1: %\documentclass[prb,preprint]{revtex4}

2: \documentclass[prb,twocolumn]{revtex4}

3: \usepackage[dvips]{graphicx}

4: \usepackage{latexsym,amsmath,amssymb,bm,euscript}

5: \usepackage[dvips]{color}

6: 

7: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

8: %%% some defs

9: \def\etal{~\textit{et~al.}} % etal

10: \def\ra{\rangle} % bra

11: \def\la{\langle} % ket

12: \def\up{\uparrow}

13: \def\dn{\downarrow}

14: \def\Hc{{\rm H.c.}}

15: 

16: 

17: \begin{document}

18: 

19: \title{D-wave correlated Critical Bose Liquids in two dimensions}

20: 

21: \author{Olexei I. Motrunich}

22: \affiliation{Department of Physics, California Institute of Technology,

23: Pasadena, CA 91125}

24: \author{Matthew P. A. Fisher}

25: \affiliation{Kavli Institute for Theoretical Physics, 

26: University of California, Santa Barbara, CA 93106}

27: 

28: \date{March 9, 2007}

29: 

30: \begin{abstract}

31: 

32: We develop a description of a new quantum liquid phase of interacting 

33: bosons confined in two dimensions which possesses relative D-wave 

34: two-body correlations.  We refer to this stable quantum phase as a 

35: D-wave Bose Liquid (DBL).  The DBL has no broken symmetries, 

36: supports gapless boson excitations which reside on ``Bose surfaces" 

37: in momentum space, and exhibits power law correlation functions 

38: characterized by a manifold of continuously variable exponents.

39: While the DBL can be constructed for bosons moving in the 2d continuum,

40: the state only respects the point group symmetries of the square lattice.

41: On the square lattice the DBL respects all symmetries and does not 

42: require a particular lattice filling.  But lattice effects do allow for 

43: the possibility of a second distinct phase, a quasi-local variant which 

44: we refer to as a D-wave Local Bose Liquid (DLBL).

45: Remarkably, the DLBL has short-range boson correlations and hence no 

46: Bose surfaces, despite sharing gapless excitations and other critical 

47: signatures with the DBL.  Moreover, both phases are metals with a 

48: resistance that vanishes as a power of the temperature.

49: We establish these results by constructing a class of many-particle 

50: wavefunctions for the DBL, which are time reversal invariant analogs of 

51: Laughlin's quantum Hall wavefunction for bosons in a half-filled 

52: Landau level.

53: A gauge theory formulation leads to a simple mean field theory and a 

54: suitable $N-$flavor generalization enables incorporation of gauge field

55: fluctuations to deduce the properties of the DBL/DLBL in a controlled 

56: and systematic fashion.

57: Various equal time correlation functions thereby obtained are in 

58: qualitative accord with the properties inferred from the variational 

59: wavefunctions.  

60: We also identify a promising microscopic Hamiltonian which might manifest

61: the DBL or DLBL, and perform a variational energetics study comparing to 

62: other competing phases including the superfluid.

63: We suggest how the D-wave Bose Liquid wavefunction can be suitably 

64: generalized to describe an itinerant non-Fermi liquid phase of electrons 

65: on the square lattice with a no double occupancy constraint, 

66: a D-wave metal phase.

67: 

68: \end{abstract}

69: 

70: \maketitle

71: 

72: 

73: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

74: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

75: \section{Introduction}

76: \label{sec:intro}

77: 

78: The principal roadblock impeding progress in disentangling the physics 

79: of the cuprate superconductors is arguably our inability to access 

80: quantum ground states of two-dimensional itinerant electrons which are 

81: qualitatively distinct from a Landau Fermi liquid.

82: Overcoming this obstruction is of paramount importance, indispensable 

83: in explaining the strange metal behavior observed near optimal doping 

84: and a likely requisite to account for the emergent pseudo-gap at lower 

85: energies.

86: A putative underlying paramagnetic Mott insulator provides the 

87: scaffolding for a popular class of theories, which view the pseudo-gap 

88: as a lightly doped spin liquid.

89: \cite{Anderson, Baskaran, KotliarLiu, IoffeLarkin, LeeNagaosaWen}  

90: Significant progress has been made in developing the groundwork and 

91: there now exists a well established theoretical framework to describe a 

92: myriad of distinct spin liquids.  For the cuprates the most promising

93: spin liquids are described in terms of fermionic spinons minimally 

94: coupled to a compact $U(1)$ gauge field and moving in various 

95: background fluxes.

96: Upon doping formidable challenges arise.  Bosonic holons carrying the 

97: electron charge become mobile carriers and lead to electrical conduction.

98: But at low temperatures Bose condensation appears inevitable and this 

99: leads to Fermi liquid behavior, either a metal with conventional Landau 

100: quasiparticles or a  BCS superconductor if the spinons are paired.

101: Accessing a pseudo-gap or a strange metal which conduct electricity 

102: despite the absence of long-lived Landau quasiparticles requires doped 

103: holons that form an uncondensed quantum Bose fluid rather than a 

104: condensed superfluid.  But is this possible, even in principle?

105: If possible, what properties would such a putative ``Bose metal'' 

106: exhibit?\cite{Feigelman, Dalidovich, Galitski, Alicea}

107: And what theoretical framework is appropriate?

108: The slave-particle gauge theory approach has been stymied by this 

109: stumbling block for over 15 years.  

110: In this paper we provide (some) answers to these questions by 

111: constructing explicit examples of such unusual phases of bosons

112: that may offer some routes out of the conundrum.

113: 

114: Our goal, then, is to access and explore uncondensed quantum phases of

115: 2d bosons which are conducting fluids but not superfluids.

116: Specifically we have in mind hard core bosons moving on a 2d square 

117: lattice, but seek to construct states which do not require particular 

118: commensurate densities.  While our construction can be implemented for 

119: bosons moving in the 2d continuum Euclidean plane, the states will only 

120: possess the reduced point group symmetry of the square lattice.  

121: Despite our interest in time reversal invariant quantum ground states, 

122: our technical approach will be strongly informed by theories of the 

123: fractional quantum Hall effect (FQHE).  In some regards, the quantum 

124: phases that we construct are time reversal invariant analogs of the 

125: Laughlin state for bosons in a half-filled Landau level.  

126: But the physical properties of the phases will be dramatically different 

127: from the incompressible FQHE states, and will have gapless excitations 

128: and metallic transport for example.

129: 

130: To motivate and illustrate our approach, it will be helpful to briefly 

131: revisit the bosonic FQHE.  Consider the Laughlin wavefunction for bosons 

132: in a half-filled Landau level,

133: \begin{equation}

134: \Psi_{\nu = 1/2} (z_1, z_2, \dots, z_N) = \prod_{i<j} (z_i - z_j)^2 ~.

135: \label{Psi_nu_1/2}

136: \end{equation}

137: Much of the physics of the Laughlin state has its origin in the structure

138: of zeroes, which reveals that any two particles upon close approach in 

139: real space are in a relative two-body $d+id$ state, $(z_i-z_j)^2$.  

140: Our first objective is to construct a time reversal invariant (real) 

141: wavefunction in which particle pairs are similarly in a relative 

142: $d_{xy}$ state.  

143: A clue is offered by noting that the Laughlin wavefunction is the square 

144: of a Vandermonde determinant, $\Psi_{\nu = 1/2} = (\det {\bf V})^2$.  

145: In the Vandermonde determinant, all pairs of particles are in a 

146: relative $p+ip$ state, $(z_i-z_j)$.  Upon squaring, the two $p+ip$ states

147: combine to form a single $d+id$ state, which is essentially just 

148: addition of angular momentum.

149: 

150: This suggests constructing a time reversal invariant boson wavefunction 

151: in zero magnetic field by simply squaring a determinant constructed from 

152: momentum states within a Fermi sea,

153: \begin{equation}

154: \Psi({\bf r}_1, {\bf r}_2, \dots, {\bf r}_N) 

155: = ( \det e^{i {\bf k}_i \cdot {\bf r}_j} )^2 ~, \quad

156: \text{(S-type)}.

157: \label{Psi_det-squared}

158: \end{equation}

159: Let's examine the nodal structure.  The generic behavior of each 

160: fermion determinant when any two particles are taken close together 

161: is dictated by Fermi statistics and reality of the wavefunction

162: and has the functional form $({\bf r}_i -{\bf r}_j) \cdot \hat{\bm l}$.

163: The unit vector $\hat{\bm l}$ will depend in a complicated way on the 

164: location of all the other particles 

165: (see Ref.~\onlinecite{Ceperley91} for a discussion and illustrations of 

166: the free fermion nodes).

167: When $\hat{\bm l} = \hat{\bf y}$, this is a $p_x$ form vanishing along a 

168: nodal line (in the relative coordinate) parallel to the $x$-axis.

169: Unfortunately, in contrast with the Laughlin case, 

170: squaring the determinant leads here to an ``extended" $s$-wave form 

171: with a quadratic nodal line rather than the desired $d$-wave.

172: 

173: But consider instead multiplying together two fermion determinants, 

174: each constructed by filling up a Fermi sea of momentum states,

175: but with different Fermi surfaces.  

176: A wavefunction with $d_{xy}$ two-particle correlations

177: can be constructed by choosing two elliptical Fermi surfaces, one with 

178: its long axis along the $x$-axis and the other rotated by 90 degrees,

179: as illustrated in Fig.~\ref{fig:introFS},

180: \begin{equation}

181: \Psi({\bf r}_1, {\bf r}_2, \dots, {\bf r}_N) = 

182: (\det)_x \times (\det)_y ~, \quad \text{(D$_{xy}$-type)} ~,

183: \label{Psi_DBL}

184: \end{equation}

185: where the short-hands $(\det)_x$ and $(\det)_y$ represent the 

186: corresponding Slater determinants.

187: In the limit of extreme eccentricity of the ellipses, the wavefunction 

188: will have the desired $d_{xy}$ form when two particles are brought 

189: close together, $(x_i-x_j)(y_i-y_j)$.  Away from this limit, the two 

190: nodal lines will not align precisely with the $x$ and $y$ axes,

191: but upon taking one particle around the other the wavefunction will 

192: exhibit the same sign structure as a $d_{xy}$ orbital, $+ - + -$,

193: changing sign twice. 

194: A picture of such sign structure as seen by a test particle is shown in 

195: Fig.~\ref{fig:nodes}.

196: 

197: 

198: \begin{figure}

199: \centerline{\includegraphics[width=\columnwidth]{figures/introFS.eps}}

200: \vskip -2mm

201: \caption{DBL wavefunction Eq.~(\ref{Psi_DBL}) is obtained by multiplying 

202: two Slater determinants corresponding to distinct but symmetry related

203: Fermi surfaces, which can be viewed as Fermi surfaces of $d_1$ and 

204: $d_2$ slave fermions.

205: The left panel shows an example with closed elliptical Fermi surfaces, 

206: while the right panel is a case with open Fermi surfaces in the first 

207: Brillouin zone of the square lattice.

208: }

209: \label{fig:introFS}

210: \end{figure}

211: 

212: 

213: \begin{figure}

214: \centerline{\includegraphics[width=3.0in]{figures/nodes.eps}}

215: \vskip -2mm

216: \caption{``Nodal picture'' of the continuum DBL wavefunction as probed 

217: by moving a single particle (whose initial location is marked by the 

218: filled circle) while keeping the rest of the particles fixed

219: (open circles).

220: The signs of the wavefunction are indicated in several places to 

221: bring out the $+ - + -$ pattern upon encircling a target particle.

222: }

223: \label{fig:nodes}

224: \end{figure}

225: 

226: 

227: Our principal thesis is that this wavefunction captures qualitative 

228: features of a new quantum liquid phase of bosons, which we call

229: ``D-wave-correlated Bose Liquid'' (DBL).

230: But a variational wavefunction does not provide a complete 

231: characterization of a quantum phase, and cannot be used to address its 

232: stability.  As for the FQHE, a field theoretic approach, such as 

233: Chern-Simons gauge theory, is both desirable and ultimately necessary.

234: Since we require time reversal invariance, Chern-Simons is inappropriate,

235: but a tractable field theoretic framework for the DBL will nevertheless

236: turn out to be a gauge theory.

237: 

238: Indeed, we follow closely the gauge theory approaches to spin liquids in 

239: quantum antiferromagnets,\cite{LeeNagaosaWen, WenPSG} which after all are 

240: lattice bosonic systems.  But there are some notable and important differences 

241: for the DBL which we discuss below.  The variational wavefunction in 

242: Eq.~(\ref{Psi_DBL}), being a product of two fermion determinants, naturally 

243: suggests expressing the boson creation operator as a product of two fermion 

244: operators,

245: \begin{equation}

246: b^\dagger ({\bf r}) = d_1^\dagger ({\bf r}) d_2^\dagger ({\bf r}) ~,

247: \end{equation}

248: with the 2d position ${\bf r}$ either continuous or denoting the 

249: discrete sites of a square lattice.

250: In a general case, this decomposition introduces a local $SU(2)$ gauge 

251: redundancy well-known in the slave-fermion treatments of the 

252: spin-1/2 antiferromagnet 

253: (see Ref.~\onlinecite{WenPSG} for a recent comprehensive discussion).

254: The S-type wavefunction corresponds to a so-called $SU(2)$ liquid,

255: while the DBL wavefunctions are $U(1)$ liquids and require a $U(1)$ 

256: gauge field which is minimally coupled to the two fermions with 

257: opposite gauge charges (the product $d_1^\dagger d_2^\dagger$

258: is then a gauge neutral composite, the physical boson).  

259: 

260: Within slave particle theory a mean field state with a fixed background 

261: gauge ``magnetic" flux is chosen.

262: This choice is constrained by symmetry:  All the physical symmetries 

263: of the boson model, even if not respected by the mean field ansatz, 

264: must be present in the full gauge theory once gauge transformations are 

265: allowed.

266: The collection of these symmetry transformations in the gauge theory

267: is called the Projective Symmetry Group (PSG).\cite{WenPSG}  

268: For lattice boson theories many different mean field states with

269: different PSG's are possible, while little is offered to guide

270: which mean field state is the correct one.

271: 

272: In the continuum, the symmetries of the Euclidean plane together with 

273: time reversal invariance are more restrictive and we can only offer 

274: two PSG mean field ansatze in this case.

275: One is a mean field state with zero gauge flux and spherical Fermi seas

276: for both fermions; this is an $SU(2)$ ansatz and corresponds to the 

277: S-type wavefunction considered earlier.

278: A different $U(1)$ ansatz is obtained by considering $d_1$ and $d_2$ 

279: fermions each moving in a uniform field but of opposite signs for the 

280: two; in this case, the boson wavefunction is schematically

281: $\Psi = (\det) \times (\det)^*$,

282: e.g., $\Psi = |\det {\bf V}|^2 = |\Psi_{\nu=1/2}|$ to give a concrete 

283: example;

284: this wavefunction is again non-negative and has off-diagonal

285: quasi-long-range order.\cite{GirvinMacDonald, Kane}

286: 

287: For bosons moving in continuous space but with only the square lattice 

288: point group symmetry, we can offer only four PSG mean fields; 

289: all of them have zero flux, but differ in the way the 90 degree rotation 

290: and mirror symmetries are realized.

291: One example has both $d_1$ and $d_2$ Fermi surfaces individually

292: respecting the symmetries but otherwise independent of each other.  

293: Two more examples are the already introduced D$_{xy}$ state,

294: Eq.~(\ref{Psi_DBL}) and Fig.~\ref{fig:introFS}, with elliptical 

295: Fermi surfaces elongated along the $x$ and $y$ axes, 

296: and a similarly constructed D$_{x^2-y^2}$ state with ellipses along 

297: $\hat{\bf x} \pm \hat{\bf y}$;

298: in both cases, the $d_1$ and $d_2$ Fermi surfaces are related by

299: the 90 degree rotation, but the two states differ in the way the mirror 

300: symmetries are realized.

301: In the fourth example, each Fermi surface is invariant under the rotation

302: but not under the mirrors, which instead are realized by interchanging 

303: the two fermions.

304: 

305: Each of the four presented ansatze can be loosely called a D-wave Bose 

306: liquid as far as the nodal pictures like that in Fig.~\ref{fig:nodes}

307: are concerned; the gauge theory analysis would also be rather similar.

308: From now on, we will focus on the D$_{xy}$ boson liquid.

309: The associated mean field wavefunction once Gutzwiller projected into

310: the physical Hilbert space, $d_1^\dagger d_1 = d^\dagger_2 d_2$, 

311: is precisely the variational wavefunction in Eq.~(\ref{Psi_DBL}).

312: A full incorporation of gauge fluctuations about the mean field 

313: performs, in principle, this projection exactly.  But in practice the gauge theory approach allows incorporation of 

314: slowly varying gauge fluxes and is different in this respect from

315: the wavefunction.  

316: Provided the gauge theory is not in a confined state and no physical 

317: symmetries are broken, an exact description of the putative DBL phase 

318: is thereby obtained.  

319: Herein, we introduce a large $N$ generalization of the gauge 

320: theory which allows a controlled and systematic treatment of the gauge 

321: fluctuations order by order in powers of $1/N$.

322: We can then perturbatively extract physical properties of the DBL phase.

323: Whether the DBL phase survives down to $N=1$ is not something that 

324: we can reliably address.

325: 

326: In describing the D$_{xy}$ liquid phase for lattice bosons there is 

327: considerable freedom when choosing the precise form of the slave fermion 

328: hopping in the mean field Hamiltonian.  The simplest choice is near 

329: neighbor hopping with different amplitudes in the $x$ and $y$ directions.

330: But even with this restriction, there are two different possible 

331: Fermi surface topologies, closed or open, as depicted in 

332: Fig.~\ref{fig:introFS}.  

333: The Fermi surface topology has rather dramatic consequences 

334: for the nature of the associated D-wave Bose fluid.

335: When the Fermi surfaces are open and have no parallel Fermi surface 

336: tangents, the resulting phase has a quasi-local character -- 

337: the boson Green's function is found to fall off exponentially in space.

338: This is a distinct quantum phase which we refer to as a 

339: D-Wave Local Bose liquid (DLBL).

340: The DLBL is intrinsically very stable against gauge fluctuations and 

341: we are fairly confident that when present it will be a stable quantum 

342: phase.  The effect of gauge fluctuations on the DBL phase, however, 

343: are more subtle.   Within our systematic large $N$ approach we do 

344: find a regime where the DBL is stable, but it is less clear that 

345: this will survive down to $N=1$.

346: The Gutzwiller wavefunction does appear to describe a DBL phase with 

347: properties that are consistent with those inferred from the gauge theory.

348: This provides some support for the stability of the DBL phase.

349: 

350: Together the gauge theory and the variational wavefunction provide a 

351: consistent and rather complete picture of both the DBL and the DLBL.

352: Here we briefly highlight some of the important characteristics.

353: In the DBL, the single particle boson Green's function, 

354: $G_b({\bf r}, \tau) \equiv \la b^\dagger ({\bf r},\tau) b({\bf 0},0) \ra$,

355: decays as an oscillatory power law at equal times,

356: \begin{eqnarray}

357: G_b({\bf r},0) &\sim & 

358: \frac{\cos[({\bf k}_{F_1} - {\bf k}_{F_2}) \cdot {\bf r}] }

359:      {|{\bf r}|^{4-\eta}} \nonumber \\

360: &+& 

361: \frac{\cos[({\bf k}_{F_1} + {\bf k}_{F_2}) \cdot {\bf r} - 3\pi/2]} 

362:      {|{\bf r}|^4} ~,

363: \label{Gb_final}

364: \end{eqnarray}

365: where the two wave vectors ${\bf k}_{F_{1,2}}$ depend on the observation 

366: direction $\hat{\bf r}$, as does the anomalous exponent $\eta > 0$.

367: As the direction of $\hat{\bf r}$ is rotated in real space, the 

368: wave vectors ${\bf k}_{F_{1,2}}$ and ${\bf k}_{F_1} \pm {\bf k}_{F_2}$ 

369: trace out closed momentum space curves, and for the DBL in the continuum 

370: these curves will have the topology of a circle.

371: 

372: If one measures the boson Green's function and finds it to be of the

373: above form, it is natural to refer to the 

374: ${\bf k}_{F_1} \pm {\bf k}_{F_2}$ as ``Bose surfaces.''

375: Within the gauge theory description the origin of these singular surfaces

376: can be traced to the Fermi surfaces of the constituent fermions.

377: Specifically, ${\bf k}_{F_{1,2}}$ are locations on the two Fermi 

378: surfaces where the surface normals are along the observation 

379: direction $\hat{\bf r}$, and the momentum space areas enclosed by 

380: each of these surfaces will equal $(2\pi)^2 \rho_b$.

381: Note that the ${\bf k}_{F_{1,2}}$ surfaces can be uniquely

382: reconstructed from the measured Bose surfaces 

383: ${\bf k}_{F_1} \pm {\bf k}_{F_2}$.

384: In this sense, the Bose surfaces in the DBL contain information

385: analogous to Luttinger's Fermi surface volume theorem in the 

386: Fermi liquid.  The Bose surface can also be extracted directly 

387: from the DBL wavefunction by using variational Monte Carlo to compute 

388: the equal time boson Green's function, and can also be indirectly 

389: inferred by using the shape of the two Fermi surfaces input into the 

390: $(\det)_x$ and $(\det)_y$ factors.  It is both reassuring and quite 

391: remarkable that these two coincide.

392: 

393: % The spatially local Green's function in the DBL falls off in time as,

394: % \begin{equation}

395: % G_b({\bf 0}, \tau) \sim \frac{1}{|\ln(\tau)|^{1/2}} 

396: % |\tau|^{-(2-3\eta_0/2)} ~.

397: % \end{equation}

398: % Here $\eta_0$ is the maximum value of the direction dependent exponent

399: % $\eta(\hat{\bf r})$.  The local boson tunneling density of states 

400: % vanishes as a power of energy, up to logarithmic factors, 

401: % $A_b(E) \sim E^{1 - 3\eta_0/2}$.

402: 

403: The spatially local Green's function, $G_b({\bf 0}, \tau)$, in the DBL 

404: falls off in time as a power law which is at least as slow as 

405: $1/\tau^2$ corresponding to the local boson tunneling density of states 

406: $A_b(E) \sim E$;

407: these are mean field results obtained by combining two fermions each 

408: with a finite density of states, but we suspect the time decay may

409: actually be slower upon including gauge fluctuations.

410: 

411: 

412: The behavior of the boson Green's function in the DLBL is quite 

413: different.

414: The time decay at ${\bf r}=0$ is a power law, 

415: $G_b({\bf 0}, \tau) \sim \tau^{-2}$,

416: which is an exact result insensitive to lattice scale details.

417: But the equal time Green's function falls off exponentially in space

418: in the DLBL, $G_b({\bf r}, 0) \sim \exp(-r/\xi)$.

419: Despite this, a two boson box correlator,

420: \begin{equation}

421: {\cal B}_b (x) \equiv

422: \la b^\dagger(0,0) b^\dagger(x,x) b(x,0) b(0,x) \ra ~,

423: \label{Bb_def}

424: \end{equation}

425: falls off as a (non-oscillatory) power law at large distances in the 

426: DLBL, ${\cal B}(x) \sim -x^{-8}$, with both the sign (negative) and the 

427: exponent being universal and insensitive to lattice scale physics.

428: Paradoxically, this seems to imply that a pair of bosons injected into 

429: the DLBL on opposite corners of a square box can move more readily 

430: than a single injected boson.

431: But in a strongly interacting quantum state such a single-particle 

432: interpretation can be misleading -- the dynamics of an injected boson 

433: is not that of a weakly interacting quasiparticle.

434: 

435: In both the DBL and the DLBL the density-density correlator

436: $D_b({\bf r}) \equiv

437:  \la \hat{\rho}_b({\bf r}) \hat{\rho}_b({\bf 0}) \ra$ 

438: behaves at large distances as,

439: \begin{equation}

440: D_b({\bf r}) \sim - \sum_{\alpha = 1,2} 

441: \frac{\cos[2{\bf k}_{F_\alpha} \cdot {\bf r} - 3\pi/2]}

442:       {|{\bf r}|^{4-\gamma_\alpha}}

443: + \frac{1}{|{\bf r}|^4} ~,

444: \label{Db_final}

445: \end{equation}

446: where again both the wave vectors ${\bf k}_{F_{1,2}}$ and the anomalous 

447: exponent $\gamma_\alpha$ depend on the observation direction 

448: $\hat{\bf r}$.   Measurement of such correlation can be used to extract 

449: ${\bf k}_{F_{1,2}}$ in the DLBL, which are not accessible from the boson 

450: Green's function here.

451: Upon rotating the unit vector $\hat{\bf r}$ in real space, 

452: ${\bf k}_{F_{1,2}}$ will again trace out the Fermi surfaces of the 

453: underlying fermions which for the DLBL will be open.

454: Both the DBL and the DLBL are conductors with a resistance varying with 

455: temperature as $R \sim T^{4/3}$.  This is particularly striking for the

456: DLBL where the boson Green's function is short-ranged.  

457: Evidently, it is not possible to understand the metallic transport in 

458: terms of the motion of weakly interacting bosons.  In a sense it is the 

459: fermionic constituents which transport the charge, but even this is not 

460: quite right since the $d_1,d_2$ fermions do not exist as well defined

461: quasiparticle excitations, being strongly scattered by the gauge 

462: fluctuations.

463: 

464: The lattice gauge theory formulation can be used to motivate a lattice

465: boson Hamiltonian which might plausibly have the DBL/DLBL as a ground 

466: state.  The simplest such model consists of a near neighbor boson 

467: hopping term supplemented by a four-site ``ring" exchange term:

468: \begin{eqnarray}

469: \label{Hring}

470: H &=& H_J + H_4 ~,\\

471: H_J &=& -J \sum_{{\bf r};\, \hat{\mu} = \hat{\bf x}, \hat{\bf y}}

472: (b^\dagger_{\bf r} b_{{\bf r} + \hat{\mu}} + h.c.) ~,\\

473: H_4 &=& K_4 \sum_{\bf r} 

474: (b^\dagger_{\bf r} b_{{\bf r} + \hat{\bf x}}

475: b^\dagger_{{\bf r} + \hat{\bf x} + \hat{\bf y}} b_{{\bf r} + \hat{\bf y}}

476: + h.c.) ~,

477: \end{eqnarray}

478: with $J,K_4 \ge 0$.  This model is fully specified by two dimensionless 

479: numbers, the ratio $J/K$ and the boson filling $0 \le \rho_b \le 1$.

480: This ring Hamiltonian with $J>0$ and $K_4<0$ was introduced and studied 

481: by Paramekanti\etal,\cite{Paramekanti} and later studied extensively

482: with quantum Monte Carlo by Sandvik\etal,\cite{Sandvik}

483: Melko\etal,\cite{Melko} and Rousseau\etal\cite{Rousseau}

484: There is no sign problem in this regime, and it was possible to

485: access large system sizes and low temperatures.

486: But for $J>0$ and $K_4>0$, there is a sign problem since the Hamiltonian 

487: does not satisfy the Marshall sign conditions, and one expects the 

488: ground state wavefunction to take both positive and negative values.

489: For this ring model we have evaluated the variational energetics

490: for the DBL/DLBL wavefunctions and compared these to the energy of a 

491: superfluid wavefunction of the usual Jastrow form.

492: Within this necessarily limited energetics study we do find a region 

493: of parameter space with small $J/K_4$ and near half filling where the 

494: DLBL wavefunction has the lowest energy.  In view of its quasi-local 

495: character,  we suspect that DMRG studies could be fruitful in helping 

496: establish whether or not the DLBL is present in the phase diagram of 

497: this (or related) ring Hamiltonian.

498: 

499: 

500: The paper is organized as follows.

501: In Sec.~\ref{sec:motiv_Wavefnc} we discuss in more detail the 

502: wavefunction motivation for considering the DBL states.

503: In Sec.~\ref{sec:MFT} we introduce the lattice gauge theory

504: description and study properties of the DBL state in the 

505: mean field theory that ignores gauge fluctuations.

506: This provides an initial guide to the singular (Bose) surfaces,

507: and is followed in Sec.~\ref{sec:Wavefnc_props} with numerical

508: characterization of the properties of the actual DBL wavefunctions.

509: In Sec.~\ref{sec:Gauge_fluct}, we consider the full gauge theory

510: description, focusing on the effects of the gauge fluctuations on

511: the singularities across the Bose surfaces, and in particular

512: obtain the long-distance properties such as

513: Eq.~(\ref{Gb_final}) and Eq.~(\ref{Db_final}) to order $1/N$ in our systematic

514: large $N$ approach.

515: In Sec.~\ref{sec:eff_FT} we address the issue of the stability

516: of the DBL by considering a putative fixed point theory.

517: We conclude in Sec.~\ref{sec:concl} with a discussion of physical

518: properties and possible future directions.

519: 

520: 

521: 

522: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

523: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

524: \section{Boson Wavefunctions and Nodes}

525: \label{sec:motiv_Wavefnc}

526: 

527: In this section, we expand on the initial motivation for considering

528: the DBL states as a way to perform (a kind of) flux attachment 

529: transformation in a time reversal invariant manner.

530: We also discuss the nodal structure of the bosonic wavefunctions,

531: providing some justification for the qualifier ``D-wave'' in the 

532: suggested wavefunction names in this work.  

533: To this end, we consider a ``relative single particle wavefunction"

534: -- more precisely, a cross-section of the many-body wavefunction --

535: defined as follows.

536: Fixing the positions of all the particles except one, we define a 

537: function which depends explicitly on the coordinates of the one 

538: ``test" particle, and implicitly on the coordinates of all the other 

539: particles,

540: \begin{equation}

541: \Phi_{{\bf r}_2, \dots, {\bf r}_N}({\bf r}) \equiv

542: \Psi({\bf r}, {\bf r}_2, \dots, {\bf r}_N) ~.

543: \label{Phi_b}

544: \end{equation}

545: For notational ease we will henceforth drop the implicit dependence 

546: on the $2(N-1)$ spatial coordinates and just use the notation 

547: $\Phi_b({\bf r})$.

548: 

549: 

550: \subsection{Laughlin $\nu=1/2$ revisited}

551: \label{subsec:Laughlin}

552: Consider the Laughlin wavefunction for bosons in a half-filled Landau 

553: level, Eq.~(\ref{Psi_nu_1/2}).  In this case, the relative single

554: particle wavefunction $\Phi(z)$ is complex and has double strength 

555: zeroes at the positions of all the other $N-1$ particles.

556: Upon close approach to a specific particle, say $z_i$, the function

557: $\Phi(z) \sim (z-z_i)^2$ is of a $d+id$ form.  It is in this sense

558: that the $\nu=1/2$ Laughlin state can be viewed as a D-wave fluid:

559: All pairs of particles, upon close approach in real space, are in a 

560: two-body $d+id$ state.

561: Our goal is to construct a time reversal invariant analog of the 

562: $\nu=1/2$ Laughlin state in which particle pairs are similarly in

563: a relative $d_{xy}$ or $d_{x^2-y^2}$ state.

564: 

565: In thinking about Laughlin states, it has been particularly instructive 

566: to view them in terms of composite particles created by 

567: ``flux attachment."

568: For example, if one re-expresses the $\nu=1/2$ Laughlin state as,

569: \begin{equation}

570: \Psi_{\nu=1/2} = \prod_{i<j} (z_i - z_j) \, \Psi_{cf} ~,

571: \end{equation}

572: the ``composite fermion" wavefunction is simply a (Vandermonde) 

573: determinant, $\det {\bf V}$, of filled Landau level orbitals,

574: \begin{equation}

575: \Psi_{cf} = \det {\bf V} = \prod_{i<j} (z_i - z_j) ~,

576: \end{equation}

577: with $V_{ij} = (z_i)^j$.

578: In a second quantized framework, flux attachment can be achieved by 

579: introducing a Chern-Simons gauge field\cite{Zhang_CS}.

580: Chern-Simons gauge theory has been a useful tool to describe the full 

581: Haldane-Halperin hierarchy of fractional quantum Hall states,

582: encoding the fractional charge and statistics of the quasiparticles as

583: well as the structure of the edge excitations\cite{MPAF_Lee, Wen_edge}.  But in discussing bosons

584: in a time reversal invariant setting (zero magnetic field), 

585: flux-attachment techniques are problematic.  For instance, the usual 

586: flux smearing mean field approximation is likely to inadvertently

587: break time reversal invariance.  

588: 

589: In seeking to avoid Chern-Simons theory, it is worth noting that

590: the $\nu=1/2$ state is a perfect square -- a square of the Vandermonde 

591: determinant,

592: \begin{equation}

593: \Psi_{\nu =1/2} = (\det {\bf V})^2 ~.

594: \label{detV2}

595: \end{equation}

596: This suggests that the $\nu=1/2$ Laughlin state can be fruitfully 

597: described within a slave particle framework.\cite{Wen_proj4FQHE}

598: Consider decomposing the boson creation operator as a product of two 

599: fermions,

600: $b^\dagger ({\bf r}) = d_1^\dagger ({\bf r}) d_2^\dagger ({\bf r})$.

601: As discussed in the Introduction, there is a local gauge redundancy,

602: and a correct treatment requires the presence of a gauge field 

603: minimally coupled to both fermions.

604: In the slave particle mean field approach to the $\nu=1/2$ boson problem,

605: one simply drops the gauge field, obtaining a problem of two fermion 

606: flavors each moving in a magnetic field.

607: If the electrical charge of the boson is divided equally, each fermion 

608: flavor is effectively in a full Landau level.

609: The mean field wavefunction is,

610: \begin{equation}

611: \Psi^{MF}_{\nu = 1/2} = \prod_{i<j} (z^{(1)}_i - z^{(1)}_j) \times 

612: \prod_{i<j} (z^{(2)}_i - z^{(2)}_j) ~.

613: \end{equation}

614: To obtain a wavefunction for the bosons it is necessary to project

615: into the physical Hilbert space, and this is achieved here by simply 

616: setting $z^{(1)}_i = z^{(2)}_i = z_i$.

617: One thereby recovers the Laughlin state as a square of the 

618: Vandermonde determinant, Eq.~(\ref{detV2}).

619: 

620: 

621: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

622: \subsection{Time Reversal Invariant Wavefunctions}

623: \label{subsec:det-squared}

624: 

625: Within a first quantized framework, the relative simplicity of a time 

626: reversal invariant bosonic superfluid as compared to the Laughlin state 

627: is the nodelessness of the ground state wavefunction.  

628: As Feynman argued many years ago, for non-relativistic bosons moving 

629: in the continuum with an interaction only depending on the particle 

630: positions, the kinetic energy of any bosonic wavefunction which has 

631: sign changes could be reduced by making all the signs positive while 

632: keeping the magnitude of the wavefunction fixed to leave the potential 

633: energy unchanged.

634: The ground state wavefunction should thus be everywhere non-negative.  

635: For the non-interacting Bose gas the ground state wavefunction is just

636: $\Psi =1$, but in the presence of interactions a popular variational

637: wavefunction is of the Jastrow form,\cite{Kane}

638: \begin{eqnarray}

639: \Psi_{\rm Jastrow}({\bf r}_1, {\bf r}_2, \dots, {\bf r}_N) 

640: \;\propto\; e^{-\sum_{i<j} u({\bf r}_i - {\bf r}_j)} ~,

641: \label{Psi_Jastrow}

642: \end{eqnarray}

643: with variational freedom in the two-particle pseudopotential 

644: $u({\bf r}_i - {\bf r}_j)$, which is usually taken to approach zero as 

645: $1/|{\bf r}|^p$ at large separations.

646: 

647: As in the case with a magnetic field present, we again consider

648: the relative single particle wavefunction.

649: For a Bose condensed  superfluid in a time reversal invariant system 

650: $\Phi_b({\bf r})$ can be taken as real and is  

651: everywhere non-negative.  If there are repulsive interactions between

652: the bosons in the superfluid, the amplitude of $\Phi_b({\bf r})$ will be 

653: reduced when the test particle is taken nearby another particle, 

654: which is implemented by the Jastrow pseudopotential in 

655: Eq.~(\ref{Psi_Jastrow}).

656: But the sign of $\Phi_b({\bf r})$ will remain positive, so in some sense 

657: all of the particle pairs are in a relative S-type state.

658: In the special case of a hard core interaction,

659: $\Phi_b({\bf r}) \to 0$ for ${\bf r} \to {\bf r}_i$, 

660: so one can then view this as an ``extended" S-type wavefunction.

661: 

662: Motivated by the preceding discussion of the $\nu=1/2$ Laughlin state,

663: in Eq.~(\ref{Psi_det-squared}) we introduced a simple time reversal 

664: invariant wavefunction for hardcore bosons which is the square of a 

665: determinant for free fermions filling a Fermi sea.

666: By construction this wavefunction is non-negative.  

667: Moreover, the wavefunction will have zeroes which coincide with the 

668: nodes of the filled Fermi sea,

669: $\Psi_{FS}({\bf r}_1, {\bf r}_2, \dots, {\bf r}_N) 

670: = \det e^{i {\bf k}_i \cdot {\bf r}_j}$.  

671: With time reversal invariance, a relative single fermion wavefunction, 

672: $\Phi_f({\bf r}) \equiv \Psi_{FS}({\bf r}, {\bf r}_2, \dots, {\bf r}_N)$, 

673: can be taken as real.

674: As a result, the nodal structure of $\Phi_f({\bf r})$ will be 

675: qualitatively different than for the filled Landau level state,

676: vanishing along nodal lines rather than at isolated points.  

677: These nodal lines will pass through the positions of all of the other 

678: fermions.\cite{Ceperley91}  

679: Upon taking the ``test" particle across a nodal line, the function 

680: $\Phi_f({\bf r})$ changes sign, vanishing linearly upon approaching the 

681: nodal line.

682: If one takes the position of the test particle close to another particle,

683: $\Phi_f({\bf r})$ will have a $p$-wave character -- in particular a 

684: $p_x$ form if we define the $x$-axis as being parallel to the 

685: nodal line.

686: 

687: Despite the $p$-wave character of the free fermion determinant, the 

688: determinant-squared wavefunction will not have a $d$-wave character. 

689: Rather, the relative single boson wavefunction will vanish 

690: quadratically upon crossing the nodal lines of the fermion determinant.

691: When the test particle is taken near another particle ${\bf r}_i$, 

692: the function vanishes quadratically,

693: $\Phi_b({\bf r}) \sim (x - x_i)^2$.  

694: The ``pair" wavefunction is thus of an ``extended S-type" form, 

695: vanishing along the residual nodal line.

696: 

697: Let us now consider the DBL many-body wavefunction Eq.~(\ref{Psi_DBL}), 

698: which is the product of two different Slater determinants for fermions 

699: that fill elliptical Fermi surfaces as shown in Fig.~\ref{fig:introFS}.

700: The nodal structure is revealed by exploring $\Phi_b({\bf r})$.

701: This function will have two sets of nodal lines, 

702: one from each of the determinants.  Due to the elliptical nature of the 

703: respective Fermi surfaces, the two sets of lines, both of which pass 

704: through all of the particles, will generally not coincide with one 

705: another.  Indeed, the fermion determinant coming from an elliptical 

706: Fermi surface will have nodal lines running preferentially perpendicular 

707: to the long axis of the ellipse.  Focusing on the behavior of 

708: $\Phi_b({\bf r})$ near a target particle ${\bf r}_i$, one anticipates a 

709: behavior of the $d_{xy}$ form,

710: $\Phi_b({\bf r}) \sim (x - x_i)(y - y_i)$.

711: Here we have assumed that the two nodal lines actually coincide with the

712: $x$ and $y$ axes.  In general, for a typical configuration of fixed 

713: particle coordinates and a given target particle, this will not 

714: precisely be the case.  More generically, the two nodal lines will 

715: intersect a particle at two angles which are not aligned with the axes.

716: But the sign of the relative wavefunction when the test particle 

717: encircles the target particle will still behave as $+ - + -$, 

718: the same sign structure as a D$_{xy}$ or D$_{x^2-y^2}$ orbital.  

719: This is illustrated in Fig.~\ref{fig:nodes}.

720: 

721: Since the DBL many-body wavefunction is not nodeless, it cannot be the 

722: ground state of a continuum Hamiltonian of bosons.  

723: If we put the coordinates on the sites of a 2d square lattice, 

724: the ground state wavefunction is only assured to be non-negative if the 

725: sign and form of the hopping matrix in the lattice tight binding 

726: Hamiltonian is such that it satisfies the Marshall sign conditions -- 

727: the requirement that a choice of gauge is possible to make all of the 

728: off-diagonal matrix elements negative.  

729: In Sec.~\ref{sec:Energetics} we consider a particular Marshall sign 

730: violating lattice boson Hamiltonian which might exhibit a ground state 

731: of the proposed D-wave form.

732: 

733: 

734: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

735: \subsection{Precedents of $(\det)_1 \times (\det)_2$ wavefunctions for 

736: spin liquids}

737: Before focusing solely on the DBL states, 

738: we want to mention that our construction of time reversal invariant 

739: bosonic wavefunctions as a product of two distinct determinants has 

740: nice precedents in the studies of spin liquids on the triangular 

741: lattice.  One can view the triangular Heisenberg antiferromagnet as a 

742: system of hard-core bosons at half-filling in the background field of 

743: flux $\pi$ through each triangle.  

744: Kalmeyer and Laughlin\cite{KL} proposed to view this in the continuum, 

745: obtaining a boson system at $\nu=1/2$, and their chiral spin liquid

746: wavefunction is precisely the lattice analog of the $\nu=1/2$ state 

747: Eq.~(\ref{Psi_nu_1/2}).  Alternatively, using a slave fermion 

748: approach,\cite{WenWilczekZee, LaughlinZou}

749: $b^\dagger = d_1^\dagger d_2^\dagger$,

750: we divide the boson charge equally between the two fermions,

751: so each sees on average a flux of $\pi/2$ per triangle;

752: the mean field where the $d_1$ and $d_2$ see the same static flux of 

753: $\pi/2$ per triangle gives a filled Landau level for each fermion

754: and reproduces precisely the Laughlin-Kalmeyer chiral spin liquid.

755:   

756: However, we can be more creative about the fluxes seen by the slave 

757: fermions while maintaining the average of $\pi/2$ flux per triangle.  

758: One example is to take different flux patterns for the two species as 

759: follows:  For the $d_1$ fermions, put $0$ flux through all up-pointing 

760: triangles and $\pi$ flux through all down-pointing triangles, 

761: while for the $d_2$ fermions interchange the locations of the $0$ and 

762: $\pi$ fluxes.

763: This state is in fact identical to the so-called 

764: $U1B \tau^1 \tau_-^0 \tau_+^1$ spin liquid found in 

765: Ref.~\onlinecite{ZhouWen}.

766: It is a time reversal invariant gapless algebraic spin liquid (ASL)

767: with Dirac nodes in the spinon spectrum;

768: it has very good energetics for the nearest neighbor triangular 

769: antiferromagnet, starting from the isotropic lattice and all the way to 

770: the limit of weakly coupled chains

771: (e.g., Ref.~\onlinecite{Yunoki06} found a different gauge-equivalent 

772: formulation of this state without realizing its ASL character).

773: 

774: Another example is obtained by taking yet different flux patterns:

775: Select one lattice direction -- chain direction in the anisotropic

776: lattice case.  For the $d_1$ fermions, put $0$ flux through triangles 

777: siding even chains and $\pi$ flux through triangles siding odd chains, 

778: while for the $d_2$ fermions interchange the locations of the $0$ 

779: and $\pi$ fluxes.  This is again a time reversal invariant ASL and is 

780: listed as $U1C \tau_+^0 \tau_-^0 \tau^1$ in 

781: Ref.~\onlinecite{ZhouWen}.

782: Unlike the $U1B$ state, there is no isotropic $U1C$ liquid, but

783: the energetics performance of $U1C$ and $U1B$ spin liquids is almost 

784: indistinguishable for weakly coupled chains and matches that of competing

785: magnetically ordered states; in this context, the time reversal 

786: invariant $U1B$ and $U1C$ liquids are significantly better than the 

787: chiral Laughlin-Kalmeyer variant.

788: 

789: Our construction of the DBL states differs in that we do not require

790: any special filling for the bosons and there is no special Heisenberg

791: spin symmetry.  Neither of these special conditions of the spin model 

792: setting are needed for the construction and subsequent gauge theory 

793: analysis to go through.  

794: Given the growing belief\cite{Rantner, WenPSG, Hermele} 

795: that critical spin liquids do exist, we do not see any reasons why the 

796: situation should be any different for our boson liquids at arbitrary 

797: incommensurate densities.  In our construction, such liquids will 

798: generically have some underlying partially filled bands and therefore 

799: Fermi surfaces of slave fermions.  

800: Of course, whether a particular state is realized in a given model 

801: requires a detailed case by case study, and in Sec.~\ref{sec:Energetics} 

802: we suggest some frustrated boson models that may stabilize the DBL phase.

803: Our primary goal in the next Secs.~\ref{sec:MFT}-\ref{sec:eff_FT}

804: will be to characterize the DBL states without worrying where to find 

805: them.

806: \\

807: 

808: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

809: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

810: \section{Mean Field Theory for the D-wave Bose Liquid}

811: \label{sec:MFT}

812: 

813: \subsection{Gauge Theory formulation}

814: \label{subsec:Gauge_formulation}

815: 

816: We next consider the challenge of constructing a field theory which 

817: can access such a $D_{xy}$-Bose liquid (DBL) state.

818: Due to our inability to implement flux attachment in a tractable 

819: time reversal invariant manner, we follow instead the slave particle 

820: approach.  As above, we decompose the hard core lattice boson as a 

821: product of two fermions, $b^\dagger = d_1^\dagger d_2^\dagger$.

822: Consider then a lattice $U(1)$ gauge theory on the square lattice 

823: in terms of these slave particles:

824: \begin{equation}

825: H_{U(1)} = H_t + H_a ~,

826: \label{HU1}

827: \end{equation}

828: with the fermion hopping Hamiltonian of the form,

829: \begin{widetext}

830: \begin{eqnarray}

831: \label{Ht}

832: H_t &=& -\sum_{\bf r}

833: \left[ 

834: t_\parallel e^{i a_x({\bf r})} d_1^\dagger({\bf r}) d_1({\bf r} + \hat{\bf x}) + 

835: t_\perp e^{i a_y({\bf r})} d_1^\dagger({\bf r}) d_1({\bf r} + \hat{\bf y})+ h.c. \right]\\

836: && -\sum_{\bf r}

837: \left[

838: t_\perp e^{-i a_x({\bf r})} d_2^\dagger({\bf r}) d_2({\bf r} + \hat{\bf x}) + 

839: t_\parallel e^{-i a_y({\bf r})} d_2^\dagger({\bf r}) d_2({\bf r} + \hat{\bf y})+ h.c. \right] ~.

840: \end{eqnarray}

841: \end{widetext}

842: The $d_1$ fermion hopping amplitudes in the $\hat{\bf x}$ and 

843: $\hat{\bf y}$ directions are $t_\parallel$ and $t_\perp$ respectively,

844: while the two amplitudes are interchanged for the $d_2$ fermions.

845: In the following, we take $t_\parallel \ge t_\perp$ for concreteness.

846: Note also that the two fermion species carry opposite gauge charges.

847: The gauge field Hamiltonian is simply

848: \begin{equation}

849: H_a = h \sum_{\bf r} \sum_{\mu = x, y} e^2_\mu({\bf r}) 

850: - K \sum_{\bf r} \cos[ (\nabla \times a)_{\bf r}  ] ~,

851: \label{Hgauge}

852: \end{equation}

853: where the lattice ``magnetic" field is

854: \begin{equation}

855: (\nabla \times a)_{\bf r} = a_x({\bf r}) + a_y({\bf r} + \hat{\bf x}) 

856: - a_x({\bf r}+ \hat{\bf y}) - a_y({\bf r}) ~.

857: \end{equation}

858: The integer-valued ``electric" field $e_\mu({\bf r})$ is canonically 

859: conjugate to the compact gauge field $a_\mu({\bf r})$ on the same link.  

860: The above Hamiltonian is supplemented by a gauge constraint on the 

861: physical states, which must satisfy Gauss' law,

862: \begin{equation}

863: (\nabla \cdot e)_{\bf r} = 

864: d^\dagger_1({\bf r}) d_1({\bf r}) - d^\dagger_2({\bf r}) d_2({\bf r}) ~,

865: \end{equation} with

866: \begin{equation}

867: (\nabla \cdot e)_{\bf r} = e_x({\bf r}) - e_x({\bf r}-\hat{\bf x})

868:  + e_y({\bf r}) - e_y({\bf r}-\hat{\bf y}) ~.

869: \end{equation} 

870: 

871: In the limit $h \gg K, t_\parallel, t_\perp$, the electric field 

872: vanishes, and Gauss' law reduces to $d^\dagger_1 d_1 = d^\dagger_2 d_2$ 

873: which projects back into the physical boson Hilbert space.  

874: In this strong coupling limit, it is possible to perturbatively 

875: eliminate the gauge field to obtain a Hamiltonian for hard core bosons 

876: hopping on the square lattice with additional ring exchange terms.

877: We pursue this in Section~\ref{sec:Energetics} where we compare the 

878: energetics of the DBL wavefunction with other states.  

879: Here we instead focus on the weak coupling limit, $K \gg h$, 

880: which suppresses the magnetic flux.

881: The usual slave particle mean field treatment corresponds to simply 

882: setting $(\nabla \times a)_{\bf r}$ equal to a constant.  

883: The simplest mean field state, and the one which should correspond to 

884: the wavefunctions in the previous section, is with zero flux through all 

885: plaquettes.  The corresponding mean field Hamiltonian describes 

886: non-interacting slave fermions.  

887: Each species has anisotropic near-neighbor hopping amplitudes,

888: but the two are related under the 90 degree rotation,

889: thus producing a boson liquid that respects the symmetries of the square 

890: lattice.

891: 

892: 

893: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

894: \subsection{Mean field results for the DBL}

895: \label{subsec:MF_props}

896: 

897: In order to focus on the effects of the underlying Fermi surfaces

898: without the complications of lattice physics such as Brillouin zone 

899: folding, we first consider fermions in the 2d continuum with anisotropic 

900: effective masses.

901: The resulting elliptical Fermi surfaces are,

902: \begin{eqnarray}

903: \label{wparam}

904: d_1: \quad (w k_x)^2 + (k_y/w)^2 &=& k_F^2 ~,\\

905: d_2: \quad (k_x/w)^2 + (w k_y)^2 &=& k_F^2 ~,

906: \end{eqnarray}

907: where the parameter $w \ge 1$ characterizes the degree of eccentricity

908: (the conventional eccentricity of the ellipses is given by 

909: $\epsilon = \sqrt{1 - 1/w^2}$).

910: In the D-wave Bose liquid, $w$ signifies the mismatch between the

911: two Fermi surfaces, and we will refer to this measure as

912: ``D-eccentricity".

913: We study equal time correlation functions since these can be compared 

914: directly with the properties of the wavefunctions, which is done in 

915: Sec.~\ref{sec:Wavefnc_props};

916: we also consider temporal dependencies within the mean field as a

917: measure of spectral properties.

918: It is useful to have in mind that much of the following analysis of 

919: long-distance properties needs only the knowledge of relevant Fermi

920: surface patches and not of the full surfaces.

921: 

922: Consider first the one-particle off-diagonal density matrix 

923: (or Green's function) for the boson,

924: $G_b({\bf r}, \tau) = 

925:  \la b^\dagger({\bf r}, \tau) b({\bf 0}, \tau) \ra$, with

926: $G_b({\bf 0}, 0) = \bar\rho$ -- the average boson density.

927: It will also be of interest to consider the momentum occupation 

928: probability,

929: \begin{eqnarray}

930: \la n_{\bf k} \ra = \la b^\dagger_{\bf k} b_{\bf k} \ra =

931: \int d{\bf r}\; G_b({\bf r}) e^{-i {\bf k} \cdot {\bf r}} ~. 

932: \end{eqnarray}

933: Within the mean field theory, the natural approximation for the order 

934: parameter correlation is

935: \begin{equation}

936: \label{Gb_MFdef}

937: G_b^{MF}({\bf r}, \tau) = 

938: G^{MF}_{d_1}({\bf r}, \tau) G^{MF}_{d_2}({\bf r}, \tau) / \bar\rho ~,

939: \end{equation}

940: where $G^{MF}_{d_\alpha}$ are the mean field (bare) fermion Green's 

941: functions.  This approximation satisfies 

942: $G_b({\bf 0}) = G_{d_\alpha}({\bf 0}) = \bar\rho$.

943: Here and below, the imaginary time $\tau$ is understood to be zero

944: if not explicitly present.

945: 

946: The fermion Green's functions are readily calculated.

947: Thus, at long distances $r \gg k_F^{-1}$, the main contribution to 

948: $G^{MF}_{d_\alpha}({\bf r})$ comes from the Fermi surface patches where 

949: the group velocity is parallel or antiparallel to the observation 

950: direction $\hat{\bf r} = {\bf r}/|{\bf r}|$.  

951: With inversion symmetry, we can denote the corresponding patch 

952: locations as $\pm {\bf k}_{F_\alpha}$ and the Fermi surface 

953: curvature as $c_\alpha$, and obtain

954: \begin{equation}

955: G^{MF}_{d_\alpha}({\bf r}) \approx \frac{1}{2^{1/2} \pi^{3/2}}

956: \frac{\cos({\bf k}_{F_\alpha} \cdot {\bf r} - 3\pi/4)} 

957:      {c_\alpha^{1/2} |{\bf r}|^{3/2}} ~.

958: \label{Gd_MF}

959: \end{equation}

960: It is important to remember that ${\bf k}_{F_1}, {\bf k}_{F_2}$, 

961: and $c_1, c_2$ depend implicitly on the direction $\hat{\bf r}$ and are 

962: different in general when the $d_1$ and $d_2$ Fermi surfaces do not 

963: coincide.

964: 

965: For the equal-time boson Green's function we thus get

966: \begin{eqnarray}

967: G_b^{MF}({\bf r}) &\sim& 

968: \frac{\cos[({\bf k}_{F_1} - {\bf k}_{F_2}) \cdot {\bf r}]}

969:      {c_1^{1/2} c_2^{1/2} |{\bf r}|^3} \nonumber \\

970: &+& 

971: \frac{\cos[({\bf k}_{F_1} + {\bf k}_{F_2}) \cdot {\bf r} - 3\pi/2]}

972:      {c_1^{1/2} c_2^{1/2} |{\bf r}|^3} ~,

973: \label{Gb_MF}

974: \end{eqnarray}

975: which decays algebraically while oscillating with 

976: $\hat{\bf r}$-direction-dependent wave vectors 

977: ${\bf k}_{F_1} + {\bf k}_{F_2}$ and ${\bf k}_{F_1} - {\bf k}_{F_2}$.

978: Such wave vectors, which are constructed by considering patches on the 

979: two Fermi surfaces that are parallel to each other, will span some 

980: new loci in the momentum space as illustrated in Fig.~\ref{fig:loci2kF}.

981: The above large-distance behavior corresponds to singularities 

982: $n_{\bf k} \sim |\delta k|^{3/2}$ across these lines.

983: In the zero eccentricity limit (i.e., when $w=1$), the two Fermi 

984: surfaces coincide, giving

985: \begin{equation}

986: G^{MF}_b (r)  \sim \frac{1 + \cos[2 k_F r - 3\pi/2]}{r^3}~,

987: \quad\quad \text{(S-type)} ~.

988: \end{equation}

989: Compared to the general case with $w>1$,

990: the ${\bf k}_{F_1} - {\bf k}_{F_2}$ locus shrinks here to zero momentum, 

991: while the singularity in $n_{\bf k}$ is hardened to $|k|$.

992: 

993: As we show in Sec.~\ref{sec:Gauge_fluct}, the power law decay of the 

994: mean field boson Green's function survives in the presence of gauge 

995: fluctuations (but with modified exponents).

996: The algebraic decay is a consequence (and a measurable indication) 

997: of the ``criticality'' of the DBL and its gapless excitations.

998: Of course, by the very construction we expect many gapless excitations

999: because of the underlying Fermi surfaces.  

1000: Some measure of the low-energy spectrum is contained in the time 

1001: dependence of the local boson Green's function,

1002: \begin{equation}

1003: G_b^{MF}({\bf 0}, \tau) = \frac{\nu_0^2}{\tau^2} ~,

1004: \label{Gtau}

1005: \end{equation}

1006: where $\nu_0$ is the density of states at the Fermi energy for each

1007: species.  The corresponding local boson spectral function is

1008: \begin{equation}

1009: A_b^{MF}({\bf r}={\bf 0}, E) = 

1010: \int_{\bf k} A_b^{MF}({\bf k}, E) = \nu_0^2 E ~.

1011: \label{Aloc}

1012: \end{equation}

1013: In the discussion of the DBL phase on the lattice in 

1014: Sec.~\ref{subsec:openFS}, we will encounter a situation where the 

1015: boson Green's function decays exponentially in space because of the 

1016: topology of the Fermi surfaces, while the above spectral signature 

1017: of the low-energy excitations depends only on the density of states

1018: which is a property of the entire Fermi surface and is insensitive 

1019: to the topology otherwise.

1020: 

1021: The most natural instability of the DBL is towards a superfluid state.  

1022: As we discuss in Section~\ref{sec:eff_FT}, in the limit of 

1023: vanishing D-eccentricity ($w=1)$, the resulting S-type Bose Liquid phase 

1024: obtained in mean field theory is, in the presence of gauge fluctuations,

1025: most probably unstable to superfluidity.  

1026: Moreover, our Sec.~\ref{sec:Wavefnc_props} analysis of the 

1027: determinant squared wavefunction appropriate to the S-Bose Liquid

1028: is consistent with off-diagonal long-range order.  

1029: These results strongly suggest that the mean field S-type Bose liquid 

1030: phase probably can not exist as a stable quantum phase.  

1031: However, with non-vanishing D-eccentricity, both the gauge theory 

1032: analysis in Sec.~\ref{sec:eff_FT} and the properties of 

1033: the corresponding $(\det)_x (\det)_y$ wavefunction which we explore

1034: in Sec.~\ref{sec:Wavefnc_props} suggest that the DBL is a stable 

1035: critical quantum phase.

1036: 

1037: It is also instructive to examine several other correlation functions 

1038: within the present mean field treatment.  

1039: Specifically, consider the boson density-density correlation function,

1040: \begin{eqnarray}

1041: D_b({\bf r}) &=&

1042: \la :\! \delta\hat\rho(0) \delta\hat\rho({\bf r}) \!: \ra  \\

1043: &=& N(N-1) \la \delta^d({\bf r}_1) \delta^d({\bf r}_2 - {\bf r}) \ra

1044: - \bar\rho^2 ~,

1045: \label{densitycorr:def}

1046: \end{eqnarray}

1047: where $N$ is the total number of particles.

1048: As defined, $D_b({\bf r})$ approaches zero for large separations, while 

1049: negative values at small distances signify a correlation hole.

1050: The density structure factor can be calculated as

1051: \begin{equation}

1052: D_b({\bf k}) = \int d{\bf r}\; D_b({\bf r}) e^{-i {\bf k} \cdot {\bf r}} 

1053: = \frac{\la |\delta\hat\rho_{\bf k}|^2 \ra}{V} - \bar\rho ~,

1054: \label{Db_k}

1055: \end{equation}

1056: where 

1057: $\delta\hat\rho_{\bf k} =

1058: \sum_j \exp[-i {\bf k} \cdot {\bf r}_j] - N\delta_{\bf k,0}$ and 

1059: $V$ is the system volume.

1060: 

1061: Microscopically, since $b^\dagger = d_1^\dagger d_2^\dagger$, for each 

1062: boson added to the system both a $d_1$ and a $d_2$ fermions are added.

1063: As such, the boson and fermion densities are equal: 

1064: $\hat\rho_b = \hat\rho_{d_1} = \hat\rho_{d_2}$.  

1065: However, in the mean field treatment, the two fermion flavors propagate 

1066: independently with different Fermi surfaces, so the corresponding 

1067: fermion density-density correlation functions, which we denote

1068: as $D_{d_\alpha}^{MF}$, do not coincide.  This ambiguity in the

1069: density correlation is an intrinsic deficiency of the 

1070: mean field theory.  As a crude measure, we approximate the boson density 

1071: correlation function as an average of that of the $d_1$ and $d_2$ 

1072: fermions, 

1073: \begin{equation}

1074: D_b^{MF}({\bf r}) \approx \frac{1}{2} 

1075: [ D^{MF}_{d_1}({\bf r}) + D^{MF}_{d_2}({\bf r}) ] ~.

1076: \label{Boson_MF_density_corr}

1077: \end{equation}

1078: The individual density correlation functions in the mean field theory 

1079: are simply,

1080: \begin{equation}

1081: D_{d_\alpha}^{MF}({\bf r}) = -|G^{MF}_{d_\alpha}({\bf r})|^2 \sim 

1082: - \frac{1 + \cos[2 {\bf k}_{F_\alpha} \cdot {\bf r} - 3\pi/2]}

1083:        {c_\alpha |{\bf r}|^3} ~,

1084: \label{DD_MF}

1085: \end{equation}

1086: where again ${\bf k}_{F_\alpha}$ is the place on the $d_\alpha$ Fermi 

1087: surface where the normal is parallel to the observation direction 

1088: $\hat{\bf r}$.

1089: We recognize the $2 k_F$ oscillation with power law envelope, which in 

1090: the momentum space translates to a singularity in the structure factor 

1091: $\sim |2k_F - k|^{3/2}$ across the $2k_F$ line 

1092: (such $2k_F$ surfaces are illustrated in Fig.~\ref{fig:loci2kF}), 

1093: while there is also a $|k|$ singularity at zero momentum.   

1094: As obtained from Eq.~(\ref{Boson_MF_density_corr}), the mean field 

1095: boson density correlator $D_b^{MF}$ has singularities at both 

1096: $2k_{F_1}$ and $2 k_{F_2}$, and ``knows" about the presence of both 

1097: Fermi surfaces.  Remarkably, this mean field approximation is 

1098: consistent with our analysis of the $(\det)_x (\det)_y$ boson 

1099: wavefunction in Sec.~\ref{sec:Wavefnc_props}, which also

1100: reveals singular behavior at both $2 k_{F_1}$ and $2 k_{F_2}$.

1101: 

1102: 

1103: 

1104: \begin{figure}

1105: \centerline{\includegraphics[width=\columnwidth]{figures/loci2kF.eps}}

1106: \vskip -2mm

1107: \caption{Singular lines in the momentum space for the boson

1108: Green's function (left) and the density correlation (right).

1109: In each panel, the thin lines show the elliptical $d_1$ and $d_2$

1110: Fermi surfaces for the parameter $w^4 = 3$, cf.~Eq.~(\ref{wparam}).

1111: In the left panel, the ${\bf k}_{F_1} \pm {\bf k}_{F_2}$ loci are 

1112: constructed by considering a pair of $d_1$ and $d_2$ fermions with 

1113: parallel ($+$ sign) or antiparallel ($-$ sign) group velocities.

1114: The axes are scaled by an ``average'' $k_F$ that would obtain if the 

1115: Fermi surfaces were circular for the same particle density.

1116: }

1117: \label{fig:loci2kF}

1118: \end{figure}

1119: 

1120: 

1121: Let us now consider the two-boson correlator,

1122: \begin{equation}

1123: G_{2b}({\bf r}_1, {\bf r}_2; {\bf r}_1^\prime, {\bf r}_2^\prime) \equiv 

1124: \la b^\dagger({\bf r}_1) b^\dagger({\bf r}_2) 

1125:     b({\bf r}_1^\prime) b({\bf r}_2^\prime ) \ra ~,

1126: \end{equation}

1127: which injects a pair of bosons at ${\bf r}_1, {\bf r}_2$ and removes a 

1128: pair at ${\bf r}_1^\prime, {\bf r}_2^\prime$.

1129: Consider first the limit where the separation between the two 

1130: injected bosons, ${\bf r} = {\bf r}_1 - {\bf r}_2$, and the two 

1131: removed bosons, ${\bf r}^\prime = {\bf r}_1^\prime - {\bf r}_2^\prime$, 

1132: are both small compared to the mean inter-particle spacing, 

1133: $|{\bf r}|, |{\bf r}^\prime| \ll k_F^{-1}$.

1134: Moreover, take the separation

1135: ${\bf R} = \frac{{\bf r}_1 + {\bf r}_2}{2} - 

1136:            \frac{{\bf r}_1^\prime + {\bf r}_2^\prime}{2}$ 

1137: between the injected and removed pair to be much larger than the 

1138: inter-particle spacing, $|{\bf R}| \gg k_F^{-1}$.  

1139: In this limit the mean field 2-boson correlator can be expressed as,

1140: \begin{equation}

1141: G_{2b}^{MF} \sim

1142: \frac{k_F^8}{ [1 + \frac{W}{2} \sin^2 (2 \phi) ]^{5/2} }

1143: \frac{\Psi_{xy}({\bf r}; \hat{\bf R}) 

1144:       \Psi_{xy}^*({\bf r}^\prime; \hat{\bf R})} 

1145:      {(k_F R)^6} ~,

1146: \end{equation}

1147: with ``pair wavefunction",

1148: \begin{equation}

1149: \Psi_{xy}({\bf r}; \hat{\bf R}) = ({\bf r} \cdot \hat{\bf R})^2 

1150: + W  \sin(2\phi) x y ~.

1151: \end{equation}

1152: Here for concreteness we specialized to the elliptical Fermi surfaces,

1153: Eq.~(\ref{wparam}), and $W \ge 0$ characterizes the D-eccentricity,

1154: \begin{equation}

1155: W = \frac{ w^4 + (1/w)^4 }{2} - 1 ~;

1156: \end{equation}

1157: $\hat{\bf R} = {\bf R}/R = \cos(\phi) \hat{\bf x} 

1158: + \sin(\phi) \hat{\bf y}$, is the unit vector pointing from one pair to 

1159: the other.  

1160: With vanishing D-eccentricity, $W=0$, the pair wavefunction is of an

1161: ``extended-$s$" form, for example $\Psi_{xy} \sim x^2$ for $\phi = 0$.

1162: This corresponds to a quadratic nodal line in the pair wavefunction.

1163: With $W=0$ the directions of the nodal lines for each pair are aligned 

1164: perpendicular to the vector connecting one pair to the other.

1165: On the other hand, in the limit of very large D-eccentricity, 

1166: $W \gg 1$, the pair wavefunction takes the $d_{xy}$ form, 

1167: $\Psi_{xy} \sim xy$, corresponding to two nodal lines along the 

1168: $x$ and $y$ axes.  Upon rotating the internal coordinate, the sign

1169: of the pair wavefunction takes the usual $d_{xy}$ form, precisely as for

1170: the Cooper pair wavefunction in a $d_{xy}$ superconductor.  

1171: Thus, the DBL has quasi--long-range order in the $d_{xy}$ boson 

1172: pair channel, although the power law decay exponent within mean field 

1173: theory is very large.

1174: 

1175: It is also interesting to examine the time decay of a ``local" two-boson 

1176: Green's function,

1177: \begin{equation}

1178: G_{2b}({\bf r}, {\bf r}^\prime; \tau-\tau^\prime) \equiv

1179: \la b^\dagger({\bf r}, \tau) b^\dagger({\bf 0}, \tau)

1180:     b({\bf r}^\prime, \tau^\prime) b({\bf 0}, \tau^\prime) \ra ~,

1181: \end{equation}

1182: with $|{\bf r}|,|{\bf r}^\prime| \ll k_F^{-1}$.  

1183: A pair of bosons is injected at positions ${\bf r}$ and ${\bf 0}$

1184: and is removed at later times at positions ${\bf r}^\prime$ and 

1185: ${\bf 0}$.

1186: Within mean field theory,

1187: \begin{equation}

1188: G^{MF}_{2b}({\bf r}, {\bf r}^\prime; \tau) \sim \frac{1}{\tau^4} 

1189: [ ({\bf r} \cdot {\bf r}^\prime)^2 + 2 W (xy) (x^\prime y^\prime) ] ~.

1190: \end{equation}

1191: For large D-eccentricity, $W \gg 1$, this factorizes into a product

1192: of two ``pair wavefunctions" of the $d_{xy}$ form.

1193: This correlator can be used to extract a local pair tunneling 

1194: density of states,

1195: \begin{equation}

1196: \rho_{xy}(E) = \int

1197: \la \hat{\cal D}_{xy}^\dagger(\tau) \hat{\cal D}_{xy}(0) \ra 

1198: e^{-E |\tau|} d\tau ~,

1199: \end{equation}

1200: where $\hat{\cal D}_{xy}^\dagger$ injects a $d_{xy}$ pair centered 

1201: at the origin,

1202: \begin{equation}

1203: \hat{\cal D}_{xy}^\dagger \equiv \int_{\bf r} \,

1204: xy \,\exp(-|{\bf r}|/\xi) b^\dagger({\bf r}) b^\dagger({\bf 0}) ~,

1205: \end{equation}

1206: with pair ``size" $\xi$.

1207: Within mean field theory one obtains a power law tunneling density of 

1208: states $\rho^{MF}_{xy}(E) \sim W E^3$,

1209: with an amplitude that grows with the D-eccentricity parameter $W$.  

1210: The tunneling density of states for an $s$-wave pair also vanishes as 

1211: $E^3$ but the amplitude is independent of $W$.  

1212: This pair tunneling density of states is perhaps the best diagnostic for 

1213: measuring the degree of local $d$-wave two-boson correlations in the DBL.

1214: 

1215: 

1216: Finally, we consider a box correlator for bosons which is defined 

1217: in the Introduction, Eq.~(\ref{Bb_def}).

1218: Within mean field this factorizes as,

1219: ${\cal B}^{MF}_b(x) = {\cal B}_{d_1}(x) {\cal B}_{d_2}(x)$, and from 

1220: Wick's theorem and 90-degree rotational invariance,

1221: \begin{equation}

1222: {\cal B}_{d_1}(x) = G_{d_1}^2(0,x) - G_{d_1}^2(x,0) 

1223: = -{\cal B}_{d_2}(x) ~.

1224: \end{equation}

1225: Thus,

1226: \begin{equation}

1227: \label{Bb_MF}

1228: {\cal B}^{MF}_b(x) = -[ G_{d_1}^2(0,x) - G_{d_1}^2(x,0) ]^2 \le 0 ~.

1229: \end{equation}

1230: Notice that in the case with zero eccentricity the mean field box 

1231: correlator vanishes, but since the $(\det)^2$ wavefunction is 

1232: non-negative, one expects ${\cal B}_b$ to be positive upon inclusion of 

1233: gauge fluctuations.

1234: With non-zero D-eccentricity, the mean field box correlator is 

1235: negative for all spatial separations, which reflects the underlying 

1236: $d_{xy}$ nodal structure, and decays as $x^{-6}$.

1237: It is possible that the inclusion of gauge fluctuations will again

1238: modify the mean field result in the case with closed Fermi surfaces

1239: because of the pairing tendencies of the $d_1$ and $d_2$ fermions.

1240: In the case with open Fermi surfaces to be discussed next,

1241: we conjecture that the exact box correlator will be negative 

1242: at large spatial separations while decaying to zero as the box size is 

1243: taken to infinity.

1244: This conjecture appears to be consistent with the box correlator 

1245: extracted from the $(\det)_x (\det)_y$ wavefunction with open Fermi

1246: surfaces in Sec.~\ref{subsec:lattice_wavefnc}.

1247: 

1248: 

1249: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1250: \subsection{Case with open Fermi surfaces}

1251: \label{subsec:openFS}

1252: 

1253: The preceding analysis holds for arbitrary Fermi surfaces,

1254: in particular, for the lattice bands obtained from Eq.~(\ref{Ht}).

1255: We only need to remember that the fermion Green's function 

1256: Eq.~(\ref{Gd_MF}) is determined by the Fermi surface patches with the 

1257: group velocities that are parallel or antiparallel to $\hat{\bf r}$. 

1258: Once $G_{d_1}$ and $G_{d_2}$ are known, the other correlation functions

1259: follow.

1260: 

1261: An interesting situation occurs on the lattice when the ratio

1262: $t_\parallel / t_\perp$ is such that the $d_1$ and $d_2$ Fermi surfaces 

1263: are open, which is illustrated in the right panel of 

1264: Fig.~\ref{fig:introFS}.

1265: In this case, for an observation direction close to, say, the $y$-axis, 

1266: there are no $d_1$ Fermi surface patches with normals in this direction, 

1267: so the $d_1$ Green's function has an exponential decay in this direction 

1268: instead of the power law Eq.~(\ref{Gd_MF}).  

1269: Since the real-space boson Green's function is the product of the two 

1270: fermion Green's functions, Eq.~(\ref{Gb_MFdef}), we conclude that it 

1271: decays exponentially in the directions near the $x$- and $y$-axes.  

1272: There may still still be directions near $\hat{\bf x} \pm \hat{\bf y}$ 

1273: in which both $G_{d_1}$ and $G_{d_2}$ show power law behavior, 

1274: and so will the boson Green's function.

1275: 

1276: Eventually, for large enough $t_\parallel / t_\perp$, the two Fermi

1277: surfaces will have no parallel patches, and the boson Green's

1278: function will decay exponentially in all directions.

1279: We will call this phase DLBL for D-wave correlated Local Boson Liquid.

1280: Note, however, that the system is still gapless and critical,

1281: as can be measured, e.g., from the local spectral function 

1282: Eq.~(\ref{Aloc}).

1283: Also, the boson density correlations still have the power law envelope, 

1284: Eq.~(\ref{DD_MF}), if one fermion field can propagate in the 

1285: observation direction.

1286: Furthermore, the boson box correlator, Eq.~(\ref{Bb_MF}), exhibits 

1287: power law behavior even though the single and pair boson Green's 

1288: functions are exponentially decaying.

1289: From the boson field perspective, the system is local and bosons

1290: have hard time to propagate; 

1291: nevertheless, the system has power law correlations in other properties 

1292: and, in particular, charges can propagate.

1293: 

1294: Based on an analysis of gauge fluctuations in Sec.~\ref{sec:Gauge_fluct},

1295: we conjecture that the $d_1$ and $d_2$ systems effectively decouple at 

1296: low energies in the DLBL, and at large distances the exact box correlator

1297: is negative and decays to zero as a non-oscillatory power law with 

1298: an integer exponent which is independent of non-universal lattice scale 

1299: physics, ${\cal B}_b(x) \sim - x^{-8}$.  This is the same as in the 

1300: mean field theory except with a larger power.

1301: 

1302: It is also worth mentioning the limiting case $t_\perp=0$ that gives 

1303: completely flat Fermi surfaces, which we will call ``extremal DLBL.''

1304: The $d_1$ fermions can move only along the $x$-axis,

1305: \begin{equation}

1306: G_{d_1}(x,y) = \delta_{y,0} \frac{\sin(k_F x)}{\pi x}~, \quad 

1307: \text{(extremal DLBL)} ~,

1308: \end{equation}

1309: while the $d_2$ fermions can move only along the $y$-axis.

1310: As a result, the boson field cannot propagate at all, not even one 

1311: lattice spacing.  However, this special system still has power law 

1312: density-density correlations, e.g.,

1313: \begin{equation}

1314: D^{MF}_{d_1}(x,y) \sim -\delta_{y,0} \frac{\sin^2(k_F x)}{x^2}, \quad 

1315: \text{(extremal DLBL)} ~,

1316: \label{Dd_MF_XTRM}

1317: \end{equation}

1318: as well as power law box correlation,

1319: \begin{equation}

1320: {\cal B}^{MF}_b(x) \sim - \frac{\sin^4(k_F x)}{x^4}, \quad

1321: \text{(extremal DLBL)} ~.

1322: \label{Bb_MF_XTRM}

1323: \end{equation}

1324: From the numerical study of the extremal DLBL wavefunction, 

1325: Sec.~\ref{subsubsec:xtrmDLBL}, we conjecture that these mean field 

1326: power laws also hold upon the Gutzwiller projection.

1327: In the gauge theory context, we would say that the fermions

1328: remain unaffected by the gauge field fluctuations.  

1329: This extremal DLBL wavefunction is of interest because of its 

1330: similarity to the so-called Excitonic Bose Liquid ground state of a pure 

1331: ring exchange model -- we discuss this in Sec.~\ref{sec:Energetics}.

1332: 

1333: 

1334: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1335: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% 

1336: \section{Properties of the wavefunctions}

1337: \label{sec:Wavefnc_props}

1338: 

1339: In this section we study the $\Psi = (\det)_x (\det)_y$ wavefunctions 

1340: directly and measure numerically their properties such as the boson 

1341: Green's function and the density correlation defined in 

1342: Sec.~\ref{subsec:MF_props}.

1343: A detailed comparison is made with the mean field, which provides an 

1344: initial guide.  This is followed by interpretations of the observed

1345: deviations from the mean field using simple ``Amperean interaction'' 

1346: rules of thumb;

1347: the actual calculations behind these rules within the gauge 

1348: fluctuations theory are given in Sec.~\ref{sec:Gauge_fluct}.

1349: 

1350: We first consider wavefunctions in the continuum so as to avoid 

1351: lattice effects and focus on the consequences of the underlying Fermi 

1352: surfaces.

1353: It appears that the S-type wavefunction, $\Psi = (\det)^2$,

1354: has off-diagonal long-range order, while a generic DBL wavefunction

1355: with non-zero D-eccentricity has only power law boson correlations.

1356: We then consider wavefunctions on the lattice where we can access 

1357: the DLBL phase with open Fermi surfaces described in 

1358: Sec.~\ref{subsec:openFS}; this features boson Green's function that 

1359: decays exponentially in real space.

1360: Finally, we discuss in some detail the extremal wavefunction 

1361: that is obtained when the Fermi surfaces are completely flat.

1362: 

1363: The calculations with the wavefunctions are performed numerically 

1364: using standard determinantal VMC (variational Monte Carlo) 

1365: techniques.\cite{vmc}  The system used in all calculations is a 

1366: square box with periodic boundary conditions.

1367: In each case, the particle number is chosen so as to fill complete 

1368: momentum shells under the Fermi surfaces.

1369: To facilitate the comparison, the presented mean field is calculated 

1370: for the same finite systems.

1371: 

1372: 

1373: 

1374: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1375: \subsection{Wavefunctions in the continuum}

1376: 

1377: Before proceeding with the numerics, the DBL structure factors 

1378: can be found analytically in some ranges in the momentum space:

1379: Thus, by expanding the boson wavefunction in terms of the orbitals that

1380: form each determinant, it is easy to see that

1381: $n_{\bf k}$ vanishes outside the ${\bf k}_{F1} + {\bf k}_{F2}$ 

1382: surface of Fig.~\ref{fig:loci2kF}, while $D({\bf k})$ vanishes

1383: outside the $2{\bf k}_{F1} + 2{\bf k}_{F2}$ surface.

1384: 

1385: 

1386: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1387: \subsubsection{S-type state}

1388: We begin by considering the limit with zero D-eccentricity, 

1389: in which case the boson wavefunction is positive everywhere 

1390: except for the nodes.

1391: Figure~\ref{fig:S-type} shows boson Green's function measured for 

1392: two systems with $N=161$ and $325$ particles.

1393: It appears that at large separations $G_b(r)$ approaches a finite 

1394: positive value which is roughly the same for both system sizes.

1395: To be more quantitative, the ${\bf k}=0$ mode contains

1396: $n_{{\bf k}=0}/N = 0.10$ and $0.083$ fractions of bosons 

1397: in the two systems.  The present data extracted from the continuum 

1398: wavefunction cannot rule out the possibility that the Green's function 

1399: vanishes in the large distance limit.

1400: But at the very least, the result of the projection is clearly dramatic 

1401: in this case with matched Fermi surfaces, since the fall-off of the 

1402: boson correlation is very slow (if any).

1403: We also remark that we unambiguously find off-diagonal long-range order

1404: when this wavefunction is studied on a lattice at fixed boson density

1405: 

1406: The extremely strong enhancement of the boson correlation over the

1407: mean field prediction seen in Fig.~\ref{fig:S-type}

1408: can be understood qualitatively as a result of the pairing of the 

1409: $d_1$ and $d_2$ fermions mediated by the gauge field.

1410: Indeed, as we discuss in Sec.~\ref{sec:Gauge_fluct},

1411: the constituents of such a (zero momentum) ``Cooper pair'' with 

1412: oppositely directed group velocities and opposite gauge charges 

1413: produce parallel gauge currents and therefore experience Amperean 

1414: attraction mediated by the gauge field.  

1415: This ``Amperean attraction'' rule of thumb for the enhancements in

1416: correlations appears to be taken to the extreme in the $(\det)^2$ 

1417: wavefunction, where we can crudely picture the fermions paired 

1418: back into $b$ and condensed, giving rise to long-range order in $G_b$.

1419: We also point out that this wavefunction has rather unusual 

1420: density-density correlations which are singular around the 

1421: $2 k_F$ circle.

1422: Since the presence of the off-diagonal order makes this state 

1423: less interesting to us, we do not consider such details any further.

1424: The observation of the order suggests that the S-type state is unstable; 

1425: the ultimate phase in this case is likely a conventional superfluid, 

1426: and a good superfluid wavefunction needs to be constructed differently, 

1427: e.g., using Jastrow approach that builds in proper density correlations.

1428: 

1429: 

1430: \begin{figure}

1431: \centerline{\includegraphics[width=\columnwidth]{figures/BdagB_CY100_PWR200_NBOS325_10dir.eps}}

1432: \vskip -2mm

1433: \caption{Boson correlation in the $(\det)^2$ wavefunction for two 

1434: systems with $N=161$ and $325$ particles.

1435: The distance is measured in units of $a = \rho^{-1/2}$, 

1436: and the plots are cut when the separation exceeds half of the box size.

1437: Since $G_b(r)$ maintains a sizable limiting value that depends only 

1438: weakly on $N$, we suggest that there is off-diagonal long-range order

1439: in this wavefunction.  

1440: The mean field result is also plotted, and the vertical scale is chosen 

1441: to accommodate its fast decay, $G_b^{MF} \sim 1/r^3$.

1442: }

1443: \label{fig:S-type}

1444: \end{figure}

1445: 

1446: 

1447: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1448: \subsubsection{DBL state}

1449: 

1450: We now consider in detail a representative case with non-zero

1451: D-eccentricity -- specifically, with $w^4 = 3$ in Eq.~(\ref{wparam}).

1452: The corresponding $d_1$ and $d_2$ Fermi surfaces are shown to scale in 

1453: Fig.~\ref{fig:loci2kF} together with the singular lines for the 

1454: order parameter and density correlations identified in the mean field,

1455: Sec.~\ref{sec:MFT}.

1456: 

1457: Figure~\ref{fig:allq} gives an overall view of $n_{\bf k}$ and 

1458: $D({\bf k})$ in the two-dimensional momentum space, and also shows a 

1459: one-dimensional cut in the $(1,0)$ k-space direction together

1460: with the mean field predictions.

1461: 

1462: \begin{figure}

1463: \centerline{\includegraphics[width=\columnwidth]{figures/allq_cy33_PWR200_NBOS325.eps}}

1464: \vskip -10mm

1465: \centerline{\includegraphics[width=\columnwidth]{figures/q10dir_cy33_PWR200_NBOS325.eps}}

1466: \vskip -2mm

1467: \caption{

1468: Top:  Mode occupation $n_{\bf k}$ and density structure factor

1469: $D({\bf k})$ measured in the continuum DBL wavefunction with $w^4=3$

1470: (note that our $D({\bf k})$ defined in Eq.~\ref{Db_k} vanishes 

1471: for large $k$).

1472: The boson number is $N=325$; the underlying Fermi surfaces and all 

1473: singular lines are shown to scale in Fig.~\ref{fig:loci2kF}.

1474: Bottom:  Cut through the data in the (1,0) direction in the k-space 

1475: and comparison with the mean field results.

1476: In the left panel, the arrows point the appropriate $k_{F1} \pm k_{F2}$ 

1477: momenta (see Fig.~\ref{fig:loci2kF}), while in the right panel the 

1478: arrows point the $2k_{F1}$ and $2k_{F2}$ momenta.

1479: }

1480: \label{fig:allq}

1481: \end{figure}

1482: 

1483: 

1484: Consider first the mode occupation function $n_{\bf k}$.

1485: Upon projection, this develops a squarish top with somewhat sharper 

1486: edges as compared with the mean field.  The latter is not shown in the 

1487: full k-space but looks more smooth, while a (1,0) cut can be seen in the 

1488: bottom panel of Fig.~\ref{fig:allq}.

1489: Coming from large momenta, significant deviations from the mean field 

1490: set in near the ${\bf k}_{F1} - {\bf k}_{F2}$ line of 

1491: Fig.~\ref{fig:loci2kF}

1492: (the corresponding location in the 1D cut is indicated with an arrow).

1493: This surface is singular in the mean field, but the singularities are 

1494: weak and are almost not visible, while they become more pronounced upon 

1495: the projection.  On the other hand, we observe no such enhancement near

1496: the ${\bf k}_{F1} + {\bf k}_{F2}$ line.

1497: 

1498: This difference between the ${\bf k}_{F1} - {\bf k}_{F2}$ and

1499: ${\bf k}_{F1} + {\bf k}_{F2}$ is also visible when examining the boson 

1500: Green's function in real space as shown in the left panels of 

1501: figures~\ref{fig:real-space-10}~and~\ref{fig:real-space-11}.

1502: The mean field $G_b^{MF}({\bf r})$, Eq.~(\ref{Gb_MF}), has both

1503: ${\bf k}_{F1} - {\bf k}_{F2}$ and ${\bf k}_{F1} + {\bf k}_{F2}$ 

1504: components equally present, while after the Gutzwiller projection one

1505: finds that the ${\bf k}_{F1} - {\bf k}_{F2}$ component dominates.

1506: Indeed, for the boson Green's function measured along the $x$-axis,

1507: Fig.~\ref{fig:real-space-10}, the relevant Fermi surface patches have 

1508: normals in the $(x, 0)$ direction and are easily located 

1509: (the relevant momentum space cut of $n_{\bf k}$ is also shown in the 

1510: left panel of Fig.~\ref{fig:allq}).

1511: By comparing with the mean field Eq.~(\ref{Gb_MF}), one finds that after 

1512: the projection the amplitude of the oscillation with the smaller 

1513: wave vector $|k_{F1} - k_{F2}|$ wins over that with the larger wave 

1514: vector $k_{F1} + k_{F2}$. 

1515: Slightly more care is needed when interpreting $G_b({\bf r})$ in the 

1516: $x=y$ diagonal direction, Fig.~\ref{fig:real-space-11}.

1517: In this case, the appropriate locations on the 

1518: ${\bf k}_{F1} - {\bf k}_{F2}$ surface are at the ``cusp'' points with 

1519: $k_x = -k_y$ in Fig.~\ref{fig:loci2kF}, since this is where the normal

1520: to the singular surface is parallel to the observation direction 

1521: $\hat{\bf r}$.  This component does not oscillate when moving along the 

1522: $x=y$ diagonal in real space, and when it is enhanced, the oscillations 

1523: due to the other ${\bf k}_{F1} + {\bf k}_{F2}$ component become less 

1524: visible, which is what what we see in Fig.~\ref{fig:real-space-11}.

1525: 

1526: 

1527: \begin{figure}

1528: \centerline{\includegraphics[width=\columnwidth]{figures/10dir_CY33_PWR200_NBOS325.eps}}

1529: \vskip -2mm

1530: \caption{Boson Green's function $G_b(r)$ and density correlation 

1531: $D_b(r)$ measured along the $r=(x,0)$ direction for the DBL wavefunction 

1532: with $N=325$ bosons and D-eccentricity characterized by $w^4=3$

1533: (this is the same system as in Fig.~\ref{fig:allq}).   

1534: The distance is measured in units of $a = \rho^{-1/2}$,

1535: and the plots are cut roughly where the signal in $G_b(r)$ approaches

1536: the noise level.  Note the logarithmic scales used, so it is the 

1537: absolute values that are plotted, while the positive/negative signs

1538: of the data are indicated with filled/open symbols respectively.

1539: In the case of the mean field data, the signs are indicated using

1540: solid/broken lines; $D^{MF}$ is always negative, see Eq.~(\ref{DD_MF}).

1541: }

1542: \label{fig:real-space-10}

1543: \end{figure}

1544: 

1545: 

1546: \begin{figure}

1547: \centerline{\includegraphics[width=\columnwidth]{figures/11dir_CY33_PWR200_NBOS325.eps}}

1548: \vskip -2mm

1549: \caption{The same as in Fig.~\ref{fig:real-space-10} but in the

1550: $x = y$ diagonal direction in real space.

1551: }

1552: \label{fig:real-space-11}

1553: \end{figure}

1554: 

1555: 

1556: Unfortunately, from this data we cannot attest whether the enhanced

1557: singularities are characterized by new exponents or whether we see

1558: only an amplitude effect.

1559: The locations where the enhancements occur are consistent with the 

1560: Amperean rules of thumb, Sec.~\ref{sec:Gauge_fluct}.

1561: Indeed, the $d_1$ and $d_2$ constituents of the boson operator 

1562: $b^\dagger = d_1^\dagger d_2^\dagger$ move in the opposite directions 

1563: when contributing to the correlation at ${\bf k}_{F1} - {\bf k}_{F2}$ 

1564: and therefore experience Amperean attraction and enhancement 

1565: (remembering that $d_1$ and $d_2$ carry opposite gauge charges), 

1566: while they move in the same direction when contributing at 

1567: ${\bf k}_{F1} + {\bf k}_{F2}$ and therefore experience Amperean 

1568: suppression.

1569: These pictures are made more precise in Sec.~\ref{sec:Gauge_fluct},

1570: where we find that within the gauge fluctuations theory, 

1571: the enhanced correlations are characterized by new exponents that 

1572: depend on the observation direction because of the varying degree 

1573: of the Fermi surface curvature matching in the $d_1$-$d_2$ pairing 

1574: channel.  The strongest such enhancement is expected along the 

1575: diagonals $x = \pm y$, which is roughly what we find in the projected 

1576: wavefunctions

1577: (we repeat again that we cannot make any statement about the exponents 

1578: from the data other than that we see increased numerical correlations 

1579: compared with the mean field).

1580: 

1581: 

1582: Consider now the density correlation $D_b(r)$ and the corresponding 

1583: structure factor $D_b({\bf k})$. 

1584: In Fig.~\ref{fig:allq}, we can see the singular $2{\bf k}_{F1}$ 

1585: and $2{\bf k}_{F2}$ lines, cf.~Fig.~\ref{fig:loci2kF};

1586: the enhancements are peaked where the two curves cross.

1587: The singular $2k_F$ points are also visible in the (1,0) k-space cut,

1588: bottom right panel of Fig.~\ref{fig:allq}.

1589: On the other hand, the $|k|$ singularity near zero momentum is 

1590: not enhanced.

1591: Examination of the density correlation in real space, right panels of

1592: figures~\ref{fig:real-space-10}~and~\ref{fig:real-space-11},

1593: shows an overall increase and dominance of the oscillatory 

1594: components over the zero-momentum component as compared with the 

1595: mean field Eq.~(\ref{DD_MF}).

1596: Again, we cannot tell whether there is a change in the exponents or 

1597: just an amplitude effect.  The $2k_F$ enhancements of the density 

1598: correlation agree qualitatively with the gauge theory expectations.  

1599: Indeed, consider the density operator $d_1^\dagger d_1$.

1600: The particle and hole constituents (which are oppositely gauge charged) 

1601: have antiparallel group velocities when contributing to the $2 k_F$ 

1602: component and therefore experience Amperean attraction, while they 

1603: move in the same direction when contributing at zero momentum and 

1604: therefore repel each other.  

1605: Again, these rules of thumb are made more precise in 

1606: Sec.~\ref{sec:Gauge_fluct}, where such enhancements in the particle-hole 

1607: channel are characterized by new direction-dependent exponents

1608: (the density correlations are found to decay more slowly when less 

1609: curved patches are involved).

1610: Note that the projection imposes 

1611: $\hat{\rho}_b = \hat{\rho}_{d_1} = \hat{\rho}_{d_2}$, and

1612: we expect $D_b$ to acquire both the $2k_{F1}$ and $2k_{F2}$ signatures 

1613: (in the gauge theory, $\rho_{d_1}$ and $\rho_{d_2}$ imprint on each other

1614: via non-singular high-energy connections that are not manifest in our 

1615: low-energy effective description of Sec.~\ref{sec:Gauge_fluct}).

1616: 

1617: 

1618: To summarize, the DBL wavefunction clearly knows about the underlying 

1619: Fermi surfaces; it also contains germs of the gauge fluctuations theory, 

1620: since the enhancements/suppressions of the various ``$2 k_F$'' lines 

1621: appear to agree with the Amperean rules.

1622: However, there are no reasons to believe that the wavefunction 

1623: and the gauge theory will have the same long-distance properties.

1624: In particular, our measurements can not tell whether the long-distance

1625: power laws are changed upon the projection compared with the mean field.

1626: Still, it is gratifying to see that the Amperean rules work for the

1627: DBL wavefunction.

1628: 

1629: 

1630: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1631: \subsection{Wavefunctions on the lattice}

1632: \label{subsec:lattice_wavefnc}

1633: 

1634: We now describe our results for various states on the lattice.

1635: As we have already mentioned, S-type wavefunction, $\Psi = (\det)^2$,

1636: with finite boson density per site, have off-diagonal long-range 

1637: order, which we confirm unambiguously using finite-size scaling.

1638: On the other hand, wavefunctions with non-zero D-eccentricity

1639: do not show such order.  

1640: While we have not performed detailed studies, we expect that the case 

1641: with closed Fermi surfaces is similar to the already described continuum 

1642: DBL wavefunction, with a slight complication that one needs to do

1643: proper Brillouin zone folding when considering singular surfaces in the 

1644: momentum space.

1645: 

1646: \subsubsection{Open Fermi surfaces -- DLBL state}

1647: Of particular interest is the case with open Fermi surfaces,

1648: which can only be realized on the lattice.

1649: Specifically, we studied a $24 \times 24$ system with $N=216$

1650: bosons ($\rho = 0.375$ per site) and the fermion hopping parameters 

1651: $t_\parallel = 1.35$, $t_\perp = 0.65$.  

1652: In this case, the $d_1$ and $d_2$ Fermi surfaces have no parallel 

1653: patches; we therefore expect that the boson Green's function decays 

1654: exponentially.

1655: When we measure $G_b(r)$, we find that it drops below our noise level 

1656: already at 3 lattice spacings, so the corresponding plots are not 

1657: particularly informative and are not shown here.  

1658: Where we can measure reliably, the values after the projection are 

1659: of the same order as the mean field values in the same system.

1660: Since the latter decay exponentially at large distances,

1661: we conjecture the same behavior in the DLBL wavefunction.

1662: 

1663: On the other hand, the boson density correlation along the $x$

1664: and $y$ axes shows oscillations with a clear power law envelope, 

1665: while the correlations are much smaller in the diagonal $x = \pm y$ 

1666: directions; this behavior is again qualitatively consistent with the

1667: mean field, since neither $d_1$ nor $d_2$ Fermi surfaces have normals 

1668: in the diagonal directions, while one or the other has normals 

1669: along the $x$ or $y$ axis.

1670: 

1671: 

1672: 

1673: \subsubsection{Flat Fermi surfaces -- extremal DLBL state}

1674: \label{subsubsec:xtrmDLBL}

1675: Finally, let us discuss the extremal case when the Fermi surfaces

1676: are completely flat.  The $d_1$ fermions can move only along the

1677: $x$-axis and $d_2$ fermions only along the $y$-axis.  

1678: Using fermion orbitals that are localized on individual rows for $d_1$

1679: (or columns for $d_2$), one can see that the boson wavefunction is 

1680: nonzero only when the number of particles on each row (or on each column)

1681: is the same.

1682: Thus, the bosons can not propagate and their inter-site 

1683: correlation is identically zero.

1684: As we discuss in Sec.~\ref{sec:Energetics}, pure ring Hamiltonian

1685: conserves boson number on each row/column, and the extremal DLBL

1686: wavefunction may be useful in this context.

1687: 

1688: We can still use the box correlator, Eq.~(\ref{Bb_def}), to

1689: characterize the state and see some ``gaplessness'' in the system.  

1690: The measurement is shown in Fig.~\ref{fig:corrBOX}, where we also plot 

1691: renormalized mean field result.  

1692: The latter is obtained by dividing Eq.~(\ref{Bb_MF}) by 

1693: $\rho^2 (1-\rho)^2$, and a crude justification for such 

1694: procedure\cite{Zhang} is as follows:  

1695: Each $d_1$ or $d_2$ mean field box calculation contains 

1696: implicitly a weight of order $\rho^2 (1-\rho)^2$, since for the ring

1697: operator to be nonzero, two specific sites need to be occupied

1698: and two need to be empty.  However, after the projection it is enough 

1699: to require that only the $d_1$ configuration is ``correct'' since the 

1700: $d_2$ fermions are tied to $d_1$.

1701: From Fig.~\ref{fig:corrBOX}, we see that the box correlator is negative 

1702: and tracks closely the renormalized mean field values.

1703: 

1704: 

1705: \begin{figure}

1706: \centerline{\includegraphics[width=\columnwidth]{figures/corrBOX_EXTRMDxDy_NVOL576_NBOS216.eps}}

1707: \vskip -2mm

1708: \caption{Box correlator ${\cal B}_b(x)$ in the extremal DLBL wavefunction

1709: with flat Fermi surfaces on a $24 \times 24$ lattice with $N=216$ bosons

1710: (the lines are guide to the eye).

1711: Note that the box correlator is negative in the mean field

1712: and is found to be negative in the DLBL state; 

1713: we plot $-{\cal B}_b(x)$ to be able to use the logarithmic scale.

1714: The mean field Eq.~(\ref{Bb_MF}) is renormalized as explained in the

1715: text using a numerical factor that removes double weighting of 

1716: configurations.  

1717: The measured values below $10^{-5}$ are at the noise threshold.

1718: }

1719: \label{fig:corrBOX}

1720: \end{figure}

1721: 

1722: 

1723: \begin{figure}

1724: \centerline{\includegraphics[width=\columnwidth]{figures/DENScorr_EXTRMDxDy_NVOL576_NBOS216.eps}}

1725: \vskip -2mm

1726: \caption{Boson density correlation for the same extremal DLBL system

1727: as in Fig.~\ref{fig:corrBOX}.  

1728: Left panel:  Real space dependence along the $x$-axis, $D_b(x,0)$, 

1729: together with the mean field expectation.

1730: Right panel:  Density structure factor in the momentum space.  

1731: }

1732: \label{fig:DENScorr}

1733: \end{figure}

1734: 

1735: We can also characterize the extremal DLBL state by measuring the 

1736: density correlations and comparing with the mean field prediction, 

1737: Eq.~(\ref{Dd_MF_XTRM}).  

1738: The results are in Fig.~\ref{fig:DENScorr}.

1739: The left panel shows $D_b(x,0)$ as a function of the distance

1740: along the $x$-axis; it has stronger oscillatory component than the

1741: mean field and swings back and forth across the zero line

1742: while the mean field only touches it, but the overall magnitudes

1743: are comparable and decay as $1/x^2$.

1744: We also find that the density correlation in the $x = \pm y$ diagonal 

1745: directions decays exponentially (not shown);

1746: the mean field predicts zero correlation unless strictly along the axes, 

1747: and we expect that after the projection this corresponds to exponential 

1748: decay.

1749: 

1750: In the right panel of Fig.~\ref{fig:DENScorr}, we show the density

1751: structure factor $D({\bf k})$.  Several features are clearly visible:

1752: rising ``towers'' draw attention to the lines $(\pm 2 k_F, k_y)$ and 

1753: $(k_x, \pm 2 k_F)$;

1754: one can also see a ``cross'' formed by the lines $(0, k_y)$ and 

1755: $(k_x, 0)$ that run along the axes.

1756: As far as we can say, the character of the singularities across these

1757: lines remains the same as in the mean field; there is an amplitude 

1758: enhancement of the $2k_F$, but no qualitative difference otherwise. 

1759: In particular, near the line $(0, k_y)$, we observe

1760: $D_{b, {\rm sing}}(k_x \to 0, k_y) = A |k_x|$ with $A$ which is 

1761: independent of $k_y$ as long as $|k_y| \gg |k_x|$.

1762: 

1763: We thus conjecture that the extremal DLBL wavefunction is adequately 

1764: described by the free fermion mean field. 

1765: Some of our findings, e.g., the cross singularity that has a long 

1766: wavelength character, can be understood semi-analytically, 

1767: since the absolute value of the wavefunction has a Jastrow form with a 

1768: peculiar pseudo-potential,

1769: $u(x, y) \sim -\ln|\sin(\pi x /L)| \delta_{y,0} 

1770:               -\ln|\sin(\pi y /L)| \delta_{x,0}$,

1771: it seems plausible that one can calculate other properties of this 

1772: wavefunction analytically.

1773: It is interesting to note that the above suggests a free fermion

1774: description of this boson state, perhaps with some constraints that become

1775: irrelevant in an infinite system.  

1776: We also note that the extremal DLBL appears to be a relative of the 

1777: so-called Excitonic Bose Liquid (EBL) phase predicted in a pure ring 

1778: model on the square lattice\cite{Paramekanti} that we discuss in 

1779: Sec.~\ref{sec:Energetics};

1780: the above therefore suggests that there may be some such description

1781: of the EBL in terms of fermions carrying fractional $1/2$ charge.

1782: 

1783: 

1784: 

1785: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1786: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1787: \section{Gauge Fluctuations}

1788: \label{sec:Gauge_fluct}

1789: 

1790: We now study the gauge theory using analytic techniques,

1791: focusing on the large $K$ limit where it is reasonable to

1792: ignore the presence of magnetic monopoles in space-time and to 

1793: expand the cosine term in Eq.~(\ref{Hgauge}) treating the 

1794: gauge field as non-compact.  

1795: Moreover, in the $K \to \infty$ limit, the fermions become free and 

1796: one can put them into a Fermi sea state.

1797: A Fermi surface of fermions minimally coupled to a non-compact $U(1)$ 

1798: gauge field has a considerable history.\cite{Holstein, Reizer, PALee, IoffeKotliar, LeeNagaosa, Polchinski, Altshuler, Nayak, YBKim, LeeNagaosaWen, Senthil, Galitski_gauge}

1799: The $2+1$D such system has been studied most notably as a theory for 

1800: the spin sector in the uniform RVB phase in the slave boson approach

1801: to the high-$T_c$ superconductors.

1802: It has been argued that such fermion systems have a stable phase that 

1803: in some crude aspects is similar to a Fermi liquid -- for example, 

1804: it has a finite long-wavelength compressibility and spin susceptibility.

1805: However, the system is strikingly different in other aspects and is 

1806: described by a new non-Fermi liquid fixed point.

1807: A scaling description of this fixed point was developed in 

1808: Refs.~\onlinecite{Polchinski, Altshuler}.

1809: 

1810: While we largely follow the earlier work, we find that it is 

1811: convenient to consider a slight reformulation in which the only 

1812: uncontrolled approximation is to assume that the gauge field dynamics 

1813: can be described correctly by the RPA approximation, retaining only 

1814: terms quadratic in the gauge field with a singular quadratic kernel.

1815: A virtue of this approach is that one thereby obtains a theory which has 

1816: an $N$ flavor extension which is soluble at $N=\infty$.  Moreover, 

1817: a controlled and systematic perturbation expansion in powers of $1/N$ 

1818: can be implemented, which allows one to compute  physical properties in 

1819: terms of non-universal ``bare" parameters such as the shapes of the 

1820: Fermi surfaces.  We employ this approach to calculate the leading

1821: $1/N$ behavior for both the boson correlator, $G_b({\bf r})$, 

1822: and the density-density correlator, $D_b({\bf r})$, in the D-wave Bose 

1823: Liquid.

1824: 

1825: In the next Sec.~\ref{sec:eff_FT}, we will consider an effective field 

1826: theory approach, which allows us to check for the stability of the 

1827: DBL phase that is present at large $N$.  Specifically, we study the 

1828: effects of residual short range attractive interactions between the 

1829: two fermion flavors.  

1830: With vanishing D-eccentricity when the Fermi surfaces for both fermion

1831: species become the same, the $s$-wave Cooper channel is ``nested", and a 

1832: possible instability towards a paired BCS state seems likely. 

1833: The resulting state is a bosonic condensate, since 

1834: $\la b^\dagger \ra = \la d_1^\dagger d_2^\dagger \ra \ne 0$.

1835: However, a non-vanishing D-eccentricity of the Fermi surfaces destroys 

1836: the BCS nesting and the possible pairing instability.

1837: Indeed, we will find that the DBL with a large D-eccentricity can exist 

1838: as a stable phase.

1839: (For smaller D-eccentricity, an instability towards a finite momentum 

1840: Bose condensate or an incommensurate charge density wave co-existing 

1841: with the DBL is a possibility.)

1842: As is discussed in Sec.~\ref{sec:Wavefnc_props}, this is nicely in line 

1843: with the properties of the associated Gutzwiller wavefunctions:

1844: The $(\det)^2$ wavefunction that obtains in the limit of zero 

1845: D-eccentricity appears to have off-diagonal long range order,

1846: whereas the general $(\det)_x (\det)_y$ wavefunction exhibits power law 

1847: correlations consistent with the DBL, as extracted from the gauge theory.

1848: 

1849: 

1850: 

1851: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1852: \subsection{Formulation}

1853: \label{subsec:Fluct_formulation}

1854: 

1855: Being interested in the low energy properties, it is legitimate to 

1856: focus on the fermions living near the Fermi surfaces, just as in 

1857: Fermi liquid theory.  Moreover, the longitudinal Coulomb interactions

1858: mediated by the gauge field are readily screened out within an RPA 

1859: treatment, generating short-ranged screened density-density interactions.

1860: But the transverse fluctuations of the gauge field -- the photon -- are 

1861: incompletely screened by the fermion particle-hole excitations.  

1862: It is thus adequate to just retain the transverse components of the 

1863: vector potential.  Working in the Coulomb gauge with 

1864: $\bm{\nabla} \cdot {\bf a} = 0$, the vector potential reduces to a 

1865: scalar, e.g.,

1866: ${\bf a}({\bf k}) = a({\bf k}) \; {\rm sign}(k_y) 

1867: (k_y \hat{\bf x} - k_x \hat{\bf y}) / |{\bf k}|$. 

1868: 

1869: Moreover, for each fermion species, it will be sufficient to focus on a 

1870: pair of Fermi surface patches with normals parallel to some axis, 

1871: say $\hat{\bf x}$, as shown in Fig.~\ref{fig:patches}.   

1872: It is the $a_x$ component of the vector potential which is minimally 

1873: coupled to these fermions.  The important wave vectors of the gauge field

1874: $a_x({\bf k})$ that are strongly Landau damped by the fermions in these 

1875: patches satisfy $k_x \ll k_y$ (see Fig.~\ref{fig:patches}), 

1876: and in this region of momentum space $a_x({\bf k}) \approx a({\bf k})$.  

1877: Conversely, the modes of the gauge field $a({\bf k})$ with $k_x \ll k_y$

1878: feed back and scatter the patch fermions.

1879: Thus, we can focus on fermion fields $d_1({\bf k}), d_2({\bf k})$ and 

1880: the gauge field $a({\bf k})$ that are confined to their respective 

1881: patches in momentum space, as shown schematically in 

1882: Fig.~\ref{fig:patches}.

1883: 

1884: For concreteness, we assume that for each fermion species there are two 

1885: relevant patches on the opposite sides of the Fermi surface that are 

1886: labelled $s = R/L = +/-$ for the right/left patch as indicated in 

1887: Fig.~\ref{fig:patches}.  We then decompose the Fermion fields into

1888: right and left movers by writing,

1889: \begin{equation}

1890: d_{\alpha} ({\bf r}) \sim \sum_s 

1891: e^{i s {\bf k}_{F_\alpha} \cdot {\bf r}} d_{\alpha s}({\bf r}) ~.

1892: \end{equation}

1893: Here, ${\bf k}_{F_\alpha}$ denote the locations on the two Fermi surfaces

1894: which have normals aligned along the $x$-axis, and the fields 

1895: $d_{\alpha s}({\bf r})$ are assumed to be slowly varying on the scale

1896: of $k_F^{-1}$.  

1897: 

1898: 

1899: The full low energy action consists of three terms describing the 

1900: dynamics of the gauge field, of the fermions, and of their interactions:

1901: \begin{equation}

1902: S^{(0)} = \int d^2{\bf r} \, d\tau \, 

1903: [ {\cal L}_d + {\cal L}^{(0)}_a + {\cal L}_{int} ] ~.

1904: \end{equation}

1905: The ``patch" Lagrangian density for the fermions is simply,

1906: \begin{equation}

1907: {\cal L}_d =  \sum_{\alpha} \sum_{s=\pm} 

1908: d^\dagger_{\alpha s} [ \partial_\tau + s v_\alpha (-i \partial_x) 

1909:                        + \frac{v_\alpha c_\alpha }{2} (-i \partial_y)^2

1910:                       ] d_{\alpha s} ~,

1911: \end{equation}

1912: which is characterized by the local Fermi velocities, $v_\alpha$, 

1913: ($\alpha=1,2$), and the local Fermi surface curvatures, $c_\alpha$.

1914: The curvature is crucial in the fermion-gauge problem since there is an 

1915: effective decoupling of the fermions outside the patch from the 

1916: gauge field, $a = a_x$, which is parallel to the patch normals.

1917: Here $s=\pm 1$ specifies the sign of the group velocity while $v$ will 

1918: always denote the absolute value.  

1919: 

1920: The dynamics for the transverse gauge field describes the free photon,

1921: \begin{equation}

1922: {\cal L}_a^{(0)} = \frac{1}{2} [ (\partial_\tau a)^2 

1923:                                  + \kappa (\bm{\nabla} a )^2 ] ~,

1924: \end{equation}

1925: whereas the gauge-fermion interactions are of the usual minimal coupling 

1926: form:

1927: \begin{equation}

1928: \label{Lint}

1929: {\cal L}_{int} = \sum_{\alpha, s} g_\alpha s v_\alpha 

1930: d^\dagger_{\alpha s} \,a\, d_{\alpha s} ~.

1931: \end{equation} 

1932: Here, $g_1 = -g_2 \equiv g$ with $g=1$, but it will be convenient for 

1933: bookkeeping purposes to retain $g$ as a dimensionless parameter in the 

1934: theory.

1935: We emphasize that both the fermion fields and the gauge field, 

1936: which enter the above action in real space, are slowly varying fields

1937: corresponding to momentum modes inside their respective patches;

1938: in what follows, we always specify the wave vectors of the slow fields 

1939: relative to their respective patch centers.

1940: 

1941: 

1942: \begin{figure}

1943: \centerline{\includegraphics[width=\columnwidth]{figures/patches.eps}}

1944: \vskip -2mm

1945: \caption{Relevant momentum patches of the strongly coupled fermion -

1946: transverse gauge field system.

1947: For simplicity, here and throughout the text, we take the Fermi surface 

1948: patches with normals in the $\hat{\bf x}$ direction; 

1949: the required properties are the Fermi velocities and surface curvatures;

1950: the latter are crucial since differently oriented patch systems

1951: effectively decouple at low energies.

1952: }

1953: \label{fig:patches}

1954: \end{figure}

1955: 

1956: 

1957: 

1958: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1959: \subsection{RPA approximation for the gauge field dynamics}

1960: \label{subsec:RPA}

1961: 

1962: The above theory is strongly coupled and cannot be treated perturbatively

1963: in $g$.  As such, it is necessary to resort to approximations to make any

1964: meaningful progress.  Here we make our central approximation, namely 

1965: replacing the dynamics of the gauge field by a Landau damped form which 

1966: results if one sums the RPA bubble of particle-hole excitations:

1967: $S_a^{(0)} \to S_a$ with

1968: \begin{equation}

1969: \label{Sa}

1970: S_a = \frac{1}{2} \int_{{\bf q}\omega} 

1971: \left(\Gamma \frac{|\omega|}{|q_y|} + \chi |q_y|^2  \right) 

1972: |a ({\bf q}, \omega)|^2 ~.

1973: \end{equation}

1974: The momentum/frequency integral is understood as 

1975: $\int_{{\bf q} \omega} \equiv \int d^2{\bf q} d\omega / (2\pi)^3$,

1976: with the momentum restricted to a region satisfying

1977: $|q_x| < \Lambda_x, |q_y| < \Lambda_y$ with $\Lambda_x \ll  \Lambda_y$.

1978: 

1979: The magnitude of the Landau damping coefficient, $\Gamma$, will be 

1980: determined by the low-energy fermions near the specified Fermi surface 

1981: patches.  Within the RPA approximation one obtains,

1982: \begin{equation}

1983: \Gamma_{RPA} = \frac{g^2}{2\pi} \left( \frac{1}{c_1} + \frac{1}{c_2} \right) ~,

1984: \label{Gamma_RPA}

1985: \end{equation}

1986: but we will retain $\Gamma$ as an independent parameter since beyond 

1987: this point we will not be using RPA in any event.  

1988: It will sometimes prove convenient to consider the limit in which the 

1989: dimensionless gauge charge,

1990: \begin{equation}

1991: e^2 \approx \frac{g^2}{\Gamma c},

1992: \end{equation}

1993: is small.  Here $c$ is a characteristic Fermi surface curvature.

1994: Indeed, within the $N$-flavor extension discussed below, while $1/N$ will

1995: be the small parameter that leads to a controlled analysis, 

1996: already at order $1/N$ there are some physical properties which we can 

1997: only compute analytically when $e^2$ is simultaneously taken as a 

1998: small parameter.

1999: 

2000: The parameter $\chi$ that enters in $S_a$ is a ``stiffness'' which

2001: has to be positive for stability.  

2002: Starting from the lattice gauge theory formulation, $\chi$ gives a 

2003: measure of the energetic cost of setting up static internal fluxes.  

2004: One can very crudely estimate $\chi$ using the expression for the 

2005: diamagnetic response of free fermions,

2006: \begin{eqnarray}

2007: \chi &\approx & \frac{g^2}{12 \pi m_d} ~,

2008: \label{chi}

2009: \end{eqnarray}

2010: where $m_d$ is some effective mass.

2011: We reiterate that this stiffness comes from some energetics that is 

2012: assumed to stabilize the Fermi surfaces of the $d$ fermions 

2013: in the first place.  Therefore, from the low-energy perspective,

2014: $\chi$ should be viewed as a phenomenological parameter that encodes 

2015: some high-energy physics.  Furthermore, the stiffness $\chi$ is 

2016: independent of the patch orientation in momentum space for the 

2017: transverse gauge field $a({\bf k})$, since gauge invariance requires 

2018: that the field energy be proportional to 

2019: $\chi B^2 = \chi (\partial_x a_y - \partial_y a_x)^2$.  

2020: In any event, the magnitude of $\chi$ will play a minor role below.

2021: Indeed, for correlation functions with power law decay as we find below,

2022: changing $\chi$ will only modify the amplitude but not the form of the 

2023: correlators.

2024: 

2025: Before proceeding with an analysis of the full theory, 

2026: $S_d + S_a + S_{int}$, it is important to note that within the above 

2027: approximation the Landau damped dynamics of the gauge field described by 

2028: the action $S_a$ is both singular and harmonic.  We have ignored possible

2029: terms in the action which involve higher powers of the gauge field.

2030: As such, at this stage we could integrate out the gauge field exactly 

2031: generating a non-local and retarded four-fermion interaction.  

2032: It is the absence of non-linearities for the gauge field dynamics which 

2033: allows us to introduce a large-$N$ generalization which is soluble at 

2034: $N=\infty$ and which facilitates a formal and systematic $1/N$ expansion,

2035: as we now describe.

2036: 

2037: 

2038: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

2039: \subsection{$N$ flavor extension}

2040: \label{subsec:N_flavor}

2041: 

2042: Purely for technical reasons, then, we generalize the theory to $N$ 

2043: flavors of fermions at the two patches on each Fermi surface, 

2044: $d_{\alpha s} \to d_{i \alpha s}$.

2045: Here $i = 1, \dots, N$ labels the flavor index, while $s = \pm$ labels 

2046: the patch location and $\alpha = 1, 2$ the two Fermion species.  

2047: The action for the fermions is simply generalized as,

2048: \begin{equation}

2049: S_d^{(N)} = \sum_{i=1}^N \sum_{\alpha,s}  \int_{{\bf x} \tau} 

2050: d^\dagger_{i\alpha s} ( \partial_\tau 

2051: - i s v_\alpha \partial_x - \frac{v_\alpha c_\alpha}{2} \partial_y^2 ) 

2052: d_{i\alpha s} ~.

2053: \end{equation}

2054: The real scalar gauge field $a({\bf r})$ is likewise generalized, 

2055: but now as an $N \times N$ Hermitian matrix of complex fields, 

2056: $a_{ij}({\bf r}) = a_{ji}^*({\bf r})$.  When expressed in momentum space,

2057: each component of the matrix, $a_{ij}({\bf q})$, has dynamics

2058: with a Landau damped form,

2059: \begin{equation}

2060: S_a^{(N)} = \frac{1}{2} \sum_{i,j=1}^N \int_{{\bf q} \omega} 

2061: \left(\Gamma \frac{|\omega|}{|q_y|} + \chi |q_y|^2  \right) 

2062: |a_{ij} ({\bf q}, \omega)|^2 ~.

2063: \end{equation}

2064: Since $a_{ij}({\bf q}, \omega) = a_{ji}^*(-{\bf q}, -\omega)$, 

2065: this can also be re-written as a trace over the flavor indices of the 

2066: matrix $\overleftrightarrow{a}({\bf q}, \omega)$:

2067: \begin{equation}

2068: S_a^{(N)} = \frac{1}{2} \int_{{\bf q}\omega} 

2069: \left(\Gamma \frac{|\omega|}{|q_y|} + \chi |q_y|^2  \right) 

2070: {\rm Tr} [ \overleftrightarrow{a}({\bf q}, \omega) 

2071:            \overleftrightarrow{a}({\bf -q}, -\omega) ] ~.

2072: \end{equation}

2073: Finally, the fermion-gauge field interaction is taken as,

2074: \begin{equation}

2075: S^{(N)}_{int} = \sum_{\alpha,s} \frac{g_\alpha}{\sqrt{N}} 

2076: s v_\alpha \sum_{i,j=1}^N \int_{{\bf x} \tau} 

2077: d^\dagger_{i\alpha s} \, a_{ij} \, d_{j \alpha s} ~.

2078: \end{equation}

2079: Notice that the magnitude of the interaction strength (the charge) has

2080: been taken to scale as $1/\sqrt{N}$.

2081: At $N=1$ the theory reduces to that discussed in the previous subsection,

2082: with the assumption of a harmonic Landau damped form for the gauge field 

2083: dynamics as the only approximation.

2084: 

2085: It is worth pointing out that the full action has an enlarged global 

2086: symmetry being invariant under, 

2087: \begin{equation}

2088: d_{i \alpha s} \to e^{i \theta_i} e^{i \phi_\alpha} d_{i \alpha s} ~;

2089: \hskip0.5cm 

2090: a_{ij} \to e^{i(\theta_i - \theta_j)} a_{ij} ~,

2091: \end{equation}

2092: for arbitrary phases $\theta_i$, $i=1, \dots, N$, and $\phi_{\alpha}$, 

2093: $\alpha = 1,2$.  Thus, the densities 

2094: $\sum_\alpha d_{i\alpha}^\dagger d_{i\alpha}$ are independently conserved

2095: for any $i$, and also $\sum_i d_{i\alpha}^\dagger d_{i\alpha}$ are

2096: independently conserved for any $\alpha$.

2097: Note also that the dynamics of each of the $N$ flavor fermions $d_{i1}$ 

2098: is identical to one another, and similarly for all the $d_{i2}$ fermions.

2099: 

2100: 

2101: Before solving the full $N$-flavor theory, 

2102: $S_d^{(N)} + S_a^{(N)} + S_{int}^{(N)}$, exactly at $N=\infty$,  

2103: we define large $N$ generalizations of the various correlation functions.

2104: As in the gauge mean field theory, we anticipate that the exact fermion 

2105: Green's function for (any) one of the $N$ fermion flavors, which we 

2106: denote $G^{(N)}_{d_\alpha}({\bf r})$, will be dominated at large 

2107: distances by the patch fermions with normals parallel or antiparallel

2108: to the observation direction $\hat{\bf r}$.

2109: We can thus expand $G^{(N)}_{d_\alpha}({\bf r})$ for large $|{\bf r}|$ 

2110: in terms of the ``patch" fermion Green's functions as,

2111: \begin{equation}

2112: G^{(N)}_{d_\alpha}({\bf r}, \tau) \approx \sum_s 

2113: e^{i s {\bf k}_{F_\alpha} \cdot {\bf r}} G^{(N)}_{\alpha s}({\bf r},\tau)

2114: ~,

2115: \end{equation}

2116: with the definition,

2117: \begin{equation}

2118: G^{(N)}_{\alpha s} ({\bf r}-{\bf r}^\prime, \tau-\tau^\prime) = 

2119: \la d^\dagger_{i\alpha s}({\bf r}, \tau) 

2120:     d_{i\alpha s}({\bf r}^\prime, \tau^\prime) \ra ~.

2121: \end{equation}

2122: Note that there is no summation over $i$; throughout, any summation over

2123: indices will be shown explicitly.

2124: 

2125: There are now $N$ flavors of bosons with creation operators,

2126: \begin{equation}

2127: b_i^\dagger({\bf r}) = d_{i1}^\dagger({\bf r}) d_{i2}^\dagger({\bf r}) ~,

2128: \end{equation}

2129: with all $N$ boson flavors having identical dynamics.

2130: The large $N$ generalization of the boson correlator is then simply,

2131: \begin{equation}

2132: G_{b;ij}^{(N)} ({\bf r}-{\bf r}^\prime, \tau-\tau^\prime) = 

2133: \la b_i^\dagger({\bf r}, \tau) b_j({\bf r}^\prime, \tau^\prime) \ra ~.

2134: \end{equation}

2135: Due to the global symmetries, the correlators which are off-diagonal in 

2136: the flavor indices vanish, leaving a uniquely defined boson Green's 

2137: function, $G_{b;ij}^{(N)} = \delta_{ij} G_{b}^{(N)}$.  

2138: As desired, $G_b^{(N)}$ reduces at $N=1$ to the boson correlator 

2139: defined in Sec.~\ref{subsec:MF_props}.

2140: 

2141: The boson density-density correlator, $D_b^{(N)}$, is readily defined in 

2142: terms of (any) of the boson densities $\hat\rho_{ib}$ as in 

2143: Eq.~(\ref{densitycorr:def}).  

2144: Similarly, we can define the fermion density-density correlators for 

2145: each species, $D^{(N)}_{d_\alpha}$, in terms of any one of the fermion 

2146: densities, $\hat\rho_{i\alpha}$.

2147: As before, we assume that the boson density-density correlator can be 

2148: approximated as an average over $\alpha$ of the fermion density-density 

2149: correlators, 

2150: $D_b^{(N)} \approx \frac{1}{2} \sum_\alpha D^{(N)}_{d_\alpha}$.

2151: 

2152: 

2153: In order to extract the above correlation functions, it is convenient to

2154: define patch fermion particle-hole and particle-particle bubbles, 

2155: which we denote as $\Pi^{(N)}_{ph}$ and $\Pi^{(N)}_{pp}$, respectively.

2156: The exact bubble correlators can be formally expressed in terms of the 

2157: exact patch fermion Green's functions and the exact particle-hole and 

2158: particle-particle vertices,

2159: \begin{widetext}

2160: \begin{eqnarray}

2161: \Pi^{(N)}_{ph;\, \alpha s; \alpha^\prime s^\prime}({\bf q}, \omega) &=& 

2162: \int_{{\bf q}^\prime \omega^\prime} 

2163: V^{(N)}_{ph;\, \alpha s; \alpha^\prime s^\prime}

2164: ({\bf q} + {\bf q}^\prime, \omega + \omega^\prime; 

2165:  {\bf q}^\prime, \omega^\prime) \,

2166: G^{(N)}_{\alpha s}({\bf q} + {\bf q}^\prime, \omega + \omega^\prime) \,

2167: G^{(N)}_{\alpha^\prime s^\prime}({\bf q}^\prime, \omega^\prime) ~, \\

2168: \Pi^{(N)}_{pp;\, \alpha s; \alpha^\prime s^\prime}({\bf q}, \omega) &=& 

2169: \int_{{\bf q}^\prime \omega^\prime} 

2170: V^{(N)}_{pp;\, \alpha s; \alpha^\prime s^\prime}

2171: ({\bf q} + {\bf q}^\prime, \omega + \omega^\prime; 

2172:  {\bf q}^\prime, \omega^\prime) \,

2173: G^{(N)}_{\alpha s}({\bf q} + {\bf q}^\prime, \omega + \omega^\prime) \,

2174: G^{(N)}_{\alpha^\prime s^\prime}(-{\bf q}^\prime, -\omega^\prime) ~.

2175: \label{bubble;vertex;Greens}

2176: \end{eqnarray}

2177: \end{widetext}

2178: Here $V^{(N)}_{ph}$ and $V^{(N)}_{pp}$ denote the fully renormalized 

2179: vertices, and $\alpha,s$ and $\alpha^\prime,s^\prime$ label the patch 

2180: locations of the two fermion lines that come out of the vertex.  

2181: The total momentum and frequency running through the bubble is 

2182: ${\bf q},\omega$, and is divided between the two fermions as shown above.

2183: At $N=\infty$ there are no vertex corrections, 

2184: $V^{(\infty)}_{ph} = 1$ and $V^{(\infty)}_{pp} = 1$, and each bubble

2185: is just a convolution of the two Green's functions.

2186: 

2187: The boson correlator at large distances and long times can be readily 

2188: expressed in terms of the bubbles,

2189: \begin{equation}

2190: G_b^{(N)}(s {\bf k}_{F_1} + s^\prime {\bf k}_{F_2} + {\bf q}, \omega) 

2191: \approx \Pi^{(N)}_{pp;\, 1s; 2s^\prime}({\bf q}, \omega) ~,

2192: \end{equation}

2193: and similarly for the fermion density-density correlator,

2194: \begin{equation}

2195: D_{d_\alpha}^{(N)}(s {\bf k}_{F_\alpha} - s^\prime {\bf k}_{F_\alpha}

2196:                    + {\bf q}, \omega) 

2197: \approx \Pi^{(N)}_{ph;\, \alpha s; \alpha s^\prime}({\bf q}, \omega) ~.

2198: \end{equation}

2199: 

2200: 

2201: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

2202: \subsection{Solution at $N=\infty$ }

2203: \label{subsec:N_infinity}

2204: 

2205: At $N=\infty$ the fermion Green's function 

2206: $G^{(N)}_{\alpha s}({\bf q}, \omega)$ can be readily obtained by 

2207: summing the nested rainbow diagrams to give,

2208: \begin{equation}

2209: G^{(\infty)}_{\alpha s}({\bf q}, \omega)

2210: = \frac{1} {i\omega \left(1 + |\Omega_\alpha /\omega|^{1/3} \right) 

2211:             - s v_\alpha q_x - \frac{v_\alpha c_\alpha}{2} q_y^2 }

2212: \label{N=infinity_Greens}

2213: \end{equation}

2214: with

2215: \begin{equation}

2216: \Omega_\alpha^{1/3} = \frac{v_\alpha}{2 \pi \sqrt{3}} 

2217: \frac{g^2}{\chi^{2/3} \Gamma^{1/3}} ~.

2218: \label{Omega}

2219: \end{equation}

2220: Because of the gauge interactions, each fermion species obtains an 

2221: anomalous self-energy of the form $-i\omega |\Omega/\omega|^{1/3}$ 

2222: which dominates over the bare term $-i\omega$ on energy scales below 

2223: $\Omega$.  Physically, the fermions cease to propagate as free particles 

2224: due to the strong scattering from the dynamical gauge field, 

2225: becoming ``incoherent".

2226: 

2227: We first consider the long distance and time behavior of 

2228: $G^{(\infty)}_{\alpha s}({\bf r},\tau)$.

2229: As in the mean field Sec.~\ref{sec:MFT}, the Green's function is 

2230: dominated at large distances by the patches of the Fermi surface with 

2231: normals along $\pm \hat{\bf r}$.  

2232: For the equal-time Green's function, we obtain,

2233: \begin{equation}

2234: G^{(\infty)}_{d_\alpha} ({\bf r}, 0) \approx \frac{3 \sqrt{\pi}}{4} 

2235: \left( \frac{v_\alpha}{\Omega_\alpha} \right)^{1/2} 

2236: \frac{1}{\sqrt{|{\bf r}|}} G_{d_\alpha}^{MF}({\bf r}) ~,

2237: \end{equation}

2238: where the velocity $v_\alpha$ and curvature $c_\alpha$ characterize the 

2239: $d_\alpha$ Fermi surface patches with normals along $\pm \hat{\bf r}$, 

2240: and $G_{d_\alpha}^{MF}({\bf r})$ is given in Eq.~(\ref{Gd_MF}).

2241: On the other hand, the long time dependence of the local fermion Green's 

2242: function is not affected by the anomalous self-energy and is determined 

2243: solely by the density of states $\nu_0$ at the Fermi energy,

2244: \begin{equation}

2245: G^{(\infty)}_{d_\alpha} ({\bf 0}, \tau) = -\frac{\nu_0}{\tau} ~.

2246: \end{equation}

2247: 

2248: 

2249: Turning to the physical correlators,

2250: at $N=\infty$ there are no vertex corrections to the fermion bubble, 

2251: so we have simply,

2252: \begin{equation}

2253: D^{(\infty)}_{d_\alpha}({\bf r}) = - |G^{(\infty)}_{d_\alpha}({\bf r})|^2

2254: ~,

2255: \end{equation}

2256: which gives for the $N=\infty$ equal-time density correlator,

2257: \begin{equation}

2258: D^{(\infty)}_{b}({\bf r}) \approx \frac{1}{2} \sum_\alpha \frac{9\pi}{16}

2259: \frac{v_\alpha}{\Omega_\alpha}

2260: \frac{1}{|{\bf r}|} D^{MF}_{d_\alpha}({\bf r}) ~,

2261: \end{equation}

2262: where $D^{MF}_{d_\alpha}({\bf r})$ is given in Eq.~(\ref{DD_MF}).

2263: Similarly, in the absence of vertex corrections at $N=\infty$ in the 

2264: particle-particle bubble, one has for the equal-time boson Green's 

2265: function, 

2266: \begin{eqnarray}

2267: G_b^{(\infty)} ({\bf r}) &=& G_{d_1}^{(\infty)} ({\bf r})  

2268: G_{d_2}^{(\infty)} ({\bf r}) \\

2269: &\approx& 

2270: \frac{9\pi}{16}

2271: \left( \frac{v_1 v_2}{\Omega_1 \Omega_2} \right)^{1/2} 

2272: \frac{1}{|{\bf r}|} G_b^{MF}({\bf r}) ~,

2273: \end{eqnarray}

2274: where $G_b^{MF}({\bf r})$ is given in Eq.~(\ref{Gb_MF}).

2275: However, the time dependence of the local boson Green's function 

2276: $G_b({\bf 0}, \tau)$ is unchanged from the mean field Eq.~(\ref{Gtau}).

2277: Finally, the box correlator is given by Eq.(\ref{Bb_MF}) with 

2278: appropriate $N=\infty$ fermion Green's functions and decays

2279: with a power law envelope of $-x^{-8}$.

2280: 

2281: 

2282: Notice that both the boson and the boson density-density correlators

2283: at $N=\infty$ fall off in space as $|{\bf r}|^{-4}$, which is faster 

2284: than their mean field counterparts.  Injecting a boson corresponds to 

2285: creating two fermions.  In the mean field, these fermions propagate 

2286: independently and as free particles.  On the other hand, in the 

2287: present $N=\infty$ theory that sums the nested rainbow diagrams, 

2288: both fermions propagate ``incoherently" due to their scattering from the 

2289: gauge field, which effectively reduces the ability of the created boson 

2290: to propagate, thereby leading to a faster decay of the boson correlator.

2291: At this level of approximation, the two injected fermions, while 

2292: scattered by the gauge fluctuations, do not scatter off one another.  

2293: The effects of interactions between the two injected fermions appear at 

2294: order $1/N$, entering as vertex corrections.  Since the two fermion 

2295: species have opposite gauge charge, one expects that they will attract 

2296: one another.  As we shall see, though, they do not form a composite 

2297: boson (a bound state, for example).

2298: But their motion becomes strongly correlated allowing them to propagate 

2299: more effectively when close by spatially.  The net result, as we find 

2300: below, is that the boson can propagate more effectively when the 

2301: motion of the fermion pair is correlated, leading to a slower decay 

2302: of the correlator compared to the $N=\infty$ result.

2303: Thus, the effect on the boson dynamics due to the decoherence 

2304: experienced by the fermions from scattering off the gauge fluctuations

2305: is compensated, perhaps only partially, by the attractive interaction 

2306: between the pair of fermions.

2307: 

2308: 

2309: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

2310: \subsection{Fermion Green's function at order $1/N$}

2311: \label{subsec:1_N_fermion_Greens}

2312: 

2313: The $1/N$ contributions to the fermion Green's function are obtained

2314: by considering the nested rainbow diagrams and replacing one gauge field 

2315: propagator 

2316: $G_a({\bf q}, \omega) = (\Gamma |\omega|/|q_y| + \chi |q_y|^2)^{-1}$ 

2317: with the bubble correction

2318: $\delta G_a \sim \frac{1}{N} G_a^2 \Pi_{ph}^{(\infty)}$.

2319: Formally, we can first calculate the rainbows using $G_a + \delta G_a$

2320: everywhere and then extract the $1/N$ piece.

2321: As we now argue, the functional form of the fermion Green's function 

2322: in Eq.~(\ref{N=infinity_Greens}) remains unchanged except for finite 

2323: $1/N$ shifts in the parameters.

2324: Indeed, evaluating the $N=\infty$ particle/hole bubble at small 

2325: wave vector and frequency gives 

2326: $\Pi^{(\infty)}_{ph}({\bf q}, \omega) \sim |\omega|/|q_y|$, 

2327: which only leads to a finite $1/N$ shift in $\Gamma$.

2328: This is not surprising, since the bare singular gauge propagator

2329: postulated at the outset in our theory is motivated by the free fermion 

2330: particle-hole bubble, and one can verify that the bubble remains unchanged

2331: also in the presence of an arbitrary fermion self-energy that depends 

2332: on the frequency only.

2333: Then, using this renormalized gauge propagator to evaluate the rainbow 

2334: diagram gives a contribution to the fermion Green's function of the form,

2335: $\delta G_{d_\alpha}^{(N)} \sim \frac{1}{N} \omega^{2/3} 

2336: [G^{(\infty)}_{d_\alpha}]^2$.

2337: This will at most give finite $1/N$ shifts to 

2338: $\Omega_\alpha, v_\alpha, c_\alpha$. 

2339: Next we consider the vertices at $1/N$, which undergo more dramatic 

2340: modifications.

2341: 

2342: 

2343: 

2344: 

2345: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

2346: \subsection{Vertices at order $1/N$}

2347: \label{subsec:1_N_vertex}

2348: 

2349: \subsubsection{Single photon contribution}

2350: \label{subsubsec:V1a}

2351: 

2352: Consider first dressing the bare vertices with a single gauge propagator.

2353: We denote the value of the corresponding Feynman diagrams as

2354: $V^{(1a)}_{ph}$ and $V^{(1a)}_{pp}$, where $(1a)$ stands for one 

2355: (Landau damped) photon exchanged.  These diagrams are the

2356: first ones in Figs.~\ref{fig:Vph}~and~\ref{fig:Vpp}.

2357: As we will discuss below, the full structure at $1/N$ is more involved, 

2358: but the calculation of $V^{(1a)}$ already captures the main qualitative

2359: physics.

2360: 

2361: For simplicity, we first focus on the situation with zero external $y$ 

2362: momentum and frequency, defining,

2363: \begin{equation}

2364: V^{(1a)}_{q_x^-} = 

2365: V^{(1a)}({\bf q}, \omega; {\bf q}^\prime, \omega^\prime)|_{q_y=q_y^\prime=\omega=\omega^\prime=0} ~,

2366: \end{equation}

2367: where ${\bf q},{\bf q}^\prime$, $\omega,\omega^\prime$ denote 

2368: the momentum/frequencies in and out of the two legs and 

2369: $q_x^- = q_x - q_x^\prime$ is the momentum running through the vertex.  

2370: For notational simplicity we have suppressed the corresponding flavor 

2371: labels, $\alpha,s; \alpha^\prime,s^\prime$.

2372: Evaluating the Feynman diagrams gives,

2373: \begin{eqnarray}

2374: V^{(1a)}_{ph;\, q_x^-} &=& 

2375: \frac{1}{N} \,{\rm sign}(g_\alpha g_{\alpha^\prime})\, \delta_{s, -s^\prime}

2376: \lambda_{ph} \ln(\Lambda_x/|q_x^-|) ~, 

2377: \label{Vph1a}

2378: \\

2379: V^{(1a)}_{pp;\, q_x^-} &=& 

2380: -\frac{1}{N} \,{\rm sign}(g_\alpha g_{\alpha^\prime})\, \delta_{s, -s^\prime}

2381: \lambda_{pp} \ln(\Lambda_x/|q_x^-|) ~,

2382: \label{Vpp1a}

2383: \end{eqnarray}

2384: revealing a logarithmic divergence cut off by the total $x$-momentum 

2385: running through the vertex.

2386: Explicit expressions for the dimensionless non-negative coefficients 

2387: $\lambda_{ph}$ and $\lambda_{pp}$ are given below.

2388: 

2389: Notice that both the particle-hole and the particle-particle 

2390: vertex corrections vanish unless the two patches have opposite group 

2391: velocities.  Moreover, the sign of the correction is given by the 

2392: sign of the product of the two charges and differs for the particle-hole 

2393: and particle-particle channels.  The origin of these signs is essentially

2394: Ampere's law familiar from electrodynamics. 

2395: Two moving charges producing parallel charge currents experience an 

2396: attractive interaction, bringing them closer together and enhancing their

2397: interaction strength, which is encapsulated by the magnitude of the 

2398: interaction vertex. 

2399: For example, a particle and a hole of the same fermion species 

2400: ($\alpha = \alpha^\prime$) residing on the opposite Fermi surface patches

2401: ($s = -s^\prime$) have an enhanced interaction strength, which is

2402: encoded in $V_{ph}^{(1a)} > 0$.

2403: On the other hand, the interaction vertex is suppressed when the

2404: charge currents are anti-parallel.

2405: For example, a pair of particles with the same charge ($g = g^\prime$) 

2406: experience a suppressed vertex interaction, $V_{pp}^{(1a)} < 0$, 

2407: when their group velocities are antiparallel.

2408: 

2409: The absence of singular vertex corrections when the two patch fermions

2410: have parallel group velocities, $s = s^\prime$, is, we believe, not just 

2411: a peculiarity of this lowest order contribution, but will be valid 

2412: generally for the exact vertices at arbitrary $N$.  This expectation is 

2413: based on the specific form of the propagators: the fermion propagator 

2414: has simple poles in the complex $q_x$ plane while the gauge field

2415: propagator is independent of $q_x$, the latter a reflection of the 

2416: transverse nature of the gauge field.

2417: For any given Feynman diagram, it is the sign of the internal frequency

2418: variables running around the loops together with $s, s^\prime$ 

2419: which will determine whether the poles in the integrand will be in the 

2420: upper or the lower complex $q_x$ planes.

2421: For vertices with $s = s^\prime$ all of the fermion lines internal to 

2422: any diagram will have the same group velocity, and with energy 

2423: (frequency) conservation one expects that at least one of the 

2424: $q_x$ variables will only have poles in the upper (or lower) half-plane, 

2425: so that the contour can be appropriately deformed to show that the 

2426: diagram vanishes.  

2427: Thus, we henceforth restrict attention exclusively to vertices with

2428: $s = -s^\prime$, and to ease the notation will drop the explicit

2429: $s, s^\prime$ indices in the following.

2430: 

2431: The dimensionless magnitudes of the vertex enhancements in the 

2432: particle-hole and the particle-particle channel, 

2433: $\lambda_{ph}$ and $\lambda_{pp}$ respectively, are given by,

2434: \begin{eqnarray}

2435: \lambda_{ph} &=& 

2436: \frac{g^2}{\sqrt{3}\pi \Gamma^{1/3} \chi^{2/3}

2437:            \left( \frac{\Omega^{1/3}}{v} 

2438:                  + \frac{\Omega^{\prime 1/3}}{v^\prime} \right)} 

2439: {\cal E}[\zeta_{ph}] ~, 

2440: \label{lambda_ph}

2441: \\

2442: \lambda_{pp} &=& 

2443: \frac{g^2}{\sqrt{3}\pi \Gamma^{1/3} \chi^{2/3}

2444:            \left( \frac{\Omega^{1/3}}{v} 

2445:                  + \frac{\Omega^{\prime 1/3}}{v^\prime} \right)} 

2446: {\cal E}[\zeta_{pp}] ~;

2447: \label{lambda_pp}

2448: \\

2449: \lambda_{ph} &=& {\cal E}[\zeta_{ph}] ~,  \quad\quad

2450: \lambda_{pp} = {\cal E}[\zeta_{pp}] ~.

2451: \label{lambda_Ninfty}

2452: \end{eqnarray}

2453: In the last line, we specialized to the $N=\infty$ expressions for 

2454: $\Omega$ and $\Omega^\prime$ (Eq.~\ref{Omega}),

2455: which is valid for extracting the order $1/N$ vertex corrections.

2456: Here, ${\cal E}[\zeta]$ is a dimensionless function of a dimensionless 

2457: argument given by

2458: \begin{widetext}

2459: \begin{equation}

2460: {\cal E}[\zeta] = \frac{3\sqrt{3}}{2\pi} \int_0^\infty \frac{t dt}{(1+t^3) (1 + \zeta^2 t^4)} 

2461:  =  \frac{3\pi\zeta^5 + 8\pi(1-\zeta^4)/\sqrt{3}

2462: - 3\sqrt{2}\pi(\zeta^{1/2} - \zeta^{7/2}) - 6\zeta^2 \ln\zeta}

2463: {8\pi (1+\zeta^6) / \sqrt{3}} ~,

2464: \label{Ffunc}

2465: \end{equation}

2466: \end{widetext}

2467: and has been normalized so that ${\cal E}[0]=1$.  

2468: This function is monotonically decreasing with increasing argument

2469: varying as ${\cal E}[\zeta] \approx \frac{3 \sqrt{3}}{8 \zeta}$ for 

2470: $\zeta \to \infty$.

2471: The parameters $\zeta_{ph}$ and $\zeta_{pp}$ are given by,

2472: \begin{eqnarray}

2473: \zeta_{ph} &=& 

2474: \frac{ \Gamma^{2/3} (c + c^\prime)}

2475:      { 2 \chi^{2/3}

2476:        \left( \frac{\Omega^{1/3}}{v} 

2477:              + \frac{\Omega^{\prime 1/3}}{v^\prime} \right) } ~, 

2478: \label{zeta_ph} \\

2479: \zeta_{pp} &=& 

2480: \frac{ \Gamma^{2/3} |c - c^\prime|}

2481:      { 2 \chi^{2/3}

2482:        \left( \frac{\Omega^{1/3}}{v} 

2483:              + \frac{\Omega^{\prime 1/3}}{v^\prime} \right) } ~;

2484: \label{zeta_pp} \\

2485: \zeta_{ph} &=& \frac{\sqrt{3}\pi}{2 g^2} \Gamma (c + c^\prime) ~, 

2486: \quad\;

2487: \zeta_{pp} = \frac{\sqrt{3}\pi}{2 g^2} \Gamma |c - c^\prime| ~.

2488: \label{zeta_Ninfty}

2489: \end{eqnarray}

2490: In the last line, we again used the $N=\infty$ values for 

2491: $\Omega,\Omega^\prime$.

2492: Notice that the vertex interaction strength between two fermions on 

2493: opposing patches depends on their particular Fermi surface curvatures 

2494: $c, c^\prime$.  Moreover, in the particle-particle channel the 

2495: interaction strength is maximal when the curvatures are equal.  

2496: This reflects a particle-particle nesting, which is responsible for the 

2497: BCS pairing instability in the Fermi liquid context.

2498: 

2499: 

2500: \begin{figure}

2501: \centerline{\includegraphics[width=2.0in]{figures/Vph.eps}}

2502: \vskip -2mm

2503: \caption{

2504: Ladder diagrams that contribute to $V_{ph}$ at $1/N$.

2505: }

2506: \label{fig:Vph}

2507: \end{figure}

2508: 

2509: 

2510: \begin{figure}

2511: \centerline{\includegraphics[width=2.0in]{figures/Vpp.eps}}

2512: \vskip -2mm

2513: \caption{

2514: Crossed diagrams that contribute to $V_{pp}$ at $1/N$.

2515: }

2516: \label{fig:Vpp}

2517: \end{figure}

2518: 

2519: 

2520: \vskip 2mm

2521: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

2522: \subsubsection{Photon ladders}

2523: 

2524: Having evaluated the one-photon corrections $V^{(1a)}$ to the bare 

2525: vertices, we now discuss the full set of diagrams that appear at 

2526: order $1/N$ in each case.

2527: First consider the fermion particle-hole vertex which enters into the

2528: density-density correlator: 

2529: $V^{(N)}_{\rho_\alpha} = V_{ph;\, \alpha, \alpha}^{(N)}$.

2530: The $1/N$ diagrams are shown in Fig.~\ref{fig:Vph} and contain an 

2531: arbitrary number of non-crossing photon lines,\cite{footnote:N_flavor} 

2532: beginning with the one-photon diagram $V_{ph}^{(1a)}$, Eq.~(\ref{Vph1a}).  

2533: We shall denote the contribution from the diagram with $n$ photon lines 

2534: as $V_{ph}^{(na)}$.  The magnitude of the singular contributions from 

2535: each diagram in the ladder sum will depend on a dimensionless gauge 

2536: charge,

2537: \begin{equation}

2538: e_\alpha^2 \equiv g^2/\Gamma c_\alpha ~.

2539: \end{equation}

2540: Above we found

2541: $V_{ph}^{(1a)} = (1/N) {\cal E}[\sqrt{3}\pi/e_\alpha^2] \ln(1/q_x^-)

2542: \equiv (1/N) A^{(1)}_1 \ln(1/q_x^-)$.

2543: The singular contributions from the diagram with two photon lines 

2544: can also be extracted from the Feynman diagram,

2545: $V_{ph}^{(2a)} = (1/N) 

2546: [ A^{(2)}_1 \ln(1/q_x^-) + A^{(2)}_2 \ln^2(1/q_x^-) ]$, 

2547: with $A^{(2)}_1, A^{(2)}_2$ universal dimensionless functions

2548: of $e_\alpha^2$.

2549: From this we infer the general structure for the diagram with $n$ 

2550: photon lines,

2551: \begin{equation}

2552: V_{ph}^{(na)} \sim \frac{1}{N} \sum_{m=1}^n A^{(n)}_m \ln^m(1/q_x^-) ~.

2553: \end{equation}

2554: This form suggests that the full set of $1/N$ ladder diagrams can

2555: be exponentiated,

2556: \begin{equation}

2557: V_{\rho_\alpha}^{(N)} 

2558: \sim \frac{1}{N} \exp[\gamma_\alpha \ln(1/q_x^-)] 

2559: \sim \frac{1}{N} |q_x - q_x^\prime|^{-\gamma_\alpha} ~,

2560: \label{ph_ladder_sum}

2561: \end{equation}

2562: with $\gamma_\alpha = \sum_{n=1}^\infty A^{(n)}_1$.

2563: The validity of this exponentiation can be justified by formally 

2564: summing the ladder series to obtain an integral equation for the full 

2565: vertex.  The integral equation has a singular kernel and its solution is 

2566: expected to have the power-law form with some exponent $\gamma_\alpha$ 

2567: which will depend on the dimensionless charge $e_\alpha^2$.

2568: 

2569: Unfortunately, we have been unable to solve the integral equation

2570: to obtain $\gamma_\alpha$.

2571: However, we observe that for small charge $e_\alpha^2$,

2572: $A^{(1)}_1 = 3 e_\alpha^2 / (8\pi) + O(e_\alpha^4)$.

2573: This gives the leading behavior even though $A^{(1)}_1$ is not analytic 

2574: in $e_\alpha^2$, as can be seen from Eq.~(\ref{Ffunc});

2575: more generally, one finds $A^{(n)}_m = O(e_\alpha^{2n})$.

2576: Thus, assuming small parameter $e_\alpha^2$, the leading contribution to 

2577: $\gamma_\alpha$ comes just from the single photon diagram,

2578: \begin{equation}

2579: \gamma_\alpha = \frac{3}{8\pi} e_\alpha^2 + O(e_\alpha^4, 1/N) ~.

2580: \label{anomalous;density;exponent}

2581: \end{equation}

2582: 

2583: For $e_\alpha^2$ of order one, the exponent $\gamma_\alpha$, while 

2584: obtained by summing diagrams at order $1/N$, is itself of order one.

2585: It is the amplitude of the vertex in Eq.~(\ref{ph_ladder_sum})

2586: which is of order $1/N$.

2587: This reflects the particular structure of the large $N$ theory,

2588: which, constructed to sum the nested rainbows for the fermion self energy

2589: at leading order, also exponentiates the ladder sum for the 

2590: particle-hole vertex at order $1/N$.

2591: Notice that $\gamma_\alpha$ monotonically decreases as the curvature

2592: of the Fermi surface is taken large, due to the increasingly restrictive 

2593: phase space requirements on the particle-hole vertex in this limit.  

2594: We expect that this trend is valid more generally, for larger 

2595: $e_\alpha^2$ and for arbitrary $N$.

2596: 

2597: 

2598: Summarizing the discussion of the particle-hole vertex, we can also 

2599: write a scaling form for more general external momenta and frequencies,

2600: \begin{equation}

2601: V^{(N)}_{\rho_\alpha} \sim |q_x-q_x^\prime|^{-\gamma_\alpha}

2602: \tilde{V}_{\rho_\alpha} 

2603: \left(\frac{q_y^2}{q_x^-}, \frac{q_y^{\prime 2}}{q_x^-},

2604:       \frac{\omega^{2/3}}{q_x^-}, \frac{\omega^{\prime 2/3}}{q_x^-} 

2605: \right) ~.

2606: \label{Vrho;scaling}

2607: \end{equation}

2608: 

2609: 

2610: \vskip 2mm

2611: 

2612: We next consider the vertex which enters in the boson-boson correlator, 

2613: $V_b^{(N)} = V^{(N)}_{pp;\, 1,2}$.

2614: The full set of diagrams that contribute to the particle-particle vertex 

2615: are shown in Fig.~\ref{fig:Vpp} and contain an arbitrary number of 

2616: maximally-crossed photon lines.\cite{footnote:N_flavor}

2617: The first term is the one-photon diagram that we evaluated in 

2618: Eq.~(\ref{Vpp1a}), 

2619: $V_{pp;\, 1,2}^{(1a)} = (1/N) {\cal E}[\zeta_{pp}] \ln(1/q_x^-)

2620: \equiv (1/N) B_1 \ln(1/q_x^-)$

2621: (plus non-singular contributions).

2622: An evaluation of the two-photon diagram gives the same form,

2623: $V_{pp;\, 1,2}^{(2a)} = (1/N) B_2 \ln(q_x^-)$.

2624: In contrast to the particle-hole case, here the two-photon diagram 

2625: does not contain a log-squared term, which indicates that the crossed 

2626: ladder series cannot be exponentiated.   While we cannot evaluate the 

2627: higher order diagrams, we expect that each term will have a log 

2628: contribution with some dimensionless amplitudes $B_n$.

2629: If this is the case, the full structure of the particle-particle vertex 

2630: up to order $1/N$ is,

2631: \begin{equation}

2632: V_b^{(N)} = 1 + \frac{\eta_1 \ln(\Lambda_x/q_x^-)}{N} + O(1/N^2) ~,

2633: \label{bosonvertex_1/N}

2634: \end{equation}

2635: with $\eta_1 = \sum_{n=1}^\infty B_n$.

2636: To check whether the latter series can be defined in principle,

2637: we again take the limit of small dimensionless charges $e_\alpha^2$.

2638: Consider the least favorable case with matched Fermi surface curvatures,

2639: $c_1 = c_2$, so $e_1^2 = e_2^2 \equiv e^2 = g^2/(\Gamma c)$.

2640: We find that the first term is independent of $e^2$,

2641: $B_1 = 1$.  However, evaluating the two-photon diagram gives

2642: $B_2 \sim (e^2)^2 \ln^3(1/e^2)$ for small $e^2$.

2643: The diagrams are expected to be smaller when $c_1 \neq c_2$

2644: (for example, even the first term $B_1 = {\cal E}[\zeta_{pp}]$

2645: can be made small when $g^2 \ll \Gamma |c_1 - c_2|$).

2646: We thus conjecture that the series converges for small enough

2647: $e_\alpha^2$.

2648: 

2649: Provided it is legitimate to exchange the order of limits $N \to \infty$ 

2650: and $q_x \to 0$ in Eq.~(\ref{bosonvertex_1/N}),

2651: the  boson vertex can be written in a power law form,

2652: \begin{equation}

2653: V_b^{(N)}({\bf q}, \omega; {\bf q}^\prime, \omega^\prime)

2654: |_{q_y=q_y^\prime=\omega=\omega^\prime=0} 

2655: \sim |q_x - q_x^\prime|^{-\eta} ~,

2656: \end{equation}

2657: with

2658: \begin{equation}

2659: \eta = \eta_1/N + O(1/N^2) ~.

2660: \label{eta_1/N}

2661: \end{equation}  

2662: We will discuss the legitimacy of this exponentiation procedure in 

2663: Sec.~\ref{sec:eff_FT}.

2664: For more general external momenta and frequencies the full boson 

2665: vertex will satisfy a scaling form,

2666: \begin{equation}

2667: V_b^{(N)}({\bf q}, \omega; {\bf q}^\prime, \omega^\prime) \sim  

2668: |q_x - q_x^\prime|^{-\eta} \,

2669: \tilde{V}_b \left(\frac{q_y^2}{q_x^{-}}, \frac{q_y^{\prime 2}}{q_x^-},

2670:                   \frac{\omega^{2/3}}{q_x^-}, 

2671:                   \frac{\omega^{\prime 2/3}}{q_x^-} \right) ~,

2672: \label{Vb;scaling}

2673: \end{equation}

2674: where we have used the fact that the vertex depends only on the

2675: total $x$-momentum running through it, $q_x^- = q_x - q_x^\prime$.

2676: The scaling function will vary as, $\tilde{V}_b(X,0,0,0) \sim X^{-\eta}$

2677: when any of its arguments $X \to \infty$.  For example, one has,

2678: $V_b \sim |q_y|^{-2\eta} \sim |\omega|^{-2\eta/3}$.

2679: 

2680: As will be detailed shortly, the singular vertices

2681: Eq.~(\ref{Vrho;scaling}) and Eq.~(\ref{Vb;scaling}) will lead to 

2682: anomalous decay exponents for the boson density-density correlator

2683: and the boson Green's function.

2684:  

2685: 

2686: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

2687: \subsection{ Boson and density correlators at finite $N$}

2688: \label{subsec:1_N_corr}

2689: 

2690: Since there are no singular $1/N$ corrections to the fermion 

2691: Green's functions, to extract the boson and density correlators up to 

2692: order $1/N$ one can use the $N=\infty$ fermion Green's functions when 

2693: evaluating the bubble diagram.

2694: Using Eq.~(\ref{N=infinity_Greens}) and the scaling form 

2695: Eq.~(\ref{Vb;scaling}) for the $V_b$ vertex 

2696: enables the particle-particle bubble to be expressed as,

2697: \begin{eqnarray*}

2698: \Pi^{(N)}_{pp} ({\bf q},\omega) \sim |q_x|^{-(2+\eta)}

2699: \int_{{\bf q}^\prime \omega^\prime}

2700: P \left( \frac{q_y^2}{q_x}, \frac{\omega^{2/3}}{q_x},

2701:          \frac{q_x^\prime}{q_x}, 

2702:          \frac{q_y^{\prime 2}}{q_x},\frac{\omega^{\prime 2/3}}{q_x} 

2703: \right) ~,

2704: \end{eqnarray*} 

2705: where $P$ is an appropriate scaling function.

2706: One can then scale out the $q_x$ dependence under the integrals 

2707: to obtain,

2708: \begin{equation}

2709: \Pi^{(N)}_{pp}({\bf q},\omega) \sim |q_x|^{1-\eta} \,

2710: \tilde{\Pi}_{pp}\left(\frac{q_y^2}{q_x}, \frac{\omega^{2/3}}{q_x} \right)

2711: ~.

2712: \end{equation}

2713: We thereby arrive at the desired expression for the boson correlator,

2714: \begin{equation}

2715: G_b^{(N)}({\bf k}_{F_1} - {\bf k}_{F_2} + {\bf q}, \omega) 

2716: \sim |q_x|^{1-\eta} 

2717: \tilde{G} \left( \frac{q_y^2}{q_x}, \frac{\omega^{2/3}}{q_x} \right) ~,

2718: \label{Gb_scaling}

2719: \end{equation}

2720: with exponent $\eta$ given in Eq.~(\ref{eta_1/N}).

2721: While this result was obtained systematically to order $1/N$, 

2722: we expect it will remain valid more generally, and in particular will 

2723: involve the same scaling combinations $q_y^2/q_x$, $\omega^{2/3}/q_x$.

2724: The scaling function $\tilde{G}$ and the dependence of the exponent 

2725: $\eta$ on the bare parameters such as the Fermi surface curvatures will 

2726: presumably vary with $N$.

2727: Near the other momentum ${\bf k}_{F_1} + {\bf k}_{F_2}$, 

2728: the boson correlator will satisfy a similar scaling form except with 

2729: $\eta=0$.

2730: As argued before, we expect that the anomalous exponent at momentum

2731: ${\bf k}_{F_1} +{\bf k}_{F_2}$ will vanish in general,

2732: not just to leading order in $1/N$.

2733: 

2734: The large distance behavior of the equal-time boson Green's function in 

2735: the DBL follows directly from Eq.~(\ref{Gb_scaling}) and was already 

2736: listed in the Introduction, Eq.~(\ref{Gb_final}).

2737: We remark that simple ``power counting'' using Eq.~(\ref{Gb_scaling}) 

2738: works here because the relevant frequencies and momenta are indeed in 

2739: the ``scaling regime''; however, it should be kept in mind that we are 

2740: dealing with line singularities and should exercise more caution with 

2741: such arguments in general.

2742: Once again, both the Fermi wave vectors ${\bf k}_{F_1}, {\bf k}_{F_2}$ 

2743: and the scaling exponent $\eta$ will depend on the location and shape of 

2744: the Fermi surface patches with normals along the observation direction 

2745: $\hat{\bf r}$.   At wave vector ${\bf k}_{F_1} - {\bf k}_{F_2}$, the 

2746: Green's function decay becomes more slow due to the Amperean attraction 

2747: between the $d_1$ and $d_2$ fermions.

2748: The fermions tend to move as pairs, but unlike a Cooper pair their 

2749: motion is not phase coherent -- the pairs are uncondensed.

2750: The boson correlator, while decaying more slowly with positive $\eta$,

2751: does not exhibit ODLRO in the DBL phase.

2752: 

2753: As we found to leading order in $1/N$, Eq.~(\ref{eta_1/N}), $\eta$ is 

2754: expected to be largest when the two Fermi surface curvatures are equal.

2755: In the DBL phase which has closed Fermi surfaces, the characteristics

2756: of the $d_1$ and $d_2$ patches with normals along either diagonal 

2757: $\hat{\bf x} \pm \hat{\bf y}$ are always equal due to the symmetry of 

2758: the square lattice.   Moreover, the decay of the boson Green's function 

2759: $G_b({\bf r})$ will be non-oscillatory along the diagonals,

2760: \begin{equation}

2761: G_b({\bf r}) \sim \frac{1}{|{\bf r}|^{4-\eta}} ~, 

2762: \hskip0.5cm \hat{\bf r} \parallel (\hat{\bf x} \pm \hat{\bf y}) ~,

2763: \end{equation}

2764: a prediction that we find consistent with the properties of the 

2765: $(\det)_x (\det)_y$ wavefunctions studied in Sec.~\ref{sec:Wavefnc_props}

2766: (see Fig.~\ref{fig:real-space-11}).

2767: For $c_1=c_2$, a very rough estimate can be extracted from the leading 

2768: behavior, $\eta = (1/N) + O(e^2/N)$ by setting $N=e^2=1$ giving 

2769: $\eta \approx 1$.

2770: This would shift the $N=\infty$ decay exponent back to its mean field 

2771: value of $1/|{\bf r}|^3$.   

2772: As noted earlier, the gauge fluctuations encapsulated already at 

2773: $N=\infty$ strongly modify the motion of each fermion leading to a 

2774: boson that moves less coherently decaying with a larger exponent than 

2775: in the mean field theory.  The Amperean attraction between the pair of 

2776: fermions which enters at order $1/N$ compensates this effect, 

2777: leading to a slower decay of the boson correlator.  

2778: For $\eta=1$ the two competing effects exactly compensate one another.

2779: 

2780: 

2781: Next we consider the density-density correlator.

2782: Just as for the boson correlator, the density correlator will

2783: satisfy a scaling form:

2784: \begin{equation}

2785: D_{d_\alpha}^{(N)}(2{\bf k}_{F_\alpha} + {\bf q}, \omega) 

2786: \sim |q_x|^{1-\gamma_\alpha} 

2787: \tilde{D}_{2{\bf k}_{F_\alpha}} 

2788: \left( \frac{q_y^2}{q_x}, \frac{\omega^{2/3}}{q_x} \right) ~,

2789: \end{equation}

2790: with an anomalous exponent that depends on the Fermi surface patches,

2791: see Eq.~(\ref{anomalous;density;exponent}).

2792: Back in real space, the dominant density correlator oscillates with 

2793: wave vector $2{\bf k}_{F_\alpha}$, with an envelope decaying as a power 

2794: law with the anomalous exponent,

2795: \begin{equation}

2796: D_{d_\alpha}^{(N)}({\bf r}) \sim 

2797: - \frac{\cos[2{\bf k}_{F_\alpha} \cdot {\bf r} - 3\pi/2]}

2798:        {|{\bf r}|^{4-\gamma_\alpha}}

2799: - \frac{1}{|{\bf r}|^4} ~.

2800: \end{equation}

2801: We have also indicated that the zero momentum component is

2802: not modified relative to the $N=\infty$ behavior.

2803: Again, both ${\bf k}_{F_\alpha}$ and $\gamma_\alpha$ are particular to 

2804: the Fermi surface patches with normals parallel to $\pm \hat{\bf r}$.

2805: In contrast to the particle-particle channel, the particle-hole channel 

2806: is not nested so that we suspect the exponents $\gamma_\alpha$ 

2807: will be smaller than $\eta$ for the physically relevant case

2808: with $N=1$ . 

2809: 

2810: In sum, at leading order in our systematic $1/N$ expansion, the boson 

2811: correlator was found to exhibit power law decay with an exponent that 

2812: depends on the bare Fermi surface curvatures, varying continuously 

2813: around the Fermi surface.

2814: Similar behavior was found for the density-density correlator.

2815: Provided this qualitative behavior persists down to the physically 

2816: relevant case of $N=1$, we conclude that the DBL phase is described 

2817: by a manifold of scale invariant theories rather than an isolated 

2818: fixed point.  Within the $1/N$ expansion, the power law form obtained 

2819: for the boson correlator relied on the exponentiation of the logarithmic 

2820: behavior on momentum and frequency of the leading $1/N$ correction.  

2821: In the next section, we check the legitimacy of this procedure 

2822: by revisiting the renormalization group approach developed earlier

2823: to describe the low energy physics of a sea of fermions coupled

2824: to a $U(1)$ gauge field.  

2825: 

2826: We conclude this section by pointing out that throughout we worked

2827: on the assumption of having both $d_1$ and $d_2$ parallel patches

2828: present as in Fig.~\ref{fig:patches}.  In the case with open Fermi 

2829: surfaces shown in the right panel of Fig.~\ref{fig:introFS}, 

2830: we encounter situations when for a given observation direction 

2831: $\hat{\bf r}$ only one Fermi surface (or even none at all) has patches 

2832: with normals in this direction.

2833: In such cases, the boson correlator will decay exponentially; 

2834: there is no inter-species gauge interactions (one species is simply

2835: absent), but the fermions that are present are still strongly 

2836: affected by the gauge field, and in particular the preceding analysis of 

2837: their density correlations remains the same.

2838: In the DLBL phase, there are no parallel patches on the entire two Fermi 

2839: surfaces, so the $d_1$ and $d_2$ fermions effectively decouple from 

2840: each other.  It is also interesting to remark that

2841: in the limit of extreme D-eccentricity when the Fermi surfaces

2842: are completely flat, we expect that the relevant gauge field

2843: is very strongly damped.  This can be seen, e.g., from the naive

2844: RPA approximation for the Landau damping coefficient $\Gamma$, 

2845: Eq.~(\ref{Gamma_RPA}), in the limit of vanishing curvatures.

2846: In this limit, the energy scale $\Omega$, Eq.~(\ref{Omega}), 

2847: below which fermions become incoherent, goes to zero, 

2848: which leads us to speculate that perhaps in this case the 

2849: $d_1$ and $d_2$ fermions behave as essentially free.

2850: 

2851: 

2852: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

2853: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

2854: \section{Effective field theory for the DBL: Criticality and Stability}

2855: \label{sec:eff_FT}

2856: 

2857: Polchinski\cite{Polchinski} and others\cite{Altshuler, Nayak} 

2858: argued that 2d fermions interacting with a $U(1)$ gauge field can be 

2859: fruitfully studied within a renormalization group analysis of a 

2860: particular effective field theory.  

2861: In the present case, the fixed point theory has an action given by,

2862: \begin{equation}

2863: S_{\rm fixed\, point} = S_d + S_a + S_{int} ~,

2864: \end{equation}

2865: with

2866: \begin{widetext}

2867: \begin{equation}

2868: S_d = \sum_{\alpha s} \int_{{\bf q}\omega} 

2869: d^\dagger_{\alpha s}({\bf q},\omega) 

2870: \left( -i \omega \left|\frac{\Omega_\alpha}{\omega}\right|^{1/3} 

2871:        + s v_\alpha q_x + \frac{v_\alpha c_\alpha}{2} q_y^2 

2872: \right) d_{\alpha s}({\bf q},\omega) ~,

2873: \end{equation}

2874: \end{widetext}

2875: while the gauge field part $S_a$ is given in Eq.~(\ref{Sa}) and

2876: the fermion-gauge coupling is specified in Eq.~(\ref{Lint}).

2877: Within this fixed point ansatz, both the gauge field and the fermions 

2878: have singular propagators, but their interaction is local.

2879: The advantage of this effective field theory is that it is possible to 

2880: define a simple renormalization group transformation which leaves the 

2881: Gaussian part of the theory invariant, and under which the interaction 

2882: strength $g$ is a marginal perturbation.  If $g$ is assumed to be small,

2883: it can be treated via a conventional perturbative RG approach as we 

2884: describe below.  One finds that the full effective field theory is 

2885: invariant under the RG, i.e., it is at a fixed point.  This strongly 

2886: suggests that exponentiating the logarithmic singularities present in the

2887: large $N$ approach, as we did to order $1/N$ above, is a correct 

2888: procedure at all orders in $1/N$.  If it is, then at order $1/N^2$ 

2889: one must find $(\log)^2$ singularities with particular coefficients

2890: such that the exponentiation procedure to obtain a power law behavior 

2891: is consistent.

2892: 

2893: An unsettling drawback with this effective field theory approach is that 

2894: it leaves unclear what constraints must be placed on additional 

2895: interactions that can be added to the theory.

2896: At the very least these interactions must be gauge invariant and 

2897: consistent with momentum conservation, but is it legitimate to require 

2898: that the interactions be local in the fermion fields?  

2899: For example, can one require that four-fermion interactions be local 

2900: with non-singular coefficients, or will singular interactions necessarily

2901: be generated?  Indeed, if one were to formally integrate out the gauge 

2902: field with its singular propagator, one would generate four-fermion 

2903: interactions with a particular singular form.

2904: In what follows we will ignore these subtleties, exploring the possible 

2905: perturbative instabilities driven by non-singular fermion interactions.  

2906: Specifically, we will consider all non-singular quartic interactions 

2907: involving four fermions living near the Fermi surfaces that are 

2908: consistent with the relevant symmetries, most importantly momentum 

2909: conservation.

2910: 

2911: The fixed point action, as it stands, has 9 parameters,

2912: \begin{eqnarray}

2913: \Gamma, \chi, v_\alpha, c_\alpha, \Omega_\alpha, g ~.

2914: \label{all_params}

2915: \end{eqnarray}

2916: But with an appropriate rescaling of the fermion fields and the gauge 

2917: field together with the momenta and frequency, it is possible to set 

2918: 5 of these parameters to unity, 

2919: $\tilde{\Gamma} = \tilde{\chi} = \tilde{g} = \tilde{v}_\alpha = 1$.  

2920: The effective field theory is then specified by 4 dimensionless 

2921: parameters, $\tilde{c}_\alpha = c_\alpha \Gamma / g^2$ and 

2922: $\tilde{\Omega}_\alpha = \Omega_\alpha \chi^2 \Gamma / (g^2 v_\alpha^3)$.

2923: Evidently, this theory is not describing a fixed ``point" per se,

2924: but constitutes a four-dimensional manifold of theories which are 

2925: invariant under the RG.  Establishing definitively that a particular 

2926: bare (lattice) gauge theory Hamiltonian is attracted to this manifold is 

2927: exceedingly difficult.  Arguably, it is even harder to deduce where on 

2928: this manifold the theory flows.  

2929: The values of the dimensionless parameters which are obtained from the 

2930: leading large $N$ analysis can, nevertheless, be used as a rough guide

2931: in addressing both questions.

2932: 

2933: The RG analysis of the fermion-gauge action, $S_{\rm fixed\, point}$, 

2934: proceeds as follows.  At each stage, the fermion fields reside in the 

2935: appropriate momentum space patches 

2936: $|q_x| < \Lambda$, $|q_y| < (\Lambda/c)^{1/2}$,

2937: where $\Lambda$ is the shell width in the direction normal to the Fermi 

2938: surface.  The corresponding restriction on the frequency is

2939: $|\omega| < (v \Lambda)^{3/2} / \Omega^{1/2}$.

2940: The gauge fields reside in similar momentum-frequency regions but 

2941: centered around zero momentum, and the overall setup is illustrated in 

2942: Fig.~\ref{fig:patches}.

2943: If the dimensionless parameters exhibited earlier are of order one

2944: (which is the case if the parameters are taken from the large-$N$ 

2945: analysis), the corresponding regions are roughly similar for all fermion

2946: and gauge fields.  

2947: Also, in practical RG calculations beyond the tree level it is 

2948: convenient to keep the cutoff only on the frequencies and perform 

2949: unrestricted integrations over the momenta.

2950: 

2951: We integrate out the high-energy fields from the shell between 

2952: $\Lambda$ and $\Lambda/b$ and then rescale the momenta and

2953: frequencies in order to recover the initial cutoff:

2954: $q_x = q_x'/b$, $q_y = q_y'/b^{1/2}$, and $\omega = \omega'/b^{3/2}$.

2955: We also perform appropriate rescaling of the fields:

2956: $d = b^2 d'$, $a = b^2 a'$.

2957: Upon such tree-level scaling, the fixed-point action remains 

2958: exactly as before, i.e., all couplings remain unchanged.

2959: It is also useful to bear in mind that even though we nominally

2960: restore all cutoffs as we proceed, in terms of the original momenta 

2961: the patches become more and more elongated in the $\hat{\bf y}$ 

2962: direction in Fig.~\ref{fig:patches}.  

2963: In particular, any overlap between two patches that correspond to nearby 

2964: but non-parallel tangents $\hat{\bf y}$ and $\hat{\bf y}'$ 

2965: goes to zero in the low-energy limit.

2966: 

2967: To proceed with the analysis beyond the tree level, we will work

2968: perturbatively in the (dimensionless) fermion-gauge coupling, 

2969: assuming that it is small.

2970: Quite generally, since the momentum shell RG cannot produce terms that 

2971: are singular at small frequencies and momenta, the couplings 

2972: $\Omega_\alpha$ and $\Gamma$ will not renormalize at any order in the 

2973: perturbation expansion.  On the other hand, the parameters 

2974: $v_\alpha$, $c_\alpha$, $\chi$, and $g$, can potentially flow.

2975: However, a direct examination shows that they do not renormalize at the 

2976: lowest (one-loop) order, and we strongly suspect that this remains true 

2977: at all orders.  The implication is that the fermion-gauge interaction is 

2978: exactly marginal and that $S_{\rm fixed\, point}$ describes a 

2979: manifold of RG fixed points parameterized by 4 dimensionless couplings.

2980: Although the fermion-gauge coupling need not be small, we expect that 

2981: this RG-invariant manifold will extend outside the perturbatively 

2982: accessible regime.

2983: This manifold describes the putative DBL phase.

2984: 

2985: In order to establish the stability of the DBL, however, we need to 

2986: consider the effects of all symmetry allowed perturbations that can be 

2987: added to $S_{\rm fixed\, point}$.

2988: Stability requires that all such perturbations are irrelevant 

2989: under the RG.  Restricting ourselves to local terms, we focus now on the 

2990: four-fermion interactions which are most likely to destabilize the 

2991: fixed point manifold.

2992: 

2993: 

2994: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

2995: \subsection{Short-range fermion interactions}

2996: \label{subsec:4_fermion}

2997: Let us consider a general quartic term,

2998: \begin{equation}

2999: W\; d^\dagger_{\alpha_1 s_1} d^\dagger_{\alpha_2 s_2}

3000:     d_{\alpha_3 s_3} d_{\alpha_4 s_4} ~,

3001: \end{equation}

3002: which can contain different fermion species labelled by $\alpha$ and 

3003: involve different patches labelled by $s$.

3004: The fermion/patch labels are assumed to be such as to satisfy the 

3005: momentum conservation.  Also, we do not show explicitly the (conserved) 

3006: frequency and momentum of the fermion fields.

3007: The amplitude for this four-fermion interaction, $W$, measures the 

3008: strength of a direct scattering between two fermions. 

3009: 

3010: At the tree level in the RG, all such short-range interactions

3011: are irrelevant,

3012: \begin{equation}

3013: W \to W' = W/b ~,

3014: \label{Wtree}

3015: \end{equation}

3016: scaling towards zero at low energies.  Physically, the incoherent motion 

3017: of the fermions, manifest in their $\omega^{2/3}$ self energy, leads to 

3018: a reduced phase space for direct two-body interactions relative to the 

3019: case for free fermions. 

3020: 

3021: 

3022: \begin{figure}

3023: \centerline{\includegraphics[width=2.5in]{figures/4ferm.eps}}

3024: \vskip -2mm

3025: \caption{One-loop diagrams needed to calculate renormalization

3026: of four-fermion interactions.

3027: }

3028: \label{fig:4ferm}

3029: \end{figure}

3030: 

3031: 

3032: At one loop, a gauge propagator can connect any two legs of the quartic 

3033: vertex which involve fermions located on patches with parallel or 

3034: antiparallel Fermi surface normals.

3035: Two such physically distinct diagrams are shown in Fig.~\ref{fig:4ferm}.

3036: Both processes correspond to an interaction between the two fermions

3037: mediated by the Landau damped gauge field.  The one-loop integrals

3038: needed to evaluate these diagrams are identical to those for the 

3039: vertex corrections in subsection~\ref{subsubsec:V1a}, and give the 

3040: following contributions to the four-fermion interaction before any 

3041: rescaling:

3042: \begin{eqnarray}

3043: \text{a):} \quad

3044: \delta W &=& \text{sign}(g_\alpha g_{\alpha^\prime})

3045: \delta_{s, -s^\prime} \lambda_{ph} \, W \, \ln(b)  ~,

3046: \label{Wph}

3047: \\

3048: \text{b):} \quad

3049: \delta W &=& - \text{sign}(g_\alpha g_{\alpha^\prime})

3050: \delta_{s, -s^\prime} \lambda_{pp} \, W \, \ln(b)  ~.

3051: \label{Wpp}

3052: \end{eqnarray}

3053: Here, the primed and unprimed parameters refer to the possibly distinct 

3054: species of the two fermions connected by the gauge propagator.

3055: The anomalous exponents $\lambda_{ph}$ and $\lambda_{pp}$ are given in 

3056: Eqs.~(\ref{lambda_ph})~and~(\ref{lambda_pp})

3057: -- these formulas are general and do not assume any relations among the 

3058: parameters, as appropriate in the present RG setup; 

3059: sometimes, which will be indicated explicitly, we will use the 

3060: $N=\infty$ expressions Eq.~(\ref{lambda_Ninfty}) to get a crude 

3061: estimate of the numerical values.

3062: 

3063: 

3064: Similarly to the analysis in Sec.~\ref{subsubsec:V1a},

3065: either diagram gives zero if the fermions involved have 

3066: parallel group velocities, whereas a nontrivial contribution is found 

3067: when the two fermion fields reside on the opposite patches with 

3068: antiparallel group velocities.

3069: The sign is again determined by Ampere's law:  Two fermions emerging 

3070: from a scattering process with parallel charge currents attract one 

3071: another, spend more time close together, and consequently enhance the 

3072: amplitude for the scattering strength $W$.  Conversely, the interaction 

3073: strength is suppressed when the pair of scattered fermions have 

3074: anti-parallel charge currents.  

3075: The difference in the absolute values of the contributions from the 

3076: diagrams a) and b) is similar to the distinction between the 

3077: particle-hole (CDW) and particle-particle (BCS) processes in a Fermi 

3078: liquid.  Thus, in the situation when $c$ and $c^\prime$ have the same 

3079: sign, the particle-hole process is not ``nested'' because of the 

3080: curvature of the Fermi surface.  On the other hand, the 

3081: particle-particle process is somewhat better nested and becomes 

3082: perfectly nested when $c=c^\prime$, which is the familiar phase space 

3083: reasoning in the theory of the BCS instability.

3084: This distinction between the processes can be clearly seen from the 

3085: arguments $\zeta_{ph}$ and $\zeta_{pp}$, 

3086: Eqs.~(\ref{zeta_ph}) and (\ref{zeta_pp}),

3087: of the function ${\cal E}[\zeta]$ in the two cases.

3088: 

3089: 

3090: With the above general expressions in hand, we now consider specific 

3091: four-fermion interactions.  Of interest is the one-loop contribution

3092: $\Upsilon \propto g^2$ to the eigenvalue of such perturbations at the 

3093: fixed point,

3094: \begin{equation}

3095: \frac{dW}{dl} = (-1 + \Upsilon) W + O(W^2) ~.

3096: \end{equation}

3097: The $O(W^2)$ terms, which in general mix different four-fermion terms,

3098: can be deduced similarly to the RG treatment of quartic interactions 

3099: around the Fermi liquid fixed point\cite{Shankar, Polchinski_FL}

3100: and in fact have very similar structure.  In the Fermi liquid case, 

3101: the interactions are marginal and there is no term linear in $W$, 

3102: so the $O(W^2)$ terms determine the physics.

3103: In the fermion-gauge case, on the other hand, the interactions can

3104: be either relevant when $\Upsilon > 1$ or irrelevant when $\Upsilon < 1$.

3105: For $\Upsilon < 1$, the DBL phase is stable to weak perturbations,

3106: but sufficiently strong bare interactions might still drive the 

3107: DBL through a quantum phase transition into another phase.  

3108: In this case the $O(W^2)$ terms could be helpful in discerning the 

3109: nature of the new phase.

3110: 

3111: As in Fermi liquid theory, the strongest constraint on the allowed

3112: quartic interactions comes from momentum conservation, since all four 

3113: fermions have to live near the Fermi surfaces.

3114: In Fermi liquid theory there are only two allowed types of vertices,

3115: the forward scattering interactions which contribute to the Landau 

3116: parameters and the Cooper vertices.  

3117: We first consider interactions in which all four fermions are of the 

3118: same species.  A general forward scattering interaction then takes the 

3119: form,

3120: \begin{equation}

3121: W^{f}_{{\bf k}, {\bf k}^\prime} 

3122: d^\dagger_{{\bf k}\alpha} d_{{\bf k}\alpha} 

3123: d^\dagger_{{\bf k}^\prime \alpha} d_{{\bf k}^\prime \alpha} ~,

3124: \end{equation}

3125: where ${\bf k}, {\bf k}^\prime$ are two wave vectors on the species 

3126: $\alpha$ Fermi surface.

3127: Consider first the generic case with ${\bf k} \ne \pm {\bf k}^\prime$.

3128: Since this interaction does not involve opposite patches on the 

3129: Fermi surface, the one-loop contribution will vanish, $\Upsilon^f = 0$.

3130: This is also the case when ${\bf k} = {\bf k}^\prime$.

3131: The forward scattering vertex with ${\bf k} = -{\bf k}^\prime$

3132: is part of the Cooper channel to which we now turn.

3133: 

3134: For a single species of fermion, there exists only the odd angular 

3135: momentum Cooper pairing channel (``triplet pairing''),

3136: \begin{equation}

3137: W^t_{{\bf k}, {\bf k}^\prime} 

3138: d^\dagger_{{\bf k} \alpha} d^\dagger_{-{\bf k} \alpha}

3139: d_{-{\bf k}^\prime \alpha} d_{{\bf k}^\prime \alpha} ,

3140: \end{equation}

3141: with 

3142: $W^{t}_{{\bf k}, {\bf k}^\prime} = -W^{t}_{-{\bf k}, {\bf k}^\prime} 

3143: = -W^{t}_{{\bf k}, -{\bf k}^\prime}$.  

3144: The one loop contribution is negative,

3145: $\Upsilon^t_{{\bf k}, {\bf k}^\prime} = 

3146: - (\lambda_{pp}^{\bf k} + \lambda_{pp}^{{\bf k}^\prime})$

3147: for ${\bf k} \ne \pm {\bf k}^\prime$, due to the Amperean repulsion

3148: between the two particles (and between the two holes) with antiparallel 

3149: group velocities.  

3150: For the special case ${\bf k} = \pm {\bf k}^\prime$ one has 

3151: $\Upsilon^t_{{\bf k}, \pm {\bf k}} = 

3152: - 2 (\lambda_{pp}^{\bf k} - \lambda_{ph}^{\bf k})$ 

3153: since there is now also an Amperean attraction between particles 

3154: and holes on opposite patches of the Fermi surface.  

3155: But since we expect $\lambda_{pp}^{\bf k} \ge \lambda_{ph}^{\bf k}$ 

3156: due to the nesting in the particle-particle channel, one has 

3157: $\Upsilon^t \le 0$ in this case as well.

3158: Thus, the one-loop contribution makes this triplet pairing vertex 

3159: more irrelevant.  Physically the repulsive gauge interaction between 

3160: two $d_\alpha$ fermions is unfavorable for their pairing.  

3161: 

3162: 

3163: We now turn to vertices which involve two $d_1$ fermions and two $d_2$ 

3164: fermions.  For simplicity, we first focus on the situation with 

3165: vanishing D-eccentricity, so that the two Fermi surfaces coincide.

3166: Consider first the forward scattering interactions coupling the 

3167: densities of the two fermion species,

3168: \begin{equation}

3169: W^{f;1,2}_{{\bf k}, {\bf k}^\prime} 

3170: d^\dagger_{{\bf k} 1} d_{{\bf k} 1} 

3171: d^\dagger_{{\bf k}^\prime 2} d_{{\bf k}^\prime 2} ~.

3172: \label{forward_scattering}

3173: \end{equation}

3174: The only non-vanishing one loop correction is when

3175: ${\bf k} = -{\bf k}^\prime$, and gives 

3176: $\Upsilon^{f;1,2}_{{\bf k}, -{\bf k}} = 

3177: 2 (\lambda^{\bf k}_{pp} - \lambda^{\bf k}_{ph}) \geq 0$.

3178: If this is greater than one, it would signal a possible instability, 

3179: but in any event we will see that this is smaller than the eigenvalue 

3180: in the conventional BCS ``singlet" pairing channel that we consider next.

3181: 

3182: The singlet BCS interaction is of the form,

3183: \begin{equation}

3184: W^{s}_{{\bf k}, {\bf k}^\prime} 

3185: d^\dagger_{{\bf k} 1} d^\dagger_{-{\bf k} 2}

3186: d_{-{\bf k}^\prime 2} d_{{\bf k}^\prime 1} ~.

3187: \end{equation}

3188: With zero D-eccentricity as assumed here, the BCS interaction is nested.

3189: Moreover, since the two particles (or the two holes) here carry opposite 

3190: gauge charge and therefore have parallel gauge currents, there will be 

3191: an Amperean attraction giving a positive contribution to the Cooper 

3192: vertex: 

3193: $\Upsilon^{s}_{{\bf k}, {\bf k}^\prime} = 

3194: \lambda_{pp}^{\bf k} + \lambda_{pp}^{{\bf k}^\prime}$,

3195: for ${\bf k} \ne \pm {\bf k}^\prime$.

3196: For a circular Fermi surface, symmetry dictates that 

3197: $\lambda_{pp}^{\bf k}$ will be ${\bf k}$-independent,

3198: but this is actually the case generally for such one-loop contribution

3199: in the case with matched $d_1$ and $d_2$ Fermi surfaces since the 

3200: curvatures on the opposing patches are equal.  

3201: Thus, as claimed earlier, the eigenvalue of the singlet BCS pairing 

3202: interaction, $\Upsilon^s = 2\lambda_{pp}$, is larger than the 

3203: corresponding eigenvalue of the forward scattering interaction coupling 

3204: the densities of the two species on opposite sides of the Fermi surface, 

3205: $\Upsilon^s > \Upsilon^{f;1,2}_{{\bf k}, -{\bf k}}$.

3206: For the special case ${\bf k} = - {\bf k}^\prime$, there will be an 

3207: additional positive contribution to the eigenvalue of the singlet BCS 

3208: pairing interaction due to the Amperean attraction between particles and 

3209: holes,

3210: $\Upsilon^s_{{\bf k}, -{\bf k}} =  

3211: 2 (\lambda_{pp}^{\bf k} + \lambda_{ph}^{\bf k})$.  

3212: But being a set of measure zero, this will not contribute to a BCS-type 

3213: pairing instability.

3214: At the present it is not clear what state would be preferred by a 

3215: large interaction of this form and what significance this might have.

3216: 

3217: It is interesting to use the $N=\infty$ parameters to deduce an 

3218: approximate value for the one-loop contribution to $\Upsilon^s$.  

3219: With zero D-eccentricity, the two Fermi surfaces coincide, 

3220: so the curvatures are equal, $c_1=c_2$, and 

3221: $\lambda_{pp}^{\bf k} = {\cal E}[0] = 1$ for all ${\bf k}$.

3222: This gives  $\Upsilon^s = 2$ and implies that the BCS interaction is 

3223: strongly relevant,

3224: \begin{equation}

3225: \frac{d W^s} {d l} = (-1 + \Upsilon^s) W^s = W^s ~,

3226: \end{equation}

3227: within this approximation.  Since we expect some bare short-range 

3228: attraction between the $d_1$ and $d_2$ particles, the runaway flows will 

3229: lead to BCS pairing.  The composite boson 

3230: $b^\dagger = d_1^\dagger d_2^\dagger$ will condense, resulting in a 

3231: superfluid phase.  Thus, based on this analysis, we suspect that the 

3232: putative S-type Bose liquid phase accessed from the gauge theory with 

3233: identical Fermi surfaces for $d_1$ and the $d_2$ will be generically 

3234: unstable towards superfluidity.

3235: 

3236: 

3237: We now consider the situation with non-zero D-eccentricity.

3238: In this case, $d_1$ and $d_2$ have different Fermi surfaces and the 

3239: Cooper vertex will no longer be nested.  This precludes a conventional 

3240: weak coupling BCS instability for the D-wave Bose liquid phase.

3241: The remaining channels that could potentially drive a weak coupling 

3242: instability are the forward scattering interactions given in 

3243: Eq.~(\ref{forward_scattering}) with the locations ${\bf k}$ 

3244: and ${\bf k}^\prime$ on the two Fermi surfaces tuned to have 

3245: anti-parallel patch normals.

3246: 

3247: For bosons on the square lattice, with increasing D-eccentricity the two 

3248: closed Fermi surfaces will eventually open up, and above some critical 

3249: eccentricity the requirement of anti-parallel $d_1$ and $d_2$ Fermi 

3250: surface normals will be no longer possible.

3251: In this large D-eccentricity regime, when the two Fermi surfaces are so

3252: dissimilar, there are no such forward scattering interactions coupling

3253: the two species which can be enhanced by the gauge fluctuations.

3254: Sometimes one can find interactions that can be enhanced via the 

3255: particle-hole attraction within the same species, e.g., 

3256: $d_{{\bf k} 1}^\dagger d_{{\bf -k} 1} 

3257:  d_{{\bf k}^\prime 2}^\dagger d_{{\bf k}^{\prime\prime} 2}$,

3258: which requires tuning both ${\bf k}^\prime$ and ${\bf k}^{\prime\prime}$

3259: to satisfy momentum conservation but this is not always possible.

3260: In any event, the corresponding one-loop $\Upsilon = \lambda_{ph}^{\bf k}$ 

3261: is not likely to change the irrelevance of this term from the tree level;

3262: e.g. the $N=\infty$ estimate of $\lambda_{ph}^{\bf k}$, 

3263: Eq.~(\ref{lambda_Ninfty}), is always smaller than 1.

3264: Regarding other interactions, none of the intra-species four fermion 

3265: terms have positive eigenvalues.  

3266: We conclude that in this large D-eccentricity regime the D-wave Local 

3267: Bose liquid (as specified in Sec.~\ref{subsec:openFS}) will exist 

3268: as a stable phase.

3269: 

3270: Returning to the case with smaller D-eccentricity, the stability of the 

3271: DBL phase will depend on the eigenvalues $\Upsilon^f(\hat{\bf n})$ of 

3272: the forward scattering interaction that couples a patch on one Fermi 

3273: surface with normal $\hat{\bf n}$ to a patch on the other Fermi surface 

3274: with anti-parallel normal.  The particles on opposing patches 

3275: experience a strong Amperean attraction, whereas a particle and hole 

3276: experience a weaker repulsion, so that the eigenvalues 

3277: $\Upsilon^f(\hat{\bf n}) = 2(\lambda_{pp} -\lambda_{ph})$ 

3278: will be positive for all orientations of $\hat{\bf n}$.  

3279: However, the magnitude of $\Upsilon^f(\hat{\bf n})$ depends on the 

3280: Fermi surface curvatures and will vary with orientation.

3281: This can be made explicit by evaluating the one-loop contribution using 

3282: the $N=\infty$ values of the various parameters, thereby we obtain an 

3283: approximate expression for the eigenvalue,

3284: \begin{eqnarray*}

3285: \Upsilon^f(\hat{\bf n}) \approx 

3286: 2{\cal E} \left[ \frac{\sqrt{3}\pi}{2 g^2} \Gamma |c_1 - c_2| \right]

3287: - 2{\cal E} \left[ \frac{\sqrt{3}\pi}{2 g^2} \Gamma (c_1 + c_2) \right]

3288: ~.

3289: \end{eqnarray*}

3290: The first contribution is maximal and equal to $2$ when the curvatures 

3291: on the two Fermi surfaces coincide, which they will when $\hat{\bf n}$ 

3292: is along a diagonal of the square lattice.

3293: We have no such estimate of the second contribution since 

3294: $g, \Gamma, c_\alpha$ are independent in our $N=\infty$ theory; 

3295: if we use in addition a crude RPA approximation Eq.~(\ref{Gamma_RPA}), 

3296: we get the total

3297: $\Upsilon^f = 2 - 2{\cal E}[\sqrt{3}] = 1.53 > 1$.

3298: In this case, the D-wave Bose liquid in this regime of D-eccentricity 

3299: will presumably be unstable, driven by the forward scattering 

3300: interactions with $\hat{\bf n} = \hat{\bf x} \pm \hat{\bf y}$ 

3301: along the diagonals of the square lattice.

3302: 

3303: 

3304: The above numerical estimate is very uncontrolled and is given only 

3305: to see possible trends.  Still, let us speculate what the resulting

3306: phase might if there is indeed such instability.

3307: Most naively, there are two guesses, obtained by a mean field 

3308: decoupling of the four fermion interaction

3309: $d^\dagger_{{\bf k} 1} d_{{\bf k} 1} 

3310:  d^\dagger_{{\bf k}^\prime 2} d_{{\bf k}^\prime 2}$,

3311: where ${\bf k}$ and ${\bf k}^\prime$ are the locations of the two

3312: patches with normals $\hat{\bf n}$ and $-\hat{\bf n}$ that

3313: give the largest $\Upsilon^f$.

3314: We can assume a non-vanishing particle-particle or particle-hole 

3315: condensate, 

3316: $\la d^\dagger_{{\bf k} 1} d^\dagger_{{\bf k}^\prime 2} \ra $ or

3317: $\la d^\dagger_{{\bf k} 1} d_{{\bf k}^\prime 2} \ra$, respectively.

3318: Both condensates carry non-zero momentum,

3319: ${\bf k} + {\bf k}^\prime$ for the particle-particle condensate, and 

3320: ${\bf k} - {\bf k}^\prime$ for the particle-hole.  

3321: But since the interaction is repulsive in the particle-hole channel,

3322: such a condensate seems rather unlikely.

3323: Also, the combination $d_1^\dagger d_2$ carries a non-zero gauge charge, 

3324: so is unphysical and cannot serve as a legitimate order parameter

3325: [but the product 

3326: $\la d^\dagger_{{\bf k} 1} d_{{\bf k}^\prime 2} 

3327:      d^\dagger_{-{\bf k}^\prime 2} d_{-{\bf k} 1} \ra$ 

3328: is gauge invariant, and if condensed would correspond to

3329: an energy density wave at momentum $2 ({\bf k} - {\bf k}^\prime)$].

3330: 

3331: If there is an instability, we think it is more likely to occur in the 

3332: particle-particle channel.  The order parameter 

3333: $\la d^\dagger_{{\bf k} 1} d^\dagger_{{\bf k}^\prime 2} \ra $ 

3334: is gauge invariant and correspond physically to a finite momentum 

3335: Bose condensate.  

3336: The situation is analogous to the FFLO problem, and as there, provided 

3337: the condensate is not too strong, it will only gap out parts of the 

3338: Fermi surfaces.  As such, this would still be a very unusual type of 

3339: superfluid, which in a mean field description would have residual 

3340: gapless fermionic excitations.  Beyond the mean field, the gauge 

3341: fluctuations would still scatter the gapless fermions, rendering them 

3342: incoherent.  This state would thus correspond to finite momentum 

3343: Bose condensation co-existing with the DBL.  

3344: 

3345: 

3346: In sum, whether or not the DBL for small D-eccentricity is unstable 

3347: to such finite momentum pairing is a quantitative issue which will 

3348: depend on the values of the anomalous dimensions.  

3349: An analysis of the actual wavefunctions in Sec.~\ref{sec:Wavefnc_props}

3350: reveals no such tendencies, so might be taken as an indication against 

3351: the instability.  

3352: Indeed, as we saw in Sec.~\ref{sec:Wavefnc_props}, the S-wave Bose 

3353: liquid wavefunction does appear generically unstable towards a 

3354: conventional superfluid, consistent with the expectation of a 

3355: zero momentum BCS instability in the gauge theory.   

3356: In the same spirit, we will see in the next section that for hard-core 

3357: bosons at half-filling the DBL wavefunction in the extreme 

3358: D-eccentricity limit (corresponding to fermions which can only hop in 

3359: one of the two directions on the square lattice) 

3360: reveals a co-existence of a commensurate $(\pi, \pi)$ CDW with a gapless 

3361: DBL Fermi surface.

3362: This indicates that the Gutzwiller wavefunctions for the DBL, 

3363: at least in some special instances, can also reveal translational 

3364: symmetry breaking instabilities,

3365: while we repeat that no such instabilities are observed for generic DBL

3366: wavefunctions.

3367: 

3368: This concludes our discussion of the effective field theory description

3369: and possible instabilities of the DBL phase.

3370: As we have seen, the main potential instability involves pairing 

3371: $d_1$ and $d_2$ fermions moving with opposite group velocities,

3372: but such BCS channel is suppressed for mismatched Fermi surfaces

3373: and is completely eliminated in the DLBL regime, which appears to

3374: be particularly stable.

3375: 

3376: 

3377: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

3378: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

3379: \section{Ring Hamiltonian Energetics for the DBL phase}

3380: \label{sec:Energetics}

3381: 

3382: In this section we ask what energetics may stabilize the DBL or DLBL

3383: phases.  Specifically, we motivate and study the $J - K_4$ Hamiltonian 

3384: Eq.~(\ref{Hring}) with competing boson hopping and ring exchange terms.

3385: When $J>0$ and $K_4>0$, this Hamiltonian does not satisfy the Marshall 

3386: sign conditions, so one expects the ground state wavefunction to take 

3387: both positive and negative values.

3388: 

3389: The Hamiltonian~(\ref{Hring}) is motivated by considering the 

3390: gauge theory description Eq.~(\ref{HU1}) of the DBL phase in the 

3391: strong coupling limit of the gauge theory, 

3392: $h \gg K, t_\parallel, t_\perp$.  

3393: In this limit, one can perturbatively eliminate the gauge field;

3394: the resulting boson Hamiltonian contains, among other terms, both $H_J$ 

3395: and $H_4$ with 

3396: $J = A\, t_\parallel t_\perp / h$ and

3397: $K_4 = A'\, K t_\parallel^4 / h^4 

3398: - A''\, t_\parallel^2 t_\perp^2 / h^3$ 

3399: with positive numerical coefficients $A, A', A''$.

3400: Only the signs of the contributions to $J$ and $K_4$ are of 

3401: interest here, since in what follows both $J$ and $K_4$ are taken as 

3402: free parameters of the Hamiltonian Eq.~(\ref{Hring}).

3403: Coming from the gauge theory, we have $J>0$ and $K_4>0$ in the regime of 

3404: primary interest when $K \sim h$ and $t_\parallel \gg t_\perp$.

3405: To define the system, we also need to specify the boson density per site 

3406: $\rho$.  The hard-core boson model has particle-hole symmetry,

3407: so it is enough to consider $\rho \leq 1/2$.

3408: 

3409: Before proceeding, we note that the above Hamiltonian with $J>0$ and 

3410: $K_4<0$ was introduced and analyzed in Ref.~\onlinecite{Paramekanti}.

3411: Since there is no sign problem in this case, extensive quantum Monte 

3412: Carlo studies of Refs.~\onlinecite{Sandvik, Melko, Rousseau} were

3413: able to map out the phase diagram of the model.

3414: In addition to a bosonic superfluid phase, the numerical studies 

3415: found two ordered phases at half-filling, a bond-ordered stripe phase 

3416: and a staggered charge density wave (CDW).  When the Monte Carlo was 

3417: performed at fixed chemical potential,\cite{Melko} varying the chemical 

3418: potential drove first-order transitions out of the commensurate solid 

3419: phases into the superfluid.

3420: On the other hand, when the Monte Carlo was performed at fixed 

3421: density,\cite{Rousseau} the superfluid phase was observed as soon as the 

3422: density dropped below half-filling, while a phase separation occurred

3423: when the density was decreased further.

3424: 

3425: The case $J>0$ and $K_4>0$ was not studied in Monte Carlo since there is 

3426: a sign problem.  Below, we present a rudimentary variational energetics 

3427: study of this regime.  

3428: The results are summarized in Fig.~\ref{fig:tKphased} and lend some 

3429: support that the DLBL phase of Sec.~\ref{subsec:openFS} may be 

3430: stabilized by such competing hopping and ring exchange terms.

3431: 

3432: 

3433: \subsection{Model with ring exchanges only}

3434: Consider first the Hamiltonian with the ring term only.

3435: In this case, we can change the sign of $K_4$ by dividing the 

3436: square lattice into four sublattices and performing a transformation 

3437: $b \to -b$ on one sublattice.  

3438: The cited quantum Monte Carlo studies did not consider the pure ring 

3439: model, but extrapolating these to $J=0$, it seems likely that at 

3440: half-filling the ground state is the $(\pi, \pi)$~CDW.

3441: However, we also expect that there is a regime away from half-filling

3442: that realizes a novel ``Excitonic Bose Liquid'' (EBL) phase

3443: discovered and studied by Paramekanti\etal,\cite{Paramekanti}

3444: which we now discuss.

3445: 

3446: The EBL was accessed by

3447: considering a rotor version of the ring Hamiltonian,

3448: \begin{eqnarray*}

3449: H_{\rm rotor} &=& \frac{\cal U}{2} \sum_{\bf r} (n_{\bf r} - \bar{n})^2

3450: \\

3451: &-& |{\cal K}|

3452: \sum_{\bf r} \cos(\phi_{\bf r}

3453:                   - \phi_{{\bf r} + \hat{\bf x}}

3454:                   + \phi_{{\bf r} + \hat{\bf x} + \hat{\bf y}} 

3455:                   - \phi_{{\bf r} + \hat{\bf y}}) ~,

3456: \end{eqnarray*}

3457: where the phase $\phi_{\bf r}$ and the boson density $n_{\bf r}$ are 

3458: canonically conjugate.

3459: When the cosine in the ring term is expanded, i.e., in the phase with 

3460: no topological defects, one finds a quasi--one-dimensional dispersion 

3461: relation for the excitons (density waves),

3462: $\omega_{\bf k}^2 \sim \sin^2(k_x/2) \sin^2(k_y/2)$.

3463: This gapless normal Bose state is expected to be stable 

3464: against becoming an insulator when $|{\cal K}|$ dominates over 

3465: ${\cal U}$.

3466: 

3467: The ring only model on the square lattice has special conservation laws: 

3468: particle numbers on each column and each row are individually conserved.

3469: It is this property that is responsible for the vanishing of 

3470: $\omega_{\bf k}$ along the lines $k_x=0$ and $k_y=0$ in the EBL phase.  

3471: Paramekanti\etal\cite{Paramekanti} showed that due to the gapless lines,

3472: EBL is a critical (power-law) 2d quantum phase with continuously

3473: varying exponents.

3474: 

3475: For our energetics study, it is useful to have a good trial wavefunction 

3476: for the EBL phase of hard-core bosons.  Motivated by the described 

3477: ``spin wave'' theory, we write the following wavefunction

3478: \begin{equation}

3479: \Psi_{\rm EBL}^{(K_4 < 0)}({\bf r}_1, {\bf r}_2, \dots)

3480: \;\;\propto\;\; \exp[-\sum_{i<j} u_{\rm EBL}({\bf r}_i - {\bf r}_j)] ~,

3481: \end{equation}

3482: with

3483: \begin{equation}

3484: u_{\rm EBL}({\bf r}-{\bf r}') = 

3485: \frac{1}{V} \sum_{\bf k} \frac{\cal W}{4 |\sin(k_x/2)| |\sin(k_y/2)|} 

3486: e^{i {\bf k} \cdot ({\bf r} - {\bf r}')} ~. 

3487: \end{equation}

3488: In the spin wave theory, we have ${\cal W} = \sqrt{\cal U/|K|}$, 

3489: but more generally ${\cal W}$ is treated as a variational parameter.

3490: The summation over ${\bf k}$ excludes the lines where either

3491: $k_x=0$ or $k_y = 0$.

3492: Such wavefunction can be defined in any sector with fixed number

3493: of particles on each row and column as is appropriate in the study

3494: of the ring-only model.

3495: We find that for moderate ${\cal W}$ and away from half-filling,

3496: this wavefunction has a density structure factor of the form,

3497: $\la \hat\rho(-{\bf k}) \hat\rho({\bf k}) \ra \sim |k_x| |k_y|$

3498: at long wavelengths as expected in the EBL.

3499: Furthermore, the EBL theory predicts that the box correlator defined in 

3500: Eq.~(\ref{Bb_def}) decays as a power law with a continuously varying

3501: exponent, and this is what we observe for the wavefunction.

3502: In the energetics study, the parameter ${\cal W}$ is varied to minimize

3503: the ring energy $H_4$.

3504: 

3505: Returning to the model with $K_4 > 0$, the trial wavefunction is

3506: \begin{equation}

3507: \Psi_{\rm EBL}^{(K_4 > 0)}({\bf r}_1, {\bf r}_2, \dots)

3508: \propto (-1)^{N_{\rm I}} 

3509: \exp[-\sum_{i<j} u_{\rm EBL}({\bf r}_i - {\bf r}_j)] ~,

3510: \label{Psi_EBL}

3511: \end{equation}

3512: where $N_{\rm I}$ is the number of bosons on the sublattice ${\rm I}$ of 

3513: the four sublattices.  One can readily translate the preceding results

3514: to this wavefunction, which has the sign structure appropriate 

3515: for the pure ring exchange model with $K_4 > 0$.

3516: 

3517: 

3518: Interestingly, the DLBL wavefunction in the extreme D-eccentricity

3519: limit, which was introduced towards the end of Sec.~\ref{subsec:openFS}

3520: and discussed further in Sec.~\ref{subsubsec:xtrmDLBL},

3521: also has the correct sign structure for the ring Hamiltonian.

3522: This follows from an observation that this wavefunction takes 

3523: opposite signs for any two boson configurations that are connected by 

3524: $H_4$.  It is important here that $H_4$ contains only unit square 

3525: plackets, and one subtlety is that such unit ring moves are not ergodic 

3526: in the sectors mentioned earlier with fixed boson numbers on each row 

3527: and column.  The signs of the EBL and extremal DLBL wavefunctions 

3528: as written will not agree everywhere.

3529: However, in the study of the ring model, it is natural to consider 

3530: smaller sectors comprising states that are connected by repeated 

3531: applications of $H_4$, and in such sectors the signs of the two 

3532: wavefunctions agree.  

3533: Thus, one needs to carefully specify what sector restriction is being 

3534: made when working with these wavefunctions.  As far as the $K_4$ 

3535: energetics is concerned, we find essentially no difference between the 

3536: wavefunctions obtained by restricting to the smaller (ergodic) sectors 

3537: or the larger sectors which fix only the boson numbers on each row and 

3538: column.

3539: The presented results are all for the latter choice of sectors;

3540: the Monte Carlo sampling is performed using arbitrary rectangular 

3541: ring moves, which are ergodic in this sector.

3542: 

3543: 

3544: Figure~\ref{fig:etrial_ring} shows the ring exchange energy per site 

3545: for the optimized EBL and extremal DLBL wavefunctions plotted vs boson 

3546: density.  The two trial energies are fairly close; the EBL state has 

3547: somewhat lower energy for $\rho \gtrsim 1/4$, and the difference is 

3548: largest near $\rho = 1/2$.

3549: Note that the extremal DLBL wavefunction has no variational 

3550: parameters, while the EBL state has one parameter.

3551: The trial DLBL energy can be somewhat improved by maintaining the 

3552: same sign structure but taking a variable power of the determinants; 

3553: such trial energies (which we do not show) approach closer to the 

3554: optimal EBL energies.

3555: Figure~\ref{fig:etrial_ring} suggests that the DLBL wavefunction 

3556: not only has the right sign structure, but also has similar short range 

3557: correlations to the EBL state.

3558: Of course, additional Jastrow-type factors can be used to further

3559: improve the DLBL energy, but here and below we emphasize the good sign 

3560: structure and correlations that are already present in the 

3561: determinantal construction itself.

3562: 

3563: 

3564: 

3565: \begin{figure}

3566: \centerline{\includegraphics[width=\columnwidth]{figures/xtrmDBL_EBL.eps}}

3567: \vskip -2mm

3568: \caption{

3569: Trial ring energy in the optimized EBL wavefunction (filled circles)

3570: and extremal DLBL wavefunction (open circles).

3571: Measurements are performed on a $24 \times 24$ system;

3572: the boson number is always a multiple of 24 since the extremal DLBL 

3573: requires that the number of particles is exactly the same in each 

3574: row/column.

3575: Away from half-filling the optimized EBL and DLBL wavefunctions

3576: produce uniform liquids, but exactly at half-filling we find a long-range

3577: $(\pi,\pi)$ CDW order in both states, which is emphasized with 

3578: large box symbols.

3579: The thin lines are guides to eye, while the thick lines are the result 

3580: of the Maxwell construction for the two trial energies that addresses

3581: the possibility of the phase separation at low densities.  

3582: From this construction, the uniform liquids appear to be stable in the 

3583: the density windows  $0.42 < \rho < 0.50$.

3584: }

3585: \label{fig:etrial_ring}

3586: \end{figure}

3587: 

3588: 

3589: The similarity between the EBL and extremal DLBL wavefunctions is also 

3590: revealed in their density structure factors, 

3591: see also Sec.~\ref{subsubsec:xtrmDLBL}.

3592: In both states away from half-filling, for small $|k_x| \ll |k_y|$, 

3593: we can write $D_b(k_x, k_y) = A(k_y) |k_x|$.  

3594: In the DLBL state, $A(k_y)$ appears to be independent of $k_y$.  

3595: On the other hand, in the EBL state, $A(k_y) \sim |k_y|$.  

3596: Thus, the character of the $k_x \to 0$ singularity is the same in both 

3597: states for any non-zero $k_y$, and it is only in the limit 

3598: $k_x, k_y \to 0$ that the two differ.

3599: Interestingly, both wavefunctions also exhibit ``$2 k_F$'' singularities

3600: in the density structure factor.

3601: Such singularities are not apparent in the naive continuum ``spin wave'' 

3602: EBL theory, but manifest themselves in the lattice hard-core boson 

3603: EBL wavefunction.

3604: 

3605: The study of density correlations also reveals that these optimized 

3606: states at half-filling in fact have spontaneous $(\pi, \pi)$ 

3607: charge order, while away from half-filling no order is observed.

3608: Thus, even though in the present crude energetics work we do not consider

3609: directly conventional ordered states of bosons such as charge or bond 

3610: density waves, the EBL and DLBL wavefunctions provide some access to 

3611: the energetics of such ordered states.  

3612: (Note, however, that these trial states likely have some degree of 

3613: EBL-ness, e.g., finite compressibility; this is analogous to the 

3614: well-known observation that when a conventional Jastrow wavefunction 

3615: is driven into a regime of charge order, it in fact represents a 

3616: supersolid that also supports off-diagonal long range order.)

3617: As we have already mentioned, Monte Carlo studies\cite{Melko, Rousseau}

3618: suggest staggered CDW at $\rho = 1/2$, and our finding of

3619: the same tendency in the considered wavefunctions lends some support

3620: to the goodness of the variational approach.

3621: Conventional boson orders are potentially important near other 

3622: commensurate boson densities but are less important at generic 

3623: $\rho$, and we will not consider charge ordered states further.

3624: 

3625: Leaving the charge orders aside, an important feature is prominent 

3626: in the energy plots in Fig.~\ref{fig:etrial_ring} for the considered 

3627: uniform liquids.  

3628: We see regions where $\epsilon(\rho)$ is concave, which signals an 

3629: instability towards phase separation.  Using the Maxwell construction 

3630: for the available data, we conclude that the system phase separates

3631: for boson densities $\rho \lesssim 0.42$.  

3632: At low densities, the system will spontaneously separate into an empty 

3633: phase and the EBL phase at density $\rho = 0.42$.

3634: In the region near $\rho = 1/2$, the considered uniform liquids

3635: appear to be stable, except perhaps very close to half-filling.

3636: 

3637: The tendency in the ring model towards phase separation at low densities

3638: was noted by Rousseau\etal\cite{Rousseau} 

3639: Indeed, if we consider two bosons on an otherwise empty lattice, 

3640: the lowest ring energy is achieved when the two reside on the same 

3641: square placket, since only in this case the ring term is nonzero.

3642: Thus, two bosons can be bound together by the ring energy.

3643: By performing exact diagonalization of small clusters up to $4 \times 4$ 

3644: with open boundary conditions and for different boson numbers, 

3645: we find that the energy per particle is the lowest when the density is 

3646: near half-filling; also, larger open clusters achieve lower energies per 

3647: boson.  

3648: One can construct eigenstates of the ring Hamiltonian at low densities 

3649: composed of such isolated clusters, and the discussed tendencies suggest 

3650: that not only two but many bosons tend to clump together in the 

3651: presence of the ring exchanges.

3652: 

3653: One more notable feature in the energy plots occurs near half-filling.  

3654: We show only the $\rho \leq 1/2$ part, while the $\rho \geq 1/2$ part 

3655: would be the mirror image due to particle-hole symmetry.

3656: Therefore, $\epsilon(\rho)$ shows a cusp precisely at half-filling --

3657: at least in the EBL case and for the system sizes studied.

3658: Since $\mu = d\epsilon/d\rho$, the cusp may be interpreted as an 

3659: evidence for a charge gap at this density, which appears to be in line 

3660: with the proposal of the CDW order by the Monte Carlo studies.

3661: 

3662: We do not know whether the above observations persist in the limit

3663: of large systems.  Thus, it may also happen that there will be

3664: no regime of the EBL phase in the specific model if the system chooses to

3665: phase separate into empty space and a half-filled charge insulator.

3666: If there is a stable EBL regime over some density window near 

3667: half-filling, then the phase separation at lower densities is likely to 

3668: be into empty space and an EBL phase of an appropriate density 

3669: determined from the Maxwell construction.

3670: The existence of an EBL phase near half-filling could presumably be 

3671: checked by careful Monte Carlo simulations for the pure ring model.

3672: It is also plausible that the regime of the stable uniform EBL phase 

3673: could be extended by adding interactions that disfavor phase separation, 

3674: but we have not explored this in the present work.

3675: 

3676: 

3677: \subsection{Full $J$-$K_4$ Hamiltonian}

3678: We now turn to the case with non-zero boson hopping $J > 0$.

3679: Consider first small $J$, and suppose we start in the uniform

3680: EBL or DLBL regime assuming such exists.  As we have discussed, 

3681: the hopping and ring terms frustrate one another.  

3682: At present, we do not know how to include the effects of

3683: hopping if we begin with the EBL.

3684: Thus, the EBL wavefunction is defined in sectors with fixed 

3685: boson numbers on each row/column, but what relative signs should 

3686: we take for the different sectors that are now mixed by the boson

3687: hopping?  If, for example, we naively extend Eq.~(\ref{Psi_EBL}) to all 

3688: boson configurations, with perhaps modified Jastrow factors, the

3689: expectation value of $H_J$ remains precisely zero.

3690: It seems that one needs to change the sign structure of the wavefunction

3691: in order to treat the frustration accurately.

3692:  

3693: On the other hand, the DLBL wavefunction provides a natural 

3694: continuation from the extremal case of the ring only model, 

3695: and we find that it is indeed capable of utilizing both the ring energy 

3696: and the hopping energy.  More precisely, for non-zero $J$ the optimized 

3697: D-eccentricity is finite and varies as $J$ is increased,

3698: moving towards zero D-eccentricity at large $J$.

3699: Within our restricted variational energetics study this is the strongest 

3700: evidence we can offer that the ground state of the ring Hamiltonian in 

3701: this regime is in the DLBL phase.

3702: 

3703: 

3704: \begin{figure}

3705: \centerline{\includegraphics[width=\columnwidth]{figures/tKphased.eps}}

3706: \vskip -2mm

3707: \caption{

3708: Schematic ``variational phase diagram'' of the $J - K_4$ model.

3709: Numerics is done for the same sizes as in Fig.~\ref{fig:etrial_ring}.

3710: The phase boundaries are obtained by examining the DBL state, allowing 

3711: for the possibility of phase separation, and the superfluid state

3712: (the latter appears to be always uniform in our study).

3713: The optimal values of $t_\perp / t_\parallel$ throughout the whole 

3714: DBL phase are such that the fermions have open Fermi surfaces,

3715: i.e., this is the DLBL phase of Sec.~\ref{subsec:openFS}.

3716: }

3717: \label{fig:tKphased}

3718: \end{figure}

3719: 

3720: 

3721: Some details of the variational study are summarized in the 

3722: ``phase diagram'' in Fig.~\ref{fig:tKphased}.

3723: At a given density, the trial energy of the DBL state is optimized

3724: using the ratio $t_\perp / t_\parallel$ as a variational parameter.  

3725: Similarly to the pure ring model, we also analyze the possibility of 

3726: phase separation, which is done in the spirit of 

3727: Fig.~\ref{fig:etrial_ring}.

3728: We find a stable uniform DBL phase in the density range roughly 

3729: similar to the ring only model.  

3730: Physically, we expect the phase separation to be suppressed when the

3731: hopping becomes non-zero, but our study is too limited to explore this 

3732: systematically.

3733: 

3734: The phase diagram Fig.~\ref{fig:tKphased} also includes a conventional 

3735: superfluid state of bosons, which is expected in the regime of large $J$.

3736: We take the Jastrow-type form for the trial wavefunction,

3737: Eq.~(\ref{Psi_Jastrow}), and use the pseudopotential 

3738: $u({\bf r} - {\bf r}') = W/|{\bf r} - {\bf r}'|^p$ 

3739: with two variational parameters $W$ and $p$.

3740: Allowing more parameters such as an independent nearest-neighbor 

3741: pseudopotential does not visibly change the optimized energies.

3742: Interestingly, as deduced from the energetics, the superfluid-DBL phase 

3743: boundary roughly coincides with the limit of vanishing D-eccentricity

3744: $t_\perp = t_\parallel$.

3745: The resulting $(\det)^2$ wavefunction is positive, and, as we discussed

3746: at length in Sec.~\ref{sec:Wavefnc_props}, this state has off-diagonal 

3747: long range order.  

3748: Of course, it does not have the same long-wavelength correlations 

3749: as the superfluid (e.g., it has unusual $2k_F$ density correlations),

3750: but it appears to have reasonable short-range correlations.

3751: 

3752: Finally, we also remark that in the present $J - K_4$ study, the 

3753: optimal DBL phase when it wins over the superfluid is in fact the DLBL 

3754: phase with open Fermi surfaces discussed in Sec.~\ref{subsec:openFS} 

3755: and is in the regime where the boson Green's function decays 

3756: exponentially in all directions.

3757: As we discussed in Secs.~\ref{sec:MFT} and \ref{sec:Wavefnc_props},

3758: despite such ``locality'' in the boson correlator, this unusual phase 

3759: is still critical with many low energy excitations.

3760: 

3761: Summarizing, the DBL wavefunction is able to interpolate 

3762: between the EBL and the superfluid regimes in the frustrated 

3763: $J - K_4$ model.  This is the main argument supporting our proposal 

3764: that the ground state of the ring Hamiltonian in the intermediate regime 

3765: is in the DLBL phase.

3766: 

3767: 

3768: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

3769: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

3770: \section{Summary and Discussion}

3771: \label{sec:concl}

3772: 

3773: Our main results were summarized in the Introduction,

3774: so here we only highlight the most interesting points and discuss

3775: possible future directions.

3776: We are searching for examples of uncondensed quantum boson liquids that respect 

3777: time reversal and occur at generic continuously varying densities.

3778: In this paper, we proposed several wavefunctions that produce such

3779: liquids and studied in detail the specific states dubbed DBL and DLBL.

3780: While the initial wavefunction construction appears somewhat ad-hoc,

3781: we argued that it is no more so than the slave particle construction

3782: of spin liquid states in frustrated antiferromagnets, which has been

3783: developed to maturity over the last two decades.  

3784: We introduced a particular slave fermion treatment of hard core bosons

3785: on a square lattice that corresponds to the proposed DBL wavefunctions.

3786: This approach allows to go beyond the trial ground state wavefunction,

3787: suggesting natural excitations, and eventually leads to a gauge theory 

3788: description of the low-energy properties of the putative boson liquid 

3789: phase.

3790: Using techniques previously developed for the so-called uniform RVB

3791: spin liquid with a Fermi sea of spinons,

3792: we argued for the possible stability of the new DBL phases to gauge 

3793: fluctuations and found the DLBL regime to be particularly stable.

3794: Within the gauge theory formulation, we also analyzed boson and boson

3795: density correlators at long distances, which reveal special singular 

3796: surfaces in the momentum space.

3797: 

3798: The singular surfaces were confirmed by measuring the properties of

3799: the proposed DBL/DLBL wavefunctions directly; in the same numerics,

3800: no potential orders were observed, indicating stability of the states.

3801: We remark here that the Gutzwiller wavefunctions and the gauge theory 

3802: should not be viewed as identical descriptions, and, in particular,

3803: the detailed long-distance properties may be different.

3804: The Gutzwiller wavefunctions can be used for crude energetics

3805: and qualitative characterization, but in our opinion the gauge theory 

3806: is more complete as far as the actual quantum phase is concerned.

3807: Indeed, the wavefunctions as constructed do not incorporate any

3808: dynamics of the gauge fluxes, which are clearly additional low-energy 

3809: ``variational'' degrees of freedoms that the quantum system is going to 

3810: explore and utilize.

3811: 

3812: From the singular momentum surfaces in the boson properties, 

3813: one can in principle recover the underlying Fermi surfaces of the 

3814: slave particle construction; the latter have correct volumes for 

3815: the boson density, offering a tantalizing possibility of some 

3816: ``Luttinger theorem'' for uncondensed liquids of interacting

3817: bosons.

3818: 

3819: Some striking thermodynamic properties of the proposed boson liquids

3820: follow from their fermion - gauge character.\cite{Holstein, Reizer, PALee, IoffeKotliar, LeeNagaosa, Polchinski, Altshuler, Nayak, YBKim, LeeNagaosaWen, Senthil} 

3821: In both the DBL and DLBL phases, because of the gapless Fermi surfaces

3822: and the gauge interactions, the specific heat at low temperatures is 

3823: expected to behave as $C \sim T^{2/3}$, while the resistivity is 

3824: expected to vanish as $R_b \sim T^{4/3}$.

3825: The last result in the DBL phase with closed Fermi surfaces is simply a 

3826: quote from earlier studies of such Fermi sea - gauge 

3827: systems,\cite{IoffeKotliar, LeeNagaosa}

3828: while the case with open Fermi surfaces requires slightly more care.  

3829: Our argument is based on the application of the Ioffe-Larkin 

3830: rule\cite{IoffeLarkin} which gives the boson resistivity as 

3831: $R_b = R_{d_1} + R_{d_2}$.

3832: For the Fermi surfaces as in the right panel of Fig.~\ref{fig:introFS},

3833: and the conductivity measured, say, in the $\hat{\bf x}$ direction,

3834: the small-angle scattering by the gauge field can effectively degrade 

3835: the $d_2$ fermion current, therefore the standard result 

3836: $R_{d_2} \sim T^{4/3}$.

3837: On the other hand, such scattering cannot completely degrade the $d_1$ 

3838: fermion current, so $R_{d_1} \ll R_{d_2}$, and $R_b \sim T^{4/3}$ 

3839: follows. 

3840: (Parenthetically, the $d_1$ current can be degraded by quartic or more 

3841: particle interactions, but such contributions vanish as $T^2$ or faster.)

3842: 

3843: Intriguingly, in the DLBL phase, the boson correlators are expected 

3844: to be short-ranged despite such manifest thermodynamic signatures of 

3845: gaplessness and criticality.

3846: On the other hand, the boson box correlator in the DLBL decays as a 

3847: power law and is negative, $-x^{-8}$, offering perhaps a more physical 

3848: glimpse of the fermionic partons in this phase.

3849: Indeed, consider inserting bosons at $(0,0)$ and $(x,x)$

3850: and removing at $(x,0)$ and $(0,x)$.  The dominant contribution

3851: to the box correlator comes from the $d_1$ fermions propagating

3852: $(0,0) \to (x,0)$ and $(x,x) \to (0,x)$, while the $d_2$ paths

3853: are $(0,0) \to (0,x)$ and $(x,x) \to (x,0)$,

3854: which is illustrated in Fig.~\ref{fig:partons}.

3855: The minus sign of the box correlator is then due to one fermionic

3856: exchange needed so that the two pairs of injected fermions are removed 

3857: with the same pairing.

3858: 

3859: \begin{figure}

3860: \centerline{\includegraphics[width=3in]{figures/partons.eps}}

3861: \vskip -2mm

3862: \caption{Boson box correlator 

3863: $\la b^\dagger(0,0) b^\dagger(x,x) b(x,0) b(0,x) \ra$

3864: in the DLBL phase reveals the fermionic character of the partons,

3865: decaying as power law and having negative sign from the fermion

3866: exchange.

3867: }

3868: \label{fig:partons}

3869: \end{figure}

3870: 

3871: The proposed DBL/DLBL wavefunctions do not satisfy Marshall signs,

3872: and their interesting sign structure is brought out by the nodal 

3873: pictures in the continuum such as Fig.~\ref{fig:nodes},

3874: which give some caricature of the wavefunction signs also on the 

3875: lattice even though the nodes are not sharply defined in this case.

3876: Clearly, these wavefunctions cannot be ground states of boson models 

3877: without frustration.

3878: To address the question of what Hamiltonians may stabilize such phases,

3879: we considered a particular frustrated hard core boson model with

3880: competing hopping and four-site ring exchange terms, 

3881: and our energetics study suggests that the DLBL state is a good 

3882: candidate in this model near half-filling.  

3883: Unfortunately, the frustrated nature of the boson motion

3884: makes this model not suitable for large system quantum Monte Carlo 

3885: studies.  We still suggest it as an interesting model for numerical

3886: studies such as exact diagonalization of small systems or DMRG; 

3887: the gapless nature of the boson liquid may make results hard to 

3888: interpret, but the short-range character of the correlations and 

3889: particularly the quasi-local nature of the DLBL phase can perhaps 

3890: facilitate DMRG to access larger systems.

3891: 

3892: 

3893: Looking ahead, we now describe one of the main drives behind the

3894: study of uncondensed boson liquids, despite its own intrinsic interest.

3895: We are searching for electronic conducting non-Fermi liquids,

3896: and the ideas of the present work suggest some avenues in this

3897: direction.

3898: As a specific example, consider the slave boson approach that is

3899: popular in the context of the $t-J$ model of high-$T_c$ superconductors.

3900: The electron operator is written as 

3901: $c^\dagger_\sigma = b^\dagger f^\dagger_\sigma$,

3902: leading to a theory of spinons and slave bosons strongly coupled via

3903: an emergent fluctuating gauge field.

3904: Phenomenologically, a Fermi sea of spinons is very

3905: appealing in the strange metal, but what about the slave bosons?

3906: If they condense, we would recover the conventional Fermi liquid.

3907: It has been argued that the fluctuating gauge field frustrates the 

3908: motion of the bosons and suppresses the tendency to 

3909: condense.\cite{IoffeLarkin, IoffeKotliar, LeeNagaosa, Feigelman, LeeNagaosaWen}

3910: If this suppression could persist to zero temperature, we would

3911: obtain a conducting non-Fermi liquid state.

3912: Such a possibility is precisely what we are trying to establish in the 

3913: present work, and our approach would be to write the slave boson in 

3914: terms of new (second generation) slave fermions, 

3915: $b^\dagger = d_1^\dagger d_2^\dagger$.

3916: 

3917: On the level of wavefunctions, we would then write an electronic state 

3918: of the form:

3919: \begin{eqnarray*}

3920: \Psi_{\rm electron}(\up, \dn)

3921: = [(\det)_x \times (\det)_y] (\up, \dn) \,\, \det(\up) \, \det(\dn) ~,

3922: \end{eqnarray*}

3923: where schematically $(\up)$ or $(\dn)$ denotes locations

3924: of all electrons with spin-up or spin-down respectively.

3925: Each determinant in the slave boson wavefunction 

3926: $(\det)_x \times (\det)_y$ evaluates appropriate $d$-particle

3927: orbitals at the locations of all electrons irrespective of their

3928: spin, which assures the no-double occupancy constraint.  

3929: There is significant freedom in specifying these orbitals,

3930: and whether such wavefunctions are useful for any electronic system

3931: requires detailed energetics studies of specific Hamiltonians.

3932: One can nevertheless develop a low-energy theory of such a $T=0$ phase 

3933: in the spirit of the present paper, thereby obtaining an itinerant 

3934: non-Fermi liquid conducting phase.

3935: If the DBL or DLBL bosonic states studied in this work appear as

3936: useful caricatures of the behavior of the electronic charge, 

3937: we may even call the resulting phase a ``D-wave metal''!

3938: Such all-fermion description of electrons on the square lattice with 

3939: no double occupancy has some phenomenological appeal, but detailed 

3940: explorations are left for future work.

3941: 

3942: 

3943: 

3944: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

3945: \acknowledgments

3946: We would like to acknowledge discussions with T.~Senthil and

3947: A.~Vishwanath, and thank Jason Alicea for help with some calculations.

3948: The work at KITP was supported by the National Science Foundation 

3949: through grants PHY-9907949 and DMR-0529399.

3950: 

3951: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

3952: 

3953: 

3954: 

3955: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

3956: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

3957: \begin{thebibliography}{10}

3958: 

3959: 

3960: \bibitem{Anderson}

3961: P. W. Anderson, Science {\bf 235}, 1196 (1987).

3962: 

3963: \bibitem{Baskaran}

3964: % The resonating valence bond state and high-T_c superconductivity

3965: % - a mean field theory

3966: G. Baskaran, Z. Zou and P. W. Anderson, 

3967: Solid State Commun. {\bf 63}, 973 (1987).

3968: 

3969: \bibitem{KotliarLiu}

3970: % Superexchange mechanism and d-wave superconductivity

3971: G. Kotliar and J. Liu,

3972: Phys. Rev. B {\bf 38}, 5142 (1988). 

3973: 

3974: \bibitem{IoffeLarkin}

3975: % Gapless fermions and gauge fields in dielectrics

3976: L. B. Ioffe and A. I. Larkin,

3977: Phys. Rev. B {\bf 39}, 8988 (1989).

3978: 

3979: \bibitem{LeeNagaosaWen}

3980: % Doping a Mott insulator: Physics of high-temperature superconductivity

3981: P. A. Lee, N. Nagaosa, and X.-G. Wen,

3982: Rev. Mod. Phys. {\bf 78}, 17 (2006).

3983: 

3984: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

3985: 

3986: 

3987: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

3988: 

3989: \bibitem{Feigelman}

3990: % Two-dimensional Bose liquid with strong gauge-field interaction

3991: M. V. Feigelman, V. B. Geshkenbein, L. B. Ioffe, and A. I. Larkin

3992: Phys. Rev. B {\bf 48}, 16641 (1993).

3993: 

3994: \bibitem{Dalidovich}

3995: % Interaction-induced Bose metal in two dimensions

3996: D. Dalidovic and P. Phillips, Phys. Rev. B 64, 052507 (2001).

3997: 

3998: \bibitem{Galitski}

3999: % Vortices and Quasiparticles near the Superconductor-Insulator 

4000: % Transition in Thin Films

4001: V. M. Galitski, G. Refael, M. P. A. Fisher, and T. Senthil

4002: Phys. Rev. Lett. {\bf 95}, 077002 (2005).

4003: 

4004: \bibitem{Alicea}

4005: % Algebraic Vortex Liquid in Spin-1/2 Triangular Antiferromagnets: 

4006: % Scenario for Cs2CuCl4

4007: J. Alicea, O. I. Motrunich, and M. P. A. Fisher,

4008: Phys. Rev. Lett. {\bf 95}, 247203 (2005);

4009: % Theory of the algebraic vortex liquid in an anisotropic spin-1/2 

4010: % triangular antiferromagnet

4011: J. Alicea, O. I. Motrunich, and M. P. A. Fisher,

4012: Phys. Rev. B {\bf 73}, 174430 (2006).

4013: 

4014: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

4015: 

4016: 

4017: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

4018: 

4019: \bibitem{Ceperley91}

4020: % Fermion Nodes

4021: D. M. Ceperley,

4022: J. Stat. Phys. {\bf 63}, 1237 (1991).

4023: 

4024: \bibitem{WenPSG}

4025: % Quantum orders and symmetric spin liquids

4026: X.-G. Wen, 

4027: Phys. Rev. B {\bf 65}, 165113 (2002); cond-mat/0107071.

4028: 

4029: \bibitem{GirvinMacDonald}

4030: % Off-diagonal long-range order, oblique confinement, and the fractional 

4031: % quantum Hall effect

4032: S. M. Girvin and A. H. MacDonald,

4033: Phys. Rev. Lett. {\bf 58}, 1252 (1987).

4034: 

4035: \bibitem{Kane}

4036: % General validity of Jastrow-Laughlin wave functions

4037: C. L. Kane, S. Kivelson, D. H. Lee, and S. C. Zhang,

4038: Phys. Rev. B {\bf 43}, 3255 (1991).

4039: 

4040: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

4041: 

4042: 

4043: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

4044: 

4045: \bibitem{Paramekanti}

4046: % Ring exchange, the exciton Bose liquid, and bosonization in two dimensions

4047: A. Paramekanti, L. Balents, and M. P. A. Fisher,

4048: Phys. Rev. B {\bf 66}, 054526 (2002).

4049: 

4050: \bibitem{Sandvik}

4051: % Striped Phase in a Quantum XY Model with Ring Exchange

4052: A. W. Sandvik, S. Daul, R. R. P. Singh, and D. J. Scalapino, 

4053: Phys. Rev. Lett. {\bf 89}, 247201 (2002).

4054: 

4055: \bibitem{Melko}

4056: % Two-dimensional quantum XY model with ring exchange and external field

4057: R. G. Melko, A. W. Sandvik, and D. J. Scalapino,

4058: Phys. Rev. B {\bf 69}, 100408 (2004).

4059: 

4060: \bibitem{Rousseau}

4061: % Ring exchange and phase separation in the two-dimensional boson 

4062: % Hubbard model

4063: V. G. Rousseau, R. T. Scalettar, and G. G. Batrouni,

4064: Phys. Rev. B {\bf 72}, 054524 (2005);

4065: % Phase Separation in the Two-Dimensional Bosonic Hubbard Model with 

4066: % Ring Exchange

4067: V. Rousseau, G. G. Batrouni, and R. T. Scalettar,

4068: Phys. Rev. Lett. {\bf 93}, 110404 (2004).

4069: 

4070: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

4071: 

4072: 

4073: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

4074: 

4075: \bibitem{Zhang_CS}

4076: % Effective-Field-Theory Model for the Fractional Quantum Hall Effect 

4077: S.C. Zhang, T.H. Hansson and S. Kivelson, 

4078: Phys. Rev. Lett. {\bf 62}, 82 (1989).

4079: 

4080: \bibitem{MPAF_Lee}

4081: % Anyon superconductivity and the fractional quantum Hall effect 

4082: D.H. Lee and M.P.A. Fisher,

4083: Phys. Rev. Lett. {\bf  63}, 903 (1989).

4084: 

4085: \bibitem{Wen_edge}

4086: % Gapless boundary excitations in the quantum Hall states and in the 

4087: % chiral spin states

4088: X.G. Wen,

4089: Phys. Rev. B {\bf 43}, 11025 (1991); 

4090: % Electrodynamical properties of gapless edge excitations in the 

4091: % fractional quantum Hall states

4092: Phys. Rev. Lett. {\bf 64}, 2206 (1990). 

4093: 

4094: 

4095: \bibitem{Wen_proj4FQHE}

4096: %Projective Construction of Non-Abelian Quantum Hall Liquids

4097: X.-G. Wen, Phys. Rev. B {\bf 60}, 8827 (1999).

4098: 

4099: \bibitem{KL}

4100: % Theory of the spin liquid state of the Heisenberg antiferromagnet

4101: V. Kalmeyer and R. B. Laughlin, Phys. Rev. B {\bf 39}, 11879 (1989);

4102: % Equivalence of the resonating-valence-bond and fractional 

4103: % quantum Hall states

4104: Phys. Rev. Lett. {\bf 59}, 2095 (1987).

4105: 

4106: \bibitem{WenWilczekZee}

4107: % Chiral spin states and superconductivity

4108: X. G. Wen, F. Wilczek, and A. Zee,

4109: Phys. Rev. B {\bf 39}, 11413 (1989).

4110: 

4111: \bibitem{LaughlinZou}

4112: % Properties of the chiral-spin-liquid state

4113: R. B. Laughlin and Z. Zou, Phys. Rev. B {\bf 41}, 664 (1990).

4114: 

4115: \bibitem{ZhouWen}

4116: % Quantum Orders and Spin Liquids in Cs$_2$CuCl$_4$

4117: Y. Zhou and X.-G. Wen, cond-mat/0210662.

4118: 

4119: \bibitem{Yunoki06}

4120: % Two spin liquid phases in the spatially anisotropic triangular 

4121: % Heisenberg model

4122: S. Yunoki and S. Sorella,

4123: Phys. Rev. B {\bf 74}, 014408 (2006).

4124: 

4125: \bibitem{Rantner}

4126: % Spin correlations in the algebraic spin liquid: 

4127: % Implications for high-Tc superconductors

4128: W. Rantner and X.-G. Wen, 

4129: Phys. Rev. B {\bf 66}, 144501 (2002).

4130: 

4131: \bibitem{Hermele}

4132: % Stability of U(1) spin liquids in two dimensions

4133: M. Hermele, T. Senthil, M. P. A. Fisher, P. A. Lee, N. Nagaosa, 

4134: and X.-G. Wen,

4135: Phys. Rev. B {\bf 70}, 214437 (2004);

4136: % Algebraic spin liquid as the mother of many competing orders

4137: M. Hermele, T. Senthil, and M. P. A. Fisher,

4138: Phys. Rev. B {\bf 72}, 104404 (2005).

4139: 

4140: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

4141: 

4142: 

4143: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

4144: 

4145: \bibitem{vmc}

4146: % Monte Carlo simulation of a many-fermion study

4147: D. M. Ceperley, G. V. Chester and M. H. Kalos,

4148: Phys. Rev. B {\bf 16}, 3081 (1977).

4149: 

4150: \bibitem{Zhang}

4151: % A renormalised hamiltonian approach for a resonant valence bond

4152: % wavefunction

4153: F. C. Zhang, C. Gros, T. M. Rice, and H. Shiba,

4154: Supercond. Sci. Technol. {\bf 1}, 36 (1988) (cond-mat/0311604).

4155: 

4156: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

4157: 

4158: 

4159: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

4160: 

4161: \bibitem{Holstein}

4162: % de Haas-van Alphen Effect and the Specific Heat of an Electron Gas

4163: T. Holstein, R. E. Norton, and P. Pincus,

4164: Phys. Rev. B {\bf 8}, 2649 (1973).

4165: 

4166: \bibitem{Reizer}

4167: % Relativistic effects in the electron density of states, 

4168: % specific heat, and the electron spectrum of normal metals

4169: M. Reizer, Phys. Rev. B {\bf 40}, 11571 (1989).

4170: 

4171: \bibitem{PALee}

4172: % Gauge field, Aharonov-Bohm flux, and high-Tc superconductivity

4173: P. A. Lee, Phys. Rev. Lett. {\bf 63}, 680 (1989).

4174: 

4175: \bibitem{IoffeKotliar}

4176: % Transport phenomena near the Mott transition

4177: L. B. Ioffe and G. Kotliar,

4178: Phys. Rev. B {\bf 42}, 10348 (1990).

4179: 

4180: \bibitem{LeeNagaosa}

4181: % Gauge theory of the normal state of high-Tc superconductors

4182: P. A. Lee and N. Nagaosa,

4183: Phys. Rev. B {\bf 46}, 5621 (1992).

4184: 

4185: \bibitem{Polchinski}

4186: % Low-energy dynamics of the spinon-gauge system

4187: J. Polchinski, Nucl. Phys. B {\bf 422}, 617 (1994).

4188: 

4189: \bibitem{Altshuler}

4190: % Low-energy properties of fermions with singular interactions

4191: B. L. Altshuler, L. B. Ioffe, and A. J. Millis

4192: Phys. Rev. B {\bf 50}, 14048 (1994) 

4193: 

4194: \bibitem{Nayak}

4195: % Non-Fermi liquid fixed point in 2 + 1 dimensions

4196: C. Nayak and F. Wilczek, Nucl. Phys. B {\bf 417}, 359 (1994);

4197: % Renormalization group approach to low temperature properties of a non-Fermi liquid metal

4198: {\it ibid.} {\bf 430}, 534 (1994).

4199: 

4200: \bibitem{YBKim}

4201: % Gauge-invariant response functions of fermions coupled to a gauge field

4202: Y. B. Kim, A. Furusaki, X. G. Wen, and P. A. Lee

4203: Phys. Rev. B {\bf 50}, 17917 (1994);

4204: % Quantum Boltzmann equation of composite fermions interacting with a 

4205: % gauge field

4206: Y. B. Kim, P. A. Lee, and X. G. Wen,

4207: Phys. Rev. B {\bf 52}, 17275 (1995).

4208: 

4209: \bibitem{Senthil}

4210: % Weak magnetism and non-Fermi liquids near heavy-fermion critical points

4211: T. Senthil, M. Vojta, and S. Sachdev,

4212: Phys. Rev. B {\bf 69}, 035111 (2004).

4213: 

4214: \bibitem{Galitski_gauge}

4215: % Metallic phase in a two-dimensional disordered Fermi system with 

4216: % singular interactions

4217: V. M. Galitski, 

4218: Phys. Rev. B {\bf 72}, 214201 (2005).

4219: 

4220: 

4221: \bibitem{footnote:N_flavor}

4222: The structure of the $1/N$ diagrams for the particle-hole and 

4223: particle-particle vertices, figures \ref{fig:Vph} and \ref{fig:Vpp}, 

4224: is specific to our $N$-flavor extension defined in Sec.~\ref{subsec:N_flavor}, 

4225: while different extensions can lead to different organization of 

4226: diagrams.

4227: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

4228: 

4229: 

4230: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

4231: 

4232: \bibitem{Shankar}

4233: % Renormalization-group approach to interacting fermions

4234: R. Shankar,

4235: Rev. Mod. Phys. {\bf 66}, 129 (1994).

4236: 

4237: \bibitem{Polchinski_FL}

4238: J. Polchinski, Effective Field Theory and the Fermi Surface,

4239: {\it in} Recent directions in particle theory,

4240: Proc. 1992 TASI, eds. J. Harvey and J. Polchinski

4241: (World Scientific, Singapore, 1993);

4242: hep-th/9210046

4243: 

4244: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

4245: 

4246: \end{thebibliography}

4247: 

4248: \end{document}

4249: