1: %\documentclass[showpacs,preprint,prl,aps]{revtex4}

2: \documentclass[showpacs,twocolumn,prl,aps]{revtex4}

3: \usepackage{graphicx}

4: 

5: \begin{document}

6: 

7: \title[Short title for running header]{Spinon-Holon Recombination in Gutzwiller Projected Wave Functions}

8: \author{Hong-Yu Yang}

9: \affiliation{Center for Advanced Study, Tsinghua University,

10: Beijing, 100084, P. R. China}

11: \author{Tao Li}

12: \affiliation{Department of Physics, Renmin University of China,

13: Beijing, 100872, P. R. China}

14: \date{{\small \today}}

15: 

16: 

17: \begin{abstract}

18: The Gutzwiller projection is shown to induce nontrivial correlation

19: between the spinon and the holon in the slave Boson theory of the

20: $t-J$ model. We find this correlation is responsible some subtle

21: differences between the slave Boson mean field theory and the

22: Gutzwiller projected wave function(GWF), among which the

23: particle-hole asymmetry in the quasiparticle weight calculated from

24: the GWF is a particular example.

25: \end{abstract}

26: 

27: \pacs{74.20.Mn,74.25.Ha,75.20.Hr}

28: \maketitle

29: 

30: Superconductivity results from Bose condensation of charged

31: particles. In the BCS theory of superconductivity, Fermionic

32: electrons are paired into Bosonic Copper pairs which then condense

33: into a superfluid. Soon after the discovery of high temperature

34: superconductors, Anderson proposed an exotic way toward

35: superconductivity in this class of materials. Anderson's proposal

36: for superconductivity is to fractionalize the electron, rather than

37: pair them up\cite{Anderson,Baskran,Zou}. According to Anderson's

38: argument, doping holes into the parent compounds of the high

39: temperature superconductors, which are two dimensional

40: antiferromagnetic insulators, would generate a liquid-type spin

41: state called RVB state. Anderson argued that the excitations above

42: the RVB state are fractionalized. More specifically, the spin and

43: the charge quantum number of the electron are carried separately by

44: two kinds of excitations, namely a spin-$\frac{1}{2}$ Fermionic

45: excitation called spinon and a charge 1 Bosonic charged excitation

46: called holon. In such a spin-charge separated system, the charged

47: holon is liberated from the Fermionic statistics of the electron and

48: is ready to condense into a superfluid.

49: 

50: A problem with Anderson's proposal is that the predicted $T_{c}$ is

51: too high\cite{Kotliar,Lee1}. The holon supercurrent is not

52: efficiently dissipated in this mechanism. In the BCS theory of

53: superconductivity, the supercurrent is dissipated by thermally

54: generated quasiparticle excitations. These quasiparticles, which are

55: charged Fermionic excitations, form a normal fluid and cause

56: dissipative response in external electromagnetic(EM) field. On the

57: other hand, in an ideal spin-charge separated system, the Fermionic

58: spinon has been deprived of the charge to cause dissipative response

59: in an EM field, while the Bosonic excitation of the holon system is

60: very inefficient in dissipating the supercurrent.

61: 

62: One suggested solution for the above problem is to recombine the

63: separated spinon and holon\cite{Lee2,Wen,Ng}. By spinon - holon

64: recombination, the Fermionic spinon excitation acquire the charge to

65: cause dissipation in an EM field, or, the charged holon regain the

66: Fermionic statistics to be transformed into a normal carrier outside

67: the condensate.

68: 

69: However, the origin and even the exact meaning of such recombination

70: are still elusive. In this paper, we address the problem of

71: spinon-holon recombination in the slave Boson theory of the $t-J$

72: model. The slave Boson theory is the most direct way to realize the

73: RVB idea. In the slave Boson theory, the constrained electron

74: operator $\hat{c}_{i\sigma}$ is rewritten as

75: $f_{i\sigma}b_{i}^{\dagger}$, in which $f_{i\sigma}$ denotes the

76: Fermionic spinon and $b_{i}^{\dagger}$ denotes the Bosonic holon. To

77: be a faithful representation of the original electron, the slave

78: particles should be subjected to the local constraint

79: $\sum_{\sigma}f_{i,\sigma}^{\dagger}f_{i,\sigma}+b_{i}^{\dagger}b_{i}=1$.

80: 

81: At the mean field level, the local constraint on the slave particle

82: is relaxed to a global one. The spinon and the holon then move

83: independently. Wen and Lee proposed that the Gaussian fluctuation

84: around the mean field solution, which are gauge fluctuations, tends

85: to recombine the spinon and the holon\cite{Wen}. However, the lack

86: of a reliable treatment of gauge fluctuation in 2+1 dimension make

87: it hard to make any quantitative prediction.

88: 

89: In this paper, we address the problem of spinon-holon recombination

90: in the Gutzwiller projected wave functions(GWF). The GWF is derived

91: from the slave Boson mean field state by filtering out its

92: unphysical components with doubly occupied sites. Technically, the

93: Gutzwiller projection, which enforces the local constraint, accounts

94: partially the effect of gauge fluctuation. Thus some kind of

95: spinon-holon recombination is expected in GWF. We find the

96: Gutzwiller projection not only enforces the local constraint, but

97: also induces longer range correlation between the spinon and the

98: holon. This correlation is given by a well defined correlation

99: function and is found to be responsible for some subtle differences

100: between the mean field theory and the GWF. For example, the

101: particle-hole asymmetry in the quasiparticle weight calculated from

102: GWF is shown to be due to this correlation.

103: 

104: We start from the $t-J$ model which is defined as

105: \begin{equation}

106:     \mathrm{H}=-t\sum_{<i,j>,\sigma}(\hat{c}_{i,\sigma}^{\dagger}\hat{c}_{j,\sigma}+h.c.)

107:     +J\sum_{<i,j>}(\mathbf{S}_{i}\mathbf{S}_{j}-\frac{1}{4}n_{i}n_{j}),

108: \end{equation}

109: in which

110: $\mathbf{S}_{i}=\sum_{\alpha\beta}\hat{c}_{i,\alpha}^{\dagger}\mathbf{\sigma}_{\alpha\beta}\hat{c}_{i,\beta}$

111: and

112: $n_{i}=\sum_{\alpha}\hat{c}_{i,\alpha}^{\dagger}\hat{c}_{i,\alpha}$.

113: The electron operator $\hat{c}_{i,\sigma}$ in (1) is subjected to

114: the constraint of no boule occupancy

115: \begin{equation}

116: \sum_{\sigma}\hat{c}_{i,\sigma}^{\dagger}\hat{c}_{i,\sigma}\leq 1.

117: \end{equation}

118: 

119: 

120: In terms of the spinon and the holon operator, the $t-J$ model

121: reads\cite{Kotliar}

122: \begin{eqnarray*}

123:    \mathrm{H} &=& -t\sum_{<i,j>,\sigma}(f_{i,\sigma}^{\dagger}f_{j,\sigma}b_{j}^{\dagger}b_{i}+h.c.) \\

124:    &+&\frac{J}{2}\sum_{<i,j>,\alpha,\beta}(f_{i\alpha}^{\dagger}f_{i\beta}f_{j\beta}^{\dagger}f_{j\alpha}

125:    -f_{i\alpha}^{\dagger}f_{i\alpha}f_{j\beta}^{\dagger}f_{j\beta}),

126: \end{eqnarray*}

127: in which $f_{i\alpha}$ and $b_{i}$ are now subjected to the

128: constraint

129: \begin{equation}

130: \sum_{\sigma}f_{i,\sigma}^{\dagger}f_{i,\sigma}+b_{i}^{\dagger}b_{i}=1.

131: \end{equation}

132: 

133: In the mean field theory, the interaction term is decoupled by

134: introducing RVB order parameter $\chi_{ij}=\sum_{\sigma}\langle

135: f_{i,\sigma}^{\dagger}f_{j,\sigma} \rangle$ and

136: $\Delta_{ij}=\sum_{\alpha}\langle

137: \epsilon_{\alpha\beta}f_{i,\alpha}^{\dagger}f_{j,\beta} \rangle$, in

138: which $\epsilon_{\alpha\beta}$ is the total antisymmetric tensor.

139: The local constraint is relaxed to a global one. The mean field

140: Hamiltonian for the spinon and the holon take the form

141: \begin{equation}

142:     \mathrm{H}_{f}=\sum_{\mathrm{k},\sigma}\xi_{\mathrm{k}}f_{\mathrm{k},\sigma}^{\dagger}f_{\mathrm{k},\sigma}+

143: \sum_{\mathrm{k},\sigma}\Delta_{\mathrm{k}}

144: (f_{\mathrm{k},\sigma}^{\dagger}f_{\mathrm{-k},\bar{\sigma}}^{\dagger}+h.c.)

145: \end{equation}

146: and

147: \begin{equation}

148:     \mathrm{H}_{h} =\sum_{\mathrm{k}}\epsilon_{\mathrm{k}}b_{\mathrm{k}}^{\dagger}b_{\mathrm{k}}

149: \end{equation}

150: in momentum space. The mean field ground state is given by the

151: following product

152: \begin{equation}

153:   |\mathrm{G}\rangle_{\mathrm{MF}}

154:   =(b_{q=0}^{\dagger})^{N_{b}}|\mathrm{BCS}\rangle,

155: \end{equation}

156: in which $|\mathrm{BCS}\rangle$ denotes the BCS pairing state of the

157: spinon. A mean field excitation, for, say, a hole-like quasipaticle,

158: is given by

159: \begin{equation}

160:   |\mathrm{k},\sigma\rangle_{\mathrm{MF}}

161:   =b_{q=0}^{\dagger}\gamma_{\mathrm{k}\sigma}^{\dagger}|\mathrm{G}\rangle_{\mathrm{MF}},

162: \end{equation}

163: in which $\gamma_{\mathrm{k}\sigma}^{\dagger}$ denotes the Bogliubov

164: quasiparticle of the BCS Hamiltonian for spinon.

165: 

166: The GWF for the ground state is given by projecting

167: $|\mathrm{G}\rangle_{\mathrm{MF}}$ into the subspace that satisfy

168: the constraint Eq.(3),

169: \begin{equation}

170:   |\mathrm{G}\rangle_{\mathrm{Var}}

171:   =\mathrm{P_{G}}(b_{q=0}^{\dagger})^{N_{b}}|\mathrm{BCS}\rangle,

172: \end{equation}

173: in which $|\mathrm{G}\rangle_{\mathrm{Var}}$ denotes the variational

174: ground state and $\mathrm{P_{G}}$ denotes the Gutzwiller projection.

175: Similarly, $\mathrm{P_{G}}|\mathrm{k}\rangle_{\mathrm{MF}}$ provides

176: a variational guess for the quasipaticle excitation on

177: $|\mathrm{G}\rangle_{\mathrm{Var}}$. The same construction for the

178: quasiparticle excitation is used in some recent works\cite{Yunoki}.

179: 

180: The spinon-holon recombination in GWF can be most easily seen by

181: inspecting the quasiparticle weight $\mathrm{Z_{k}}$. In the mean

182: field theory, $\mathrm{Z_{k}}$ scales linearly with hole density ,

183: since the coherent spectral weight is caused by holon condensation.

184: However, $\mathrm{Z_{k}}$ calculated from GWF vanishes more slowly

185: near half filling. For example, $\mathrm{Z_{k}}$ calculated from the

186: one dimensional GWF vanishes as $x^{1/2}$ near half

187: filling\cite{Gebhard}. Similar behavior is also observed in two

188: dimensional systems\cite{Yang}. In some cases, $\mathrm{Z_{k}}$

189: calculated from GWF can even be nonzero at half filling\cite{TKLee}.

190: This discrepancy can be naturally explained by invoking spinon-holon

191: recombination in GWF.

192: 

193: Now we define spinon-holon recombination in GWF more explicitly. For

194: clarity's sake, we first consider a half filled system and assume

195: the spinon pairing term to be absent\cite{note1}. The mean field

196: ground state of the system is then given by a half filled spinon

197: Fermi sea, while a hole-like quasiparticle is generated as follow

198: \begin{equation}

199:     |\mathrm{k},\uparrow

200:     \rangle_{\mathrm{MF}}=b_{\mathrm{q=0}}^{\dagger}f_{-\mathrm{k},\downarrow}|\mathrm{FS}\rangle,

201: \end{equation}

202: in which $|\mathrm{FS}\rangle$ denotes the half filled Fermi sea. At

203: the mean field level, $|\mathrm{k},\uparrow\rangle_{\mathrm{MF}}$ is

204: interpreted as a state with an added holon and a hole of spinon on

205: the background of $|\mathrm{FS}\rangle$. In real space,

206: \begin{equation}

207:     |\mathrm{k},\uparrow\rangle_{\mathrm{MF}}=\frac{1}{N}\sum_{i,j}e^{i\mathrm{k}(r_{i}-r_{j})}b_{i}^{\dagger}f_{j\downarrow}|\mathrm{FS}\rangle.

208: \end{equation}

209: Thus, the added holon and the hole of spinon move independently on

210: $|\mathrm{FS}\rangle$ and have a chance of only $\frac{1}{N}$ to

211: recombine into the original electron. This explains the vanishing of

212: the quasiparticle weight at half filling in the mean field theory.

213: 

214: Now let's see what happens when the local constraint is enforced by

215: the Gutzwiller projection. The projected wave function for the above

216: quasiparticle is given by

217: \begin{equation}

218:     |\mathrm{k},\uparrow\rangle_{\mathrm{Var}}

219:     =\frac{1}{N}\sum_{i,j}e^{i\mathrm{k}(r_{i}-r_{j})}\mathrm{P_{G}} b_{i}^{\dagger}f_{j\downarrow}|\mathrm{FS}\rangle.

220: \end{equation}

221: Following the convention of the mean field theory, we interpret the

222: index $i$ and $j$ as the locations of the added holon and the hole

223: of spinon. The difference between

224: $|\mathrm{k},\uparrow\rangle_{\mathrm{Var}}$ and

225: $|\mathrm{k},\uparrow\rangle_{\mathrm{MF}}$ can be understood as

226: follows. In the mean field theory, the relative motion between the

227: added holon and the hole of spinon is described by plane wave

228: factor. The background particles in $|\mathrm{FS}\rangle$ do not

229: contribute to their correlation. After the Gutzwiller projection,

230: thees added particles get correlated with the background particles

231: through the local constraint. Thus the background particles in

232: $|\mathrm{FS}\rangle$ can also contribute to the correlation between

233: the added holon and the hole of spinon in

234: $|\mathrm{k},\uparrow\rangle_{\mathrm{Var}}$.

235: 

236: First, we assume $i \neq j$. Due to the local constraint, spinons in

237: $|\mathrm{FS}\rangle$ must keep away from the site $i$, which is

238: already assigned to the added holon. At the same time, site $j$

239: should be doubly occupied in $|\mathrm{FS}\rangle$, since it

240: accommodates a spin in the projected state. The sites other than $i$

241: and $j$ should be all singly occupied in $|\mathrm{FS}\rangle$.

242: Thus, the probability of finding the holon at site $i$ while the

243: hole of spinon at site $j$, which is denoted as $P_{ij}$, is given

244: by that of finding site $i$ empty, site $j$ doubly occupied, and all

245: other sites singly occupied in $|\mathrm{FS}\rangle$. For $i=j$, in

246: which case a real electron is removed from site $i$, the

247: corresponding probability($P_{ii}$) is given by that of finding site

248: $i$ occupied by a down spin and all other sites singly occupied in

249: $|\mathrm{FS}\rangle$. The ratio between $P_{ij}$ and $P_{ii}$ is

250: given by

251: \begin{equation}

252: \frac{P_{ij}}{P_{ii}}=\frac{\sum_{\beta}|\psi_{\beta}|^{2}}{\sum_{\alpha}|\psi_{\alpha}|^{2}}=

253:    \frac{\sum_{\alpha}|\psi_{\alpha}|^{2}

254:    \frac{|\psi_{\beta}|^{2}}{|\psi_{\alpha}|^{2}}}{\sum_{\alpha}|\psi_{\alpha}|^{2}}.

255: \end{equation}

256: Here, $\psi_{\alpha}$ denotes the amplitude of $|\mathrm{FS}\rangle$

257: for a general configuration $|\alpha\rangle$ in the subspace of no

258: double occupancy, $\psi_{\beta}$ denotes the amplitude for the

259: configuration derived from $|\alpha\rangle$ by moving an electron(of

260: either spin) from site $i$ to site $j$. This statistical sum can be

261: evaluated easily with Variational Monte Carlo method.

262: 

263: To see the value of the thus defined correlation function, we

264: consider some special cases in which the effect of spinon-holon

265: recombination is manifest. As a trivial example, we consider the

266: case of removing an electron from a fully polarized spin background.

267: According to the slave Boson mean field theory, the quasipaticle

268: weight is given by the hole density and vanishes at half filling.

269: However, since the spin is fully polarized, the state in

270: consideration is in fact a mean field state in which the

271: quasiparticle weight should be exactly one. This discrepancy can be

272: easily resolved by noting that $P_{ij}$ is nonzero only for $i=j$ on

273: a fully polarized spin background. This indicates that the added

274: holon and the hole of spinon must occupy the same site and thus

275: recombine into a real electron on that site.

276: 

277: As a less trivial example, we consider a state with

278: antiferromagnetic long range order. As we will show below, $P_{ij}$

279: decays exponentially in this case. The exponential decay of $P_{ij}$

280: can be understood by noting the opening of a SDW gap between

281: configuration $|\beta\rangle$ and $|\alpha\rangle$ in Eq. (12). Thus

282: the added holon and the hole of spinon will form a well defined

283: bound state in such a spin background. Since the bound state has a

284: nonzero overlap with a bare electron, the quasiparticle weight

285: should be finite even at half filling. Calculation in \cite{TKLee}

286: does find a finite quasiparticle weight for such a state.

287: 

288: \begin{figure}[h!]

289: \includegraphics[width=8cm,angle=0]{1d.eps}

290: \caption{Spinon - holon correlation function for the one dimensional

291: projected Fermi sea. The inset show the data in logarithmic scale.}

292: \label{fig1}

293: \end{figure}

294: 

295: Now we present the numerical results. Figure 1 shows the correlation

296: function for the one dimensional projected Fermi sea. The

297: correlation function decays as $1/r$ at large distance and is

298: unnormalizable. Thus the quasiparticle weight should vanish at half

299: filling. The result for the two dimensional d-wave RVB state is

300: shown in Figure 2. The correlation function now decays as

301: $1/r^{\frac{3}{2}}$ at large distance, which is also unnormalizable.

302: Thus the quasiparticle weight should also vanish at half filling in

303: this state. In Figure 3, we show the result for the antiferromagetic

304: ordered state. As mentioned above, the correlation function decays

305: exponentially in this case. However, the short range behavior is

306: quite similar to that of the d-wave RVB state.

307: 

308: \begin{figure}[h!]

309: \includegraphics[width=9cm,angle=0]{dwave.eps}

310: \caption{(a)Spinon - holon correlation function for a d-wave RVB

311: state. The d-wave RVB order parameter $\Delta/t=0.25$. The

312: calculation is done on a $20\times20$ lattice. (b)Power law decay of

313: the spinon - holon correlation function. L denotes the linear size

314: of the lattice, $r_{\mathrm{max}}$ is largest distance that can be

315: defined on a periodic lattice of size L.} \label{fig2}

316: \end{figure}

317: 

318: \begin{figure}[h!]

319: \includegraphics[width=9cm,angle=0]{af.eps}

320: \caption{(a)Same as Figure 2, but now for an antiferromagnetic

321: ordered state. The SDW and d-wave order parameters are

322: $\Delta_{AF}/t=0.1$ and $\Delta/t=0.25$. (b)Exponential decay of the

323: spinon - holon correlation function.} \label{fig3}

324: \end{figure}

325: 

326: The spinon-holon recombination discussed above refers to the

327: correlation between the added particles rather than that between the

328: background particles. At finite doping, a holon condensate is

329: established and the background spinon has a chance of $x$ to behave

330: as a coherent quasipaticle. In fact, the latter is the only way that

331: the system can build a finite quasiparticle weight on the particle

332: side of the spectrum. Following essentially the same steps detailed

333: above, one easily sees that the added slave paricles(or hole of

334: them) during the process of injecting an electron have no

335: correlation with each other. Since the quasiparticle weight due to

336: holon condensation is by definition particle-hole symmetric, the

337: particle-hole asymmetry in the quasiparticle weight should be

338: attributed to the spinon-holon recombination effect discussed in

339: this paper\cite{Yang}.

340: 

341: Especially, while the quasiparticle weight for adding an electron is

342: constrained by the sum rule to vanish at half filling, the

343: quasiparticle weight for removing an electron can be finite at half

344: filling, provided that the added holon and the hole of spion form

345: well defined bound state, as is the case in the antiferromagnetic

346: ordered state. In such a case the quasiparticle can contribute a

347: substantial part of the tunneling asymmetry, although for a full

348: understanding of the latter one also should take into account the

349: contribution from incoherent spectral weight\cite{Anderson1}. When

350: the added holon and the hole of spinon are less tightly bounded, the

351: particle-hole asymmetry in the quasiparticle weight should be less

352: dramatic, as is found in a recent work on the tunneling asymmetry of

353: high temperature superconductors\cite{Yang}.

354: 

355: We now discuss some other effects of spinon-holon recombination. By

356: recombining with a spinon excitation, a hole changes the local spin

357: environment around it. This effect can be invoked to release the

358: kinetic energy of the holes in certain spin background through

359: spontaneous generation of spinon excitations in the system. An

360: example in this respect is provided by a recent variational study on

361: the slightly doped cuprates\cite{TKLee}. In that work, a $t-J$ model

362: with next-nearest-neighboring hoping term is considered. At small

363: doping , it is found that a variational state with spontaneously

364: generated spinon excitations give lower energy than a state without

365: spinon excitations. It is also found that the presence of these

366: spinon excitations changes the charge correlation dramatically. This

367: can also be understood in terms of spinon-holon recombination

368: defined in this paper. A detailed analysis of these issues will

369: appear in separate paper.

370: 

371: Finally, we revisit the problem of supercurrent dissipation. In the

372: presence of spinon excitations, a holon tend to bind with them and

373: form a Fermionic composite object. A holon is thus transformed into

374: a normal charge carrier out of its condensate. The superconductivity

375: is gone when all holons in the condensate are transformed into

376: normal carriers. At low doping, a small number of spinon excitation

377: is enough to kill the superconductivity which is not expected to

378: alter the spin correlation significantly. This is proposed as a

379: mechanism for the pseudogap in underdoped cuprates\cite{Lee2,Wen}.

380: 

381: In conclusion, an explicit definition for spinon-holon recombination

382: is given on the GWF. The thus defined spinon-holon recombination is

383: shown to be responsible for the particle-hole asymmetry in the

384: quasiparticle weight calculated from the GWF and some other subtle

385: differences between the mean field theory and the GWF. A new

386: mechanism for the dissipation of supercurrent in the RVB state is

387: proposed based on the spinon-holon recombination defined in this

388: paper.

389: 

390: This work is supported by NSFC Grant No.90303009.

391: 

392: 

393: \begin{thebibliography}{99}

394: 

395: 

396: \bibitem {Anderson}P. W. Anderson, Science 235 1196 (1987).

397: 

398: \bibitem {Baskran}G. Baskaran Z. Zou and P. W. Anderson, Solid State Commu. 63

399: 973 (1987).

400: 

401: \bibitem {Zou}Z. Zou and P. W. Anderson, Phys. Rev. B37, 627 (1988).

402: 

403: \bibitem {Kotliar}G. Kotliar and J. Liu, Phys. Rev. B 38 5142 (1988).

404: 

405: \bibitem {Lee1}P. A. Lee and N. Nagaosa, Phys. Rev. B 46, 5621-5639 (1992).

406: 

407: \bibitem {Lee2}P. A. Lee and X.-G. Wen, Phys. Rev. Lett. 78, 4111-4114 (1997).

408: 

409: \bibitem {Wen}X.-G. Wen and P. A. Lee, Phys. Rev. Lett. 80, 2193-2196 (1998).

410: 

411: \bibitem {Ng}T. K. Ng, cond-mat/0408374.

412: 

413: \bibitem {Yunoki}S. Yunoki, E. Dagotto, and S. Sorella, Phys. Rev. Lett. 94, 037001 (2005).

414: 

415: \bibitem {Gebhard} F. Gebhard and D. Vollhardt, Phys. Rev. B {\bf 38}, 6911 (1988).

416: 

417: \bibitem {Yang}H-Y.Yang, F. Yang, Y-J. Jiang and T. Li, Cond-mat/0604488.

418: 

419: \bibitem {TKLee}T. K. Lee, C. M. Ho, and N. Nagaosa, Phys. Rev. Lett. 90,

420: 

421: \bibitem {note1}One easily verifies that the only effect of a nonzero pairing

422: term is to replace $|\mathrm{FS}\rangle$ with $|\mathrm{BCS}\rangle$

423: in the following discussion, provided that a cannonical ensamble is

424: used.

425: 

426: \bibitem {Anderson1}P.W. Andeson and N.P. Ong, Cond-mat/0405518.

427: 

428: \end{thebibliography}

429: 

430: 

431: 

432: 

433: 

434: \end{document}

435: