mu)=

505: -{\mu^2\over g}+

506: {\mu^2\over16\pi}\left[L+{1\over2}\right]\;,

507: \eqa

508: where the coupling constant is evaluated at the scale $M$, $g=g(M)$.

509: \subsection{Two-loop free energy}

510: \vspace{-1.5cm}

511: \begin{figure}

512: %\psfig{figure=b2.eps} 

513: \psfig{figure=b2.eps,width=9cm,height=14cm}

514: %\vspace{2mm} 

515: \vspace{-7.5cm}

516: \caption{\narrowtext

517: Two-loop vacuum graphs contributing to the free energy ${\cal F}$.}

518: \label{2loop}

519: \end{figure}

520: 

521: 

522: The two-loop contribution to the thermodynamic potential $\Omega_2(\mu,v)$

523: isgiven by the one-particle irreducible graphs

524: shown in Fig.~\ref{2loop}. The expression for the diagrams is

525: \bqa\nonumber

526: \Omega_2(\mu,v)&=&{g\mu\over8}J

527: +{g\over32}

528: \left[3I_{1,1}^2+2I_{-1,-1}I_{1,1}

529: \right.

530: \\&&

531: \left.

532: +3I_{-1,-1}^2\right]\;,

533: \eqa

534: where 

535: \bqa

536: J=6J_{0,0,1}-J_{-1,-1,1}-3J_{1,1,1}-2J_{-1,0,0}\;, 

537: \eqa

538: and

539: integrals $J_{l,m,n}$ are defined in the appendix.

540: The first correction $v_1$ is given by Eq.~(\ref{v1}):

541: \bqa

542: v_1=-{gv_0\over8\mu}\left[3I_{1,1}+I_{-1,-1}\right]\;.

543: \eqa

544: The two-loop contribution to the free energy is then 

545: \bqa\nonumber

546: {\cal F}_2(\mu)&=&

547: {g\mu\over8}J

548: +{g\over16}

549: \left[I_{-1,-1}^2-2I_{-1,-1}I_{1,1}-3I_{1,1}^2\right]

550: \\

551: &&

552: \label{f2j}

553: +\Delta_2{\cal F}\;.

554: \eqa

555: The counterterm $\Delta_2{\cal F}$ is

556: \bqa\nonumber

557: \Delta_2{\cal F}&=&

558: {1\over2}I_{1,1}\Delta_1\mu

559: -{\mu\over g}\Delta_2\mu+{\mu^2\over2g^2}\Delta_2g

560: \\

561: &&

562: -{1\over2g}\left(\Delta_1\mu\right)^2+{\mu\over g^2}\Delta_1\mu\Delta_1g

563: -{\mu^2\over2g^3}\Delta_1g^2

564: \;.

565: \eqa

566: At the two-loop level, the counterterms are given by

567: \bqa

568: \label{dm2}

569: \Delta_2\mu&=&0\;, \\

570: \label{dg2}

571: \Delta_2g&=&{g^3\over64\pi^2\epsilon^2}\;.

572: \eqa

573: The integrals $J_{l,m,n}$ are ultraviolet divergent. After subtracting the

574: divergent part, the remainder must be calculated numerically. We evaluate

575: the expression in the appendix. The result is

576: \bqa

577: J=-2I_{0,1}\left[I_{1,1}+I_{-1,-1}\right]+J_{\rm num}\;,

578: \eqa

579: where $J_{\rm num}=-3.52\times10^{-5}\mu$.

580: The final result for the two-loop contribution to the free energy is

581: \bqa

582: \label{f2}

583: {\cal F}_2(\mu)&=&-{g\mu^2\over64\pi^2}\left[1+C\right]

584: \;.

585: \eqa 

586: Here $C=2.78\times10^{-3}$.

587: Adding Eqs.~(\ref{f0}),~(\ref{f1}), and~(\ref{f2}), 

588: we obtain our final result for the free energy to second order

589: in the quantum loop expansion

590: \bqa

591: \label{f012}

592: {\cal F}_{0+1+2}(\mu)&=&

593: -{\mu^2\over2g}+{\mu^2\over16\pi}\left[L+{1\over2}\right]

594: -{g\mu^2\over64\pi^2}\left[1+C\right]

595: \;,

596: \eqa

597: \subsection{Pressure to two loops}

598: The pressure ${\cal P}$ is given by $-{\cal F}$. The pressure through

599: two loops is given by minus the sum of 

600: Eqs.~({\ref{f0}}),~(\ref{f1}) and~(\ref{f2}):

601: \bqa

602: \label{pres2}

603: {\cal P}_{0+1+2}(\mu)={\mu^2\over2g}-{\mu^2\over16\pi}\left[L+{1\over2}\right]

604: +{g\mu^2\over64\pi^2}\left[1+C\right]

605: \;.

606: \eqa

607: The coupling constant $g$ in Eq.~(\ref{act}) 

608: satisfies% a renormalization group equation.

609: \bqa

610: \label{rg2}

611: M{\!\!d\over dM}\;g&=&\beta(g)\;,%{g^2\over8\pi}+{\cal O}(g^3)\;.

612: \eqa

613: where the $\beta$-function is a polynomial in $g$.

614: Normally, the $\beta$-functions are

615: known only up to a certain order in the quantum loop

616: expansion. In the present case, the one-loop result for 

617: the $\beta$-function is exact

618: and $\beta(g)=g^2/4\pi$~\cite{berg}. 

619: From Eq.~(\ref{rg2}), one can easily

620: check that our result

621: Eq.~(\ref{pres2}) for the two-loop pressure is independent of the scale $M$

622: up to correction of order $g^2$.

623: \section{Energy density to two loops}

624: In this section, we derive the energy density ${\cal E}$ 

625: as a function of the density $\rho$ and the renormalized coupling

626: $g$ to two loops.

627: \subsection{Mean field energy density}

628: Using Eqs.~(\ref{rhodef}) and~(\ref{f0}) the density in the mean-field 

629: approximation is 

630: \bqa

631: \label{mu0}

632: \rho_{0}(\mu)={\mu\over g}\;.

633: \eqa

634: The chemical potential is obtained by inverting Eq.~(\ref{mu0}): 

635: \bqa

636: \mu_{0}(\rho)&=&g\rho\;.

637: \eqa

638: Using Eqs.~(\ref{rela}) and~(\ref{mu0}), the energy density 

639: in the mean-field approximation is given by

640: \bqa

641: \label{e0}

642: {\cal E}_0(\rho)&=&{1\over2}g\rho^2\;.

643: \eqa

644: \subsection{One-loop energy density}

645: Using Eqs.~(\ref{rhodef}) and~(\ref{f1})  

646: we obtain the density in the

647: one-loop approximation

648: \bqa

649: \label{rho1}

650: \rho_{0+1}(\mu)=

651: {\mu\over g}-{\mu\over8\pi}\bigg[L+1\bigg]\;.

652: \eqa

653: Inverting Eq.~(\ref{rho1}) to obtain $\mu$ as a function of $\rho$, one finds

654: \bqa

655: \label{inrho1}

656: \mu_{0+1}(\rho)=g\rho+{g^2\rho\over8\pi}\bigg[\bar{L}+1\bigg]\;,

657: \eqa 

658: where $\bar{L}=\log\left({g\rho/2M^2}\right)$

659: and $g=g(M)$.

660: Using Eqs.~(\ref{rela}),~(\ref{f1}) and~(\ref{inrho1}), 

661: the energy density in the one-loop approximation becomes

662: \bqa

663: \label{e1}

664: {\cal E}_{0+1}(\rho)={1\over2}g\rho^2

665: +{g^2\rho^2\over16\pi}\bigg[\bar{L}+{1\over2}\bigg]\;.

666: \eqa

667: This agrees with the result obtained 

668: by Lozano~\cite{loz}, and by Haugset and 

669: Ravndal~\cite{finn}. 

670: \subsection{Two-loop energy density}

671: Using Eqs.~(\ref{rhodef}) and~(\ref{f012}), we obtain the density in the

672: two-loop approximation:

673: \bqa%\nonumber

674: \rho_{0+1+2}(\mu)&=&{\mu\over g}

675: -{\mu\over8\pi}\bigg[L+1\bigg]

676: +{g\mu\over32\pi^2}\left[1+C\right]

677: \;.

678: \label{rho2}

679: \eqa

680: %where the prime denotes differentiation with respect to $\mu$.

681: Inverting Eq.~(\ref{rho2}), we obtain the two-loop expression for the

682: chemical potential:

683: \bqa\nonumber

684: \mu_{0+1+2}(\rho)&=&g\rho+{g^2\rho\over8\pi}\bigg[\bar{L}+1\bigg]

685: \\

686: &&

687: +{g^3\rho\over64\pi^2}\left[\bar{L}^2+3\bar{L}-2C\right]

688: \;.

689: \eqa 

690: The final result for the two-loop energy density then becomes

691: \bqa\nonumber

692: {\cal E}_{0+1+2}(\rho)&=&{1\over2}g\rho^2

693: +{g^2\over16\pi}\rho^2\left[\bar{L}+{1\over2}\right]

694: \\

695: &&

696: +{g^3\over128\pi^2}\rho^2\left[\bar{L}^2+

697: 2\bar{L}-1-2C\right]\;.

698: \label{final}

699: \eqa

700: Using the renormalization group equation~(\ref{rg2}) 

701: for the

702: running coupling constant, we see that the two-loop results for the

703: density, chemical potential,

704: and energy density are independent

705: of the renormalization scale $M$ up to corrections of order $g^4$. 

706: \section{Renormalization group}

707: In Secs.~III and IV, we showed that the results 

708: for the pressure, density, chemical potential, and

709: energy density are independent of the arbitrary renormalization

710: scale $M$ that is introduced with dimensional regularization.

711: 

712: The two-loop results for the pressure and density include a logarithm

713: $L$.

714: The renormalization group can be used to absorb

715: this logarithm.

716: The solution to the renormalization group equation~(\ref{rg2}) is

717: \bqa

718: \label{sol}

719: g(M_0)={g(M)\over{1-{g(M)\over8\pi}\log{\left(M_0^2\over M^2\right)}}}\;,

720: \eqa

721: By substituting the running coupling constant into the expressions for the

722: pressure and density in the mean-field approximation

723: and choosing the renormalization scale $M_0=\sqrt{\mu/2}$, we 

724: absorb this logarithm.

725: In a similar manner, we can sum logs of the form

726: $g^{n+1}L^n$ ($n=1,2,3...$) by using the running coupling in the

727: remaining terms in expressions for the pressure and density.

728: These terms are generated by expanding out Eq.~(\ref{sol}) in powers of ${L}$. 

729: The result is

730: \bqa

731: \rho(\mu)

732: &=&{\mu\over g_0}-{\mu\over8\pi}+{g_0\mu\over32\pi^2}\left[1+C\right] \;,

733: \\

734: {\cal P}(\mu)&=&

735: {\mu^2\over2g_0}-{\mu^2\over32\pi}+{g_0\mu\over64\pi^2}\left[1+C\right] \;,

736: \eqa

737: where  $g_0\equiv g(M_0)$.

738: In the same way, we can sum leading (of the form 

739: $g^{n+1}\bar{L}^n$, where $n=1,2,3...$) and subleading logs

740: (of the form 

741: $g^{n+m}\bar{L}^{n}$, where $n=1,2,3...$, and $m=2,3...$,)

742: from all orders of perturbation theory 

743: in the chemical potential and energy density

744: by using a running coupling constant 

745: $g(M_0)$ and choosing the renormalization scale $M_0=\sqrt{\rho/2}$.

746: These terms are generated by expanding out Eq.~(\ref{sol}).

747: The result is

748: \bqa\nonumber

749: \mu(\rho)&=&g_0\rho+{g^2_0\rho\over8\pi}\left[\log(g_0)+1\right]

750: \\

751: &&

752: +{g^3_0\rho\over64\pi^2}

753: \left[\log^2(g_0)+3\log(g_0)-2C\right]\;,\\ \nonumber

754: {\cal E}(\rho)&=&

755: {1\over2}g_0\rho+{g^2_0\rho\over16\pi}\left[\log(g_0)+{1\over2}\right]

756: \\

757: &&

758: +{g^3_0\rho\over128\pi^2}

759: \left[\log^2(g_0)+2\log(g_0)-1-2C\right]\;,

760: \eqa

761: where $g_0\equiv g(M_0)$.

762: 

763: 

764: We have approximated the two-body potential by local interactions.

765: In real systems, however,  the interactions have a finite range $a$ and

766: so $1/a$ provides a natural ultraviolet cutoff $M$. 

767: In the dilute gas limit, we have

768: $|\log(\mu a^2/2)|\gg1$ and $|\log(\rho a^2/2)|\gg1$

769: ~\cite{schiff,popov1,popov2,fis,rg2,fried},

770: and the running coupling constant

771: is approximately equal to either

772: $-8\pi[\log(\mu a^2/2)]^{-1}$

773: or $-8\pi[\log(\rho a^2/2)]^{-1}$.

774: This is essentially the two-body $T$-matrix and so this is the relevant

775: expansion parameter.

776: By substituting these

777: expressions for the coupling constant into to the expressions

778: for the pressure, density, chemical potential, and energy density, we 

779: obtain the renormalization group improved results

780: \bqa

781: \label{pfinal}

782: {\cal P}(\mu)&=&-{\mu^2L\over16\pi}

783: \left[1+{1\over2}L^{-1}+2(1+C)L^{-2}\right]\;, \\

784: \rho(\mu)&=&-{\mu L\over8\pi}

785: \left[1+L^{-1}+2(1+C)L^{-2}\right]\;,

786: \\

787: \nonumber

788: \mu(\rho)&=&-{8\pi\rho\over\bar{L}}

789: \left[

790: 1+

791: \left(\log[-\bar{L}/8\pi]-1\right)\bar{L}^{-1}

792: \right. \\ \nonumber

793: &&

794: +\log^2\left(\log[-\bar{L}/8\pi]\right)L^{-2}

795: -{3}\left(\log[-\bar{L}/8\pi]\right)L^{-2}

796: \\

797: &&\left.

798: -2C\bar{L}^{-2}

799: \right]

800: \label{myfinal}

801:  \;, \\

802: \nonumber

803: {\cal E}(\rho)&=&-{4\pi\rho^2\over\bar{L}}

804: \left[

805: 1+{1\over2}\left(2\log[-\bar{L}/8\pi]-1\right)\bar{L}^{-1}

806: \right.\nonumber

807: \\ \nonumber

808: &&

809: +\log^2\left(\log[-\bar{L}/8\pi]\right)L^{-2}

810: -{2}\left(\log[-\bar{L}/8\pi]\right)L^{-2}

811: \\

812: &&

813: \left.

814: -(1+2C)\bar{L}^{-2}

815: \right]

816: \label{efinal}

817: \;,

818: \eqa

819: where where $L=\left[\log({\mu a^2}/2)\right].$

820: and

821: $\bar{L}=\left[\log(\rho a^2/2)\right]$.

822: Eqs.~(\ref{pfinal})-(\ref{efinal}) are the main result of the 

823: present paper. The 

824: leading order results 

825: were first obtained by Schick~\cite{schiff}, while 

826: the leading corrections have been calculated

827: in Refs.~\cite{popov1,popov2,fis,rg2,fried,hines}.

828: The last terms are new result.

829: \section{Summary}

830: In the present paper, we have studied a two-dimensional interacting 

831: homogeneous Bose gas

832: at zero temperature. We have calculated 

833: the ground state pressure and energy density to second order in the

834: quantum loop expansion. The results are independent of the

835: arbitrary renormalization scale $M$.

836: 

837: We have applied the renormalization group to sum up leading and subleading

838: logarithms from all orders in perturbation theory. 

839: The renormalization group improved pressure and density are 

840: expansions in $\left[\log(\mu a^2/2)\right]^{-1}$,

841: while the chemical potential and energy density are expansions in 

842: $\left[\log(\rho a^2/2)\right]^{-1}$.

843: These are essentially expansions in the two-body $T$-matrix,

844: and we have obtained the ground state pressure, density, chemical

845: potential, and energy density 

846: to next-to-next-to-leading order.

847: 

848: 

849: 

850: \section*{Acknowledgments}

851: This work was supported by the Stichting voor

852: Fundamenteel Onderzoek der Materie

853: (FOM), which is supported by the Nederlandse Organisatie voor Wetenschapplijk

854: Onderzoek (NWO). 

855: \appendix

856: \renewcommand{\theequation}{\thesection.\arabic{equation}}

857: \setcounter{equation}{0}

858: \section{Formulas}

859: The loop integrals that appear in our calculations involve integrations

860: over the energy $\omega$ and the spatial momentum ${\bf p}$.

861: The energy integrals are evaluated using contour integration. 

862: 

863: The specific one-loop integral needed is

864: \bqa

865: \label{e11}

866: \int{d\omega\over2\pi}{1\over[\omega^2-\epsilon^2(p)+i\epsilon]}&=&

867: -{i\over2\epsilon(p)}\;.

868: \eqa

869: The specific two-loop integrals needed are

870: \bqa

871: \nonumber

872: \int{d\omega_1\over2\pi}\int{d\omega_2\over2\pi}

873: {1\over\left[\omega_1^2-\epsilon^2(p)+i\epsilon\right]}

874: \times

875: &&\\ \nonumber

876: {1\over[\omega_2^2-\epsilon^2(q)+i\epsilon][(\omega_1+\omega_2)^2-\epsilon^2(r)+i\epsilon]}

877: &=&\\

878: %\hspace{-5cm}

879: {1\over4\epsilon(p)\epsilon(q)\epsilon(r)(\epsilon(p)+\epsilon(q)+

880: \epsilon(r))}\;,\\ \nonumber

881: \int{d\omega_1\over2\pi}\int{d\omega_2\over2\pi}

882: {\omega_1\omega_2\over[\omega_1^2-\epsilon^2(p)+i\epsilon]}

883: \times

884: && \\ \nonumber

885: {1\over[\omega_2^2-\epsilon^2(q)+i\epsilon][(\omega_1+\omega_2)^2-\epsilon^2(r)+i\epsilon]}

886: &=&

887: \\

888: {1\over4\epsilon(r)[\epsilon(p)+\epsilon(q)+

889: \epsilon(r)]}\;.

890: &&

891: \eqa

892: Here $r=|{\bf p}+{\bf q}|$.

893: 

894: Some of the one-loop momentum

895: integrals are infrared divergent or ultraviolet divergent or

896: both.

897: They can be written in terms of the integral $I_{m,n}$, which is defined by

898: \bqa

899: \label{idef}

900: I_{m,n}=\left({e^{\gamma}M^2\over4\pi}\right)^{\epsilon}

901: %M^{2\epsilon}

902: \int{d^dp\over(2\pi)^d}{p^{2m}\over p^n(p^2+\Lambda^2)^{n/2}}\;.

903: \eqa

904: Here, $M$ is a renormalization scale that ensures that $I_{m,n}$ has the

905: canonical dimension also for $d\neq2$.

906: $\gamma\approx0.5772$ is the Euler-Mascheroni constant. 

907: With dimensional regularization, $I_{m,n}$ is given by the formula

908: \bqa\nonumber

909: I_{m,n}&=&{\Omega_d\over(2\pi)^d}

910: \left({e^{\gamma}M^2\over4\pi}\right)^{\epsilon}

911: \Lambda^{d+2m-2n}

912: \\

913: &&

914: {\Gamma({{d-n\over2}+m})\Gamma(n-m-{d\over2})\over2\Gamma({n\over2})}\;,

915: \eqa

916: where $\Omega_d=2\pi^{d/2}/\Gamma[d/2]$ is the 

917: area of the $d$-dimensional sphere.

918: 

919: The integrals $I_{m,n}$ satisfy the relations

920: \bqa

921: \label{alge}

922: {\!\!\!d\over d\Lambda^2}I_{m,n}&=&-{n\over2}I_{m+1,n+2}\;,\\

923: \left(d+2m-n\right)I_{m,n}&=&nI_{m+2,n+2}\;,\\

924: \Lambda^2I_{m,n}&=&I_{m-1,n-2}-I_{m+1,n}\;.

925: \eqa

926: The first relation follows directly from the definition of $I_{m,n}$.

927: The second relation follows

928: from integration by parts, while the last

929: is simply an algebraic relation.

930: 

931: In two dimensions, these integrals have logarithmic and power ultraviolet

932: divergences. The power divergences are set to zero in dimensional 

933: regularization, while the logarithmic divergences appear as poles in 

934: $\epsilon$. The specific integrals are

935: \bqa\nonumber

936: I_{0,-1}&=&-{\Lambda^4\over32\pi}

937: \left\{{1\over\epsilon}-L

938: -{1\over2}

939: \right.\\

940: &&

941: \left.

942: +{1\over2}\left[

943: L^2+L-{5\over2}+{\pi^2\over2}

944: \right]

945: \epsilon+

946: {\cal O}\left(\epsilon^2\right)

947: \right\}

948: \label{i1}

949: \;,\\ \nonumber

950: I_{-1,-1}&=&{\Lambda^2\over8\pi}\left\{{1\over\epsilon}

951: -L+1

952: \right.\\

953: &&

954: \left.

955: +{1\over2}\left[

956: \left(L-1\right)^2

957: +1+{\pi^2\over2}\right]\epsilon

958: +{\cal O}\left(\epsilon^2\right)

959: \right\}\;,\\

960: \label{i3}\nonumber

961: I_{1,1}&=&-{\Lambda^2\over8\pi}\left\{{1\over\epsilon}

962: -L-1

963: \right.\\ 

964: &&

965: \left.

966: +

967: {1\over2}\left[

968: (L+1)^2-4+{\pi^2\over2}

969: \right]\epsilon

970: +{\cal O}\left(\epsilon^2\right)

971: \right\}

972: \label{i2}

973: \;, \\

974: I_{0,1}&=&{1\over4\pi}\left\{{1\over\epsilon}

975: -L

976: +{1\over2}\left[

977: L^2%\left(L-1\right)^2

978: +{\pi^2\over2}

979: \right]\epsilon

980: +{\cal O}\left(\epsilon^2\right)

981: \right\}

982: \;,

983: \label{i4}

984: \eqa

985: where $L=\log(\Lambda^2/4M^2)$.

986: 

987: The two-loop integrals needed can be expressed in terms of $J_{l,m,n}$

988: \bqa\nonumber

989: \label{jlmn}

990: J_{l,m,n}&=&\left({e^{\gamma}M^2\over4\pi}\right)^{2\epsilon}

991: \int{d^dp\over(2\pi)^d}\int{d^dq\over(2\pi)^d}

992: \times

993: \\

994: &&

995: \hspace{-1.4cm}

996: {\Big[p/\sqrt{p^2+\Lambda^2}\Big]^{l}

997: \Big[q/\sqrt{q^2+\Lambda^2}\Big]^{m}

998: \Big[r/\sqrt{r^2+\Lambda^2}\Big]^{n}

999: \over p\sqrt{p^2+\Lambda^2}+q\sqrt{q^2+\Lambda^2}+r\sqrt{r^2+\Lambda^2}}

1000: \;.

1001: \eqa 

1002: In two dimensions, these integrals have quadratic and double logarithmic

1003: divergences that cancel in the particular combination

1004: in Eq.~(\ref{f2j}), 

1005: leaving us with a logarithmically divergent integral.

1006: We write the integral as

1007: \bqa

1008: J=J_{\rm div}+J_{\rm num}\;,

1009: \eqa

1010: where 

1011: the ultraviolet divergence of 

1012: the integral $J$ has been isolated:

1013: \bqa\nonumber

1014: J_{\rm div}&=&2\left({e^{\gamma}M^2\over4\pi}\right)^{2\epsilon}

1015: \int{d^dp\over(2\pi)^d}

1016: \left[2-{p\over\sqrt{p^2+\Lambda^2}}

1017: \right.\\&&

1018: \left.

1019: -{\sqrt{p^2+\Lambda^2}\over p}

1020: \right]

1021: \int{d^dq\over(2\pi)^d}{1\over q\sqrt{q^2+\Lambda^2}}

1022: \label{div}

1023: \;.

1024: \eqa

1025: The first term inside the square brackets in Eq.~(\ref{div})

1026: has only a power divergence, and so it is set to zero in 

1027: dimensional regularization. In terms of $I_{m,n}$, the remaining terms

1028: can be written as

1029: \bqa

1030: J_{\rm div}=-2I_{0,1}\left[

1031: I_{1,1}+I_{-1,-1}

1032: \right]\;.

1033: \eqa

1034: The remaining finite part of the integral $J$ can be evaluated directly

1035: in two dimensions and reads

1036: \bqa\nonumber

1037: J_{\rm num}&=&%\left({e^{\gamma}M^2\over4\pi}\right)^{2\epsilon}

1038: \int{d^2p\over(2\pi)^2}\int{d^2q\over(2\pi)^2}\Bigg\{

1039: \left[{6p\over \sqrt{p^2+\Lambda^2}}

1040: -{2\over p}\sqrt{p^2+\Lambda^2}

1041: \right.\\ \nonumber

1042: &&

1043: -{3pqr\over \sqrt{p^2+\Lambda^2}\sqrt{q^2+\Lambda^2}\sqrt{r^2+\Lambda^2}}

1044: \\ \nonumber

1045: &&

1046: \left.

1047: -{r\sqrt{p^2+\Lambda^2}\sqrt{q^2+\Lambda^2}\over

1048: pq\sqrt{r^2+\Lambda^2}}\right]

1049: \times

1050: \\ \nonumber

1051: &&

1052: {1\over p\sqrt{p^2+\Lambda^2}+q\sqrt{q^2+\Lambda^2}

1053: +r\sqrt{r^2+\Lambda^2}}

1054: \\\nonumber

1055: &&

1056: -{2\over q\sqrt{q^2+\Lambda^2}}\left[

1057: 2

1058: -{p\over\sqrt{p^2+\Lambda^2}}

1059: \right.

1060: \\

1061: &&

1062: \left.

1063: -{\sqrt{p^2+\Lambda^2}\over p}

1064: \right]

1065: \Bigg\}

1066: \;.

1067: \label{jint}

1068: \eqa

1069: Since the only scale in the integrand in Eq.~(\ref{jint}) is $\Lambda$,

1070: it follows from dimensional analysis that

1071: $J_{\rm  num}$ is proportional to $\Lambda^2$. The numerical value is

1072: \bqa

1073: J_{\rm num}=-1.76\times10^{-5}\Lambda^2\;.

1074: \eqa

1075: 

1076: \begin{thebibliography}{99}

1077: \bibitem{bec1} M.H. Anderson, J.R. Ensher, M.R. Matthews, C.E. Wieman

1078: and E.A. Cornell, Science {\bf 269}, 198 (1995).

1079: \bibitem{bec2}

1080: K.B. Davis, M.O. Mewes, M.R. Andrews, N.J. van Druten,

1081: D.S. Durfee, D.M. Kurn and W. Ketterle, Phys. Rev. Lett. {\bf 75}, 

1082: 3969 (1995).

1083: \bibitem{bec3}

1084: C.C. Bradley, C.A. Sackett, J.J. Tollett and R.G. Hulet, 

1085: Phys. Rev. Lett. {\bf 75}, 1687 (1995).

1086: 

1087: \bibitem{string} F. Dalfovo, S. Giorgini, L.P. Pitaevskii, and S. Stringari, Rev. Mod. Phys. {\bf 71}, 463 (1999).

1088: \bibitem{grif}H. Shi and A. Griffin, Phys. Rep. {\bf 304}, 1 (1998).

1089: 

1090: \bibitem{leeyang}T.D. Lee and C.N. Yang, Phys. Rev. {\bf 105}, 1119 (1957).

1091: \bibitem{to} T.T. Wu, Phys. Rev. {\bf 115}, 1390 (1959); N.M. Hugenholz

1092: and D. Pines, Phys. Rev. {\bf 116}, 489 (1959); K. Sawada, 

1093: Phys. Rev. {\bf 116}, 1344 (1959).

1094: \bibitem{eric}E. Braaten and A. Nieto, Eur. J. Phys, (1997).

1095: \bibitem{e2}E. Braaten, H.-W. Hammer and S. Hermans, Phys. Rev. {\bf A63}, 063609 (2001). 

1096: 

1097: \bibitem{schiff}M. Schick, Phys. Rev. {\bf A3}, 1067 (1971).

1098: 

1099: \bibitem{popov1} V.N. Popov, Theor. Math. Phys. {\bf 11}, 565 (1972).

1100: \bibitem{popov2} V.N. Popov {\it Functional Integrals in Quantum Field Theory and Statistical Physics}, (Reidel, Dordrecht, 1983). 

1101: \bibitem{fis}D.S. Fisher and P.C. Hohenberg, Phys. Rev. {\bf B37}, 4936 (1988).\bibitem{rg2}E.B. Kolomeisky and J.P. Straley, Phys. Rev. {\bf B46}, 11749 (1992).

1102: \bibitem{fried} C.-C. Chang and R. Friedberg, Phys. Rev. {\bf B51}, 1117 (1995).

1103: \bibitem{hines} D.F. Hines, N.E. Frankel, and D.J. Mitchell, Phys. Lett. {\bf A68}, 12 (1978).

1104: \bibitem{lieb1}E.H. Lieb and J. Yngvason, e-print math-ph/0002014.

1105: \bibitem{lieb2}E.H. Lieb, R. Seiringer, and J. Yngvason, e-print 

1106: cond-mat/0005026.

1107: 

1108: 

1109: \bibitem{berg}O. Bergman, Phys. Rev. {\bf D46}, 5474 (1992). 

1110: \bibitem{loz}G. Lozano, Phys. Lett. {\bf B283}, 70 (1992).

1111: \bibitem{finn}T. Haugset and F. Ravndal, Phys. Rev. {\bf D49}, 4299 (1994).

1112: 

1113: 

1114: \end{thebibliography}

1115: 

1116: \end{multicols}

1117: \end{document}

1118: 

1119: 

1120: