1: \documentclass[twocolumn, prl]{revtex4}
2:
3: \usepackage{graphicx}% Include figure files
4: \usepackage{dcolumn}% Align table columns on decimal point
5: \usepackage{bm}% bold math
6:
7:
8: \begin{document}
9:
10: \title{ N\'eel and Valence-Bond Crystal phases of the Two-Dimensional Heisenberg Model on the Checkerboard Lattice}
11:
12: \author {S. Moukouri }
13:
14: \affiliation{ Department of Physics and Michigan Center for
15: Theoretical Physics \\
16: University of Michigan 2477 Randall Laboratory, Ann Arbor MI 48109}
17:
18: \begin{abstract}
19: I use an improved version of the two-step density matrix renormalization
20: group method to study ground-state properties of the 2D Heisenberg model on the
21: checkerboard lattice. In this version the Hamiltonian is projected on a tensor
22: product of two-leg ladders instead of chains. This allows investigations of
23: 2D isotropic models. I show that this method can describe both the
24: magnetically disordered and ordered phases. The ground-state phases of
25: the checkerboard model
26: as $J_2$ increases are: (i) N\'eel with $Q=(\pi,\pi)$, (ii) a valence
27: bond crystal (VBC)
28: of plaquettes, (iii) N\'eel with $Q=(\pi/2,\pi)$, and (iv) a VBC of crossed
29: dimers. In agreement with previous results, I find that at the isotropic
30: point $J_2=J_1$, the ground state is made of weakly
31: interacting plaquettes with a large gap $\Delta \approx 0.67 J_1$ to triplet
32: excitations.
33:
34:
35: \end{abstract}
36:
37: \maketitle
38:
39: Frustration-induced magnetically disordered phases in two dimensions (2D)
40: recently have attracted substantial interest \cite{lhuillier}.
41: Frustrated magnets are known to display unconventional ground states
42: with, in some cases, a large set of low-lying degenerate singlet excitations
43: that are still not well understood. Among models of frustrated systems, the
44: Heisenberg model on the checkerboard lattice (HMCL) has recently been
45: intensively studied by various techniques \cite{lieb,palmer,fouet,
46: canals,berg,sfb}. This model is seen as
47: a first step in the investigation of the 3D pyrochlore model.
48: The emerging picture is that at the isotropic point
49: ($J_1=J_2$), the HMCL spontaneously breaks the lattice's translational
50: symmetry. The ground state is a singlet made of a collection of weakly coupled
51: plaquettes with a large gap, $\Delta \approx 0.7 J_1$, to triplet excitations.
52: Away from the isotropic point, the situation is less clear. There
53: is no single method which can capture the full phase diagram.
54:
55: In this letter, I introduce an improved version of the two-step density-matrix
56: renormalization group (TSDMRG) \cite{moukouri-TSDMRG, moukouri-TSDMRG2} which,
57: as I will
58: show, is very convenient in the study of the HMCL and other 2D frustrated
59: models. This new version is based on using the two-leg ladder, instead of chains,
60: as the starting point to build the 2D lattice.
61: The main insight in using the two-leg ladder to construct the
62: 2D lattice comes from large $N$ predictions
63: \cite{read-sachdev} that frustration often induces ground states in which the
64: translationaly symmetry is broken. In the strong-coupling regime of the
65: disordered phase of $S=1/2$ systems, the system is made of a collection
66: of singlets or plaquettes. This strong coupling regime cannot be described
67: starting from independent chains which are gapless.
68: Starting from a single chain, small transverse perturbations can yield a
69: gap within the TSDMRG. But this gap is often small
70: and it is difficult to obtain reliable extrapolations. The two-leg
71: ladder does not present this problem. It does already present a large gap
72: $\Delta \approx 0.5$ even in absence of frustration. Coupled ladders naturally
73: evolve toward the 2D N\'eel state as the number of legs increases. Hence,
74: in principle, disordered and ordered phases could be described within a
75: two-leg ladder version of the TSDMRG. This suggests that the two-leg ladder is a more
76: natural starting point to describe ground state phases of 2D antiferromagnets
77: than the single chain.
78:
79: Additional insights into this idea came from my comparative study of
80: coupled chains with half-integer and integer spins \cite{moukouri-TSDMRG3}.
81: In Ref.\cite{moukouri-TSDMRG3}, when starting from single chains, I found
82: that although chains with $S=1$ display the Haldane gap, $\Delta \approx 0.4$,
83: they converge much faster to the N\'eel state than those with $S=1/2$.
84: Furthermore, when a frustration induced disordered phase is present, it can be
85: much more easily found in the case $S=1$. Hence, following the equivalence
86: between the two-leg ladder and the Haldane spin chain, suggested by the
87: Affleck-Kennedy-Lieb-Tasaki construction \cite{aklt}, it
88: would be better to adopt the two-leg ladder as the building block for
89: two-dimensional lattices.
90:
91: I will now illustrate this idea in the case of the HMCL. Following the
92: usual notation, the HMCL is given by:
93:
94: \begin{eqnarray}
95: H=J_1 \sum_{<i,j>}{\bf S}_i{\bf S}_j+
96: J_2 \sum_{[i,j]}{\bf S}_i{\bf S}_j,
97: \label{hamiltonian}
98: \end{eqnarray}
99:
100: \noindent where $<i,j>$ represents nearest-neighbor sites and $[i,j]$
101: stand for next-nearest neighbors on every other plaquette. $J_1$ is set as
102: the unit energy.
103:
104: The TSDMRG with ladders is similar to the method with
105: chains. So I refer the reader to Ref.\cite{moukouri-TSDMRG,
106: moukouri-TSDMRG2} for a complete exposition of
107: the algorithm. Here, I will discuss only briefly the main points of the
108: algorithm. I start by dividing the 2D lattice into two-leg ladders; the
109: Hamiltonian (\ref{hamiltonian}) is written as,
110:
111: \begin{eqnarray}
112: H=\sum_{ladders}H_{ladder}+H_{int},
113: \label{hamiltonian2}
114: \end{eqnarray}
115:
116: \noindent where $H_{ladder}$ is the Hamiltonian of a single two-leg ladder,
117: $H_{int}$ contains the inter-ladder part. In the first step of the
118: method, the usual DMRG method is applied to generate a low
119: energy Hamiltonian of an isolated ladder of $N_x$ sites keeping $m_1$ states.
120: Then $m_2$ low-lying states of the superblock states, the corresponding
121: energies, and all the local spin operators are kept. These energies represent
122: the renormalized low energy Hamiltonian of a single ladder. The Hamiltonian
123: (\ref{hamiltonian2}) is then projected onto the tensor product basis of
124: independent ladders,
125:
126: \begin{equation}
127: \Psi=\prod_{ladders} \Phi_{ladder},
128: \end{equation}
129:
130: \noindent where $\Phi_{ladder}$ is an eigenfunction of $H_{0,ladder}$.
131: This yield an effective Hamiltonian,
132:
133: \begin{eqnarray}
134: H_{eff}=\sum_{ladders}H_{0,ladder}+{\tilde H}_{int}.
135: \label{hamiltonian3}
136: \end{eqnarray}
137:
138: \begin{figure}
139: \begin{center}
140: $\begin{array}{c@{\hspace{0.5in}}c}
141: \multicolumn{1}{l} {}\\ [-0.23cm]
142: \includegraphics[width=1.5 in, height=1.5 in]{gap_lad.eps}
143: \hspace{0.25cm}
144: \includegraphics[width=1.5 in, height=1.5 in]{eg.eps}
145: \end{array}$
146: \end{center}
147: \caption{(a) Spin gap of the two-leg ladder for $J_2=0$ (circles),
148: $0.5$ (squares), $1$ (diamonds), $1.1$ (triangles up), and $2$
149: (triangles down). (b) Ground-state energies as function of the system size
150: for a two-leg ladder for $J_2=1$ (circles), $J_2=2$ (diamonds), and for
151: the 2D lattice for $J_2=1$ (squares), $J_2=2$ (triangle up).}
152:
153: \vspace{0.5cm}
154: \label{gaplad-eg}
155: \end{figure}
156:
157:
158: The resulting effective coupled ladder problem which is 1D is studied
159: again by the DMRG method in the transverse direction. The TSDMRG is
160: variational, as the original DMRG method, the subspace spanned
161: by the wavefunctions of the
162: form $\Psi$ is a subspace of the full Hilbert space of
163: Hamiltonian (\ref{hamiltonian}). Its convergence depends on $m_1$ and $m_2$,
164: the error is given by $max(\rho_1,\rho_2)$, where $\rho_1$ and $\rho_2$
165: are the truncation errors in the first and second steps respectively.
166: $m_2$ fixes the energy band-width $\delta E$. The method is accurate only
167: when the inter-ladder couplings are small with respect to $\delta E$.
168: In the present simulations $\delta E \approx 4$. Since
169: for the HMCL the inter-ladder and intra-ladder are of the same magnitude,
170: in principle this approach would be plagued by the same deficiencies the
171: block RG method. But if the starting point is chosen so that
172: the essential physics is already contained at the level of the ladder, the
173: effective strength of the inter-ladder couplings will be small even if the
174: bare couplings are not. This is particularly the case of models with
175: frustration in which the competing interactions largely cancel each other
176: in the strong frustration regime, yielding weakly coupled sub-clusters.
177:
178:
179:
180: \begin{figure}
181: \begin{center}
182: $\begin{array}{c@{\hspace{0.5in}}c}
183: \multicolumn{1}{l} {}\\ [-0.23cm]
184: \includegraphics[width=1.5 in, height=1.5 in]{src_cb1.eps}
185: \hspace{0.25cm}
186: \includegraphics[width=1.5 in, height=1.5 in]{src_cb2.eps}
187: \end{array}$
188: \end{center}
189: \caption{Short-range correlations $C_l$ (circles), $C_r$ (squares), $C_d$
190: (diamonds) for the two-leg ladder for uncrossed (a) and crossed (b)
191: plaquettes as function of $J_2$}
192: \vspace{0.5cm}
193: \label{src}
194: \end{figure}
195:
196: The ground state properties of an isolated ladder can readily be obtained.
197: I keep up to $m_1=96$ and $N_x=16$ and I target spin sectors from
198: $S_z=0$ to $S_z=\pm 4$ and used open boundary conditions (OBC).
199: The maximum error is $\rho_1= 1\times 10^{-4}$. There
200: is a gap $\Delta$ for all values of $J_2$ investigated between $0$ and
201: $2$. The finite size behavior of gaps for some typical values of $J_2$ are shown
202: in Fig.\ref{gaplad-eg}. The case $J_2=0$ reduces to the usual two-leg ladder
203: which has been widely studied in the literature \cite{acm}. For
204: $J_2=0$, $\Delta \approx 0.5$ . As $J_2$ increases, $\Delta$ has a
205: non-monotonous behavior. This suggests a rich structure which is
206: revealed more clearly by the analysis of the correlation functions.
207: I computed the following short-range correlation functions: the bond
208: strength along a leg $C_{l_{u,c}}=\langle {\bf S}_{i,1}
209: {\bf S}_{i+1,1} \rangle_{u,c}$ for uncrossed (u) and crossed (c) plaquettes,
210: the diagonal correlation
211: $C_{d_{u,c}}=\langle {\bf S}_{i,1}{\bf S}_{i+1,2} \rangle_{u,c}$, and
212: bond strength along the rungs $C_{r}=\langle {\bf S}_{i,1}
213: {\bf S}_{i,2} \rangle$. Note that I have introduced a second index to
214: the local spin. These correlations are shown in Fig.\ref{src}. Four
215: regions can be identified: (i) region I (rung dimers): $0 \alt J_2 \alt 0.6$,
216: $C_{l_{u,c}} < 0$, $C_{l_u} \approx C_{l_c}$, $C_{d_{u,c}} > 0$, $C_r <0$,
217: and $|C_r| > |C_{l_u}|$; the dominant spin-spin correlations are along
218: the rungs. The ground state properties of the ladder in this
219: region are identical to those of the unfrustrated ladder ($J_2=0$).
220: (ii) Region II (plaquettes I): $0.6 \alt J_2 \alt 1$,
221: $C_{l_{u,c}} < 0$, $|C_{l_u}| > |C_{l_c}|$, $C_{d_{u,c}} > 0$, $C_r <0$,
222: and $|C_r| < |C_{l_u}|$; the physics is dominated by that of the isotropic
223: point. At this point, the ground state is a collection of weakly interacting
224: uncrossed plaquettes. Both $C_{d_c}$ and $C_{l_c}$ vanish at $J_2=1$.
225: In this region, the local spin configuration is the same on all the
226: uncrossed plaquettes
227: as shown in Fig.\ref{ph}(b). (iii) region III (plaquettes II):
228: $1 \alt J_2 \alt 1.3$, $C_{l_{u}} < 0$, $C_{l_c} > 0$, $C_{d_{u}} > 0$,
229: $C_{d_{c}} < 0$, $C_r <0$, and $|C_r| < |C_{l_u}|$; in this region,
230: the ground state is again dominated by uncrossed plaquettes. But now the
231: local spin configurations on two consecutive uncrossed plaquettes are images
232: of one another by reflection with respect to a plane passing through the
233: middle of the crossed plaquette between them.
234: Region IV (crossed dimers): $1.3 \alt J_2 $, $C_{l_{u}} < 0$,
235: $C_{d_{c}} < 0$, $C_{l_c}=C_r =C_{d_{u}} \approx 0$, and
236: $|C_r| < |C_{l_u}|$; the ground state is dominated by the crossed dimers on
237: crossed plaquettes as shown in Fig.\ref{ph}(d). The sketch of the
238: spin structure corresponding to each region is summarized in Fig.\ref{ph}.
239: Since I applied OBC, for a given size, there are two possible
240: ground states depending on the plaquette pattern: {\bf (a)} $ucu...ucu$ or
241: {\bf (b)} $cuc...cuc$. In region I, the configurations {\bf (a)} and {\bf (b)}
242: have nearly the same energy. This is consistent with the fact that the
243: translational symmetry is not broken. But in Region II and III, {\bf (a)} has
244: the lowest energy, since it has a larger number of uncrossed plaquettes.
245: By contrast, in Region IV where dimer order is dominant, it is {\bf (b)}
246: that has the lowest energy.
247:
248: \begin{figure}
249: \includegraphics[width=2. in, height=1.5 in]{cb_ladder.eps}
250: \caption{ The four phases of the two-leg ladder: rung dimers (a),
251: plaquette I (b), plaquette II (c), and crossed dimers (d).}
252: \vspace{0.5cm}
253: \label{ph}
254: \end{figure}
255:
256: The 2D systems are obtained by applying the DMRG on $H_{eff}$ in the
257: transverse direction.
258: I studied systems of size $N_x \times N_y=4\times6,~8\times10,
259: ~12\times14$ and $16\times18$.
260: I kept up to $m_2=96$ and used OBC. Inter-ladder interactions will have
261: very different effects depending on
262: whether they correspond to a magnetic regime or a disordered regime.
263: I will first consider their effects on region II, which includes the
264: isotropic point. Recently, there have been a number of studies which strongly
265: suggest that the physics of the 2D systems is identical to that displayed
266: by the two-leg ladder. In other words, the ground state is essentially made of weakly
267: interacting plaquettes. If this is the case, it means that the inter-ladder
268: interactions will not strongly modify the ground state wave function
269: of decoupled ladders. Fig.\ref{gaplad-eg} shows that the
270: ground state energy and $\Delta$ remain very close to that of an isolated
271: plaquette. Thus in the vicinity of $J_2=1$, inter-ladder interactions do
272: not strongly renormalize the properties of an isolated ladder which
273: themselves are close to those of an isolated plaquette. The extrapolated
274: gap is found to be $\Delta =0.67 J_1$ which is in good agreement with
275: the prediction from exact diagonalization \cite{fouet}. The same conclusion
276: is seen in Fig.\ref{gaplad-eg} for region IV where the crossed-dimer ground state
277: found for the ladder is also the ground state of the 2D lattice. In both
278: cases, the wave function made of the tensor product of the wave function
279: of single two-leg ladders is a good variational wave function for the 2D
280: system. In each case, the ground state energy of the 2D system remains
281: very close to that of individual plaquettes or crossed dimers. This
282: can be explained as follows: when ladders are brought together to build
283: the 2D lattice, the dominant local correlations are $C_{l_u}$
284: in region II and $C_{d_c}$ in region IV; during this process,
285: magnetic energy cannot efficiently be gained. For region II, this
286: is because the two
287: neighboring plaquettes of an uncrossed plaquette in the direction of the
288: rungs involve frustrated bonds. Hence the system prefers the original
289: configuration to avoid increasing its energy. For region IV, $C_r$ is
290: very small. The system cannot increase it when the ladders are coupled, because
291: the spins are already involved in strong diagonal dimers. There is,
292: however, the possibility to gain magnetic energy by forming N\'eel order
293: along the direction of the diagonal bonds ($J_2$ direction) as suggested
294: in Ref.\cite{sfb}. This is unlikely, however, because once such
295: a phase is reached, I do not see how the system could go to crossed dimers
296: at larger $J_2$. The action of $J_1$ which act as frustration in this regime
297: decreases as $J_2$ increases. Hence once this hypothetical N\'eel phase along
298: the $J_2$ bonds is reached, there is no obvious mechanism that could destroy
299: it as $J_2$ increases to yield the crossed-dimer phase as suggested in
300: Ref.\cite{sfb}. Such a N\'eel phase would be favored only
301: when $J_2 \gg J_1$. I made rough calculations with $J_2=4,8$ and I found
302: that the system remains in the crossed-dimer phase. The situation
303: is apparently identical to the $J_1-J_2$ chain where the independent
304: chains regime is only reached in the infinite $J_2$ limit.
305:
306: \begin{figure}
307: \begin{center}
308: $\begin{array}{c@{\hspace{0.5in}}c}
309: \multicolumn{1}{l} {}\\ [-0.23cm]
310: \includegraphics[width=1.5 in, height=1.5 in]{gap_2D.eps}
311: \hspace{0.25cm}
312: \includegraphics[width=1.5 in, height=1.5 in]{corlgp1.25l16.eps}
313: \end{array}$
314: \end{center}
315: \caption{ (a) 2D gaps as function of the system size $J_2=0$ (circles),
316: $0.5$ (squares), $1$ (diamonds), $1.1$ (triangles up), and
317: $2$ (triangles down). (b) Correlation function along the legs as function
318: of the distance for $J_2=1.25$.}
319: \vspace{0.5cm}
320: \label{gap-lrc}
321: \end{figure}
322:
323:
324: The situation is very different for regions I and III. In region I,
325: the dominant local correlation is $C_r$; when the ladders are brought
326: together, magnetic energy can be gained by
327: an antiferromagnetic arrangement along the rungs. This enhances the
328: local antiferromagnetic order which exists along the legs and ultimately leads to
329: a N\'eel order with $Q=(\pi,\pi)$. This is seen in the vanishing of
330: the spin gap for $J_2=0$ and $J_2=0.5$ shown in Fig.\ref{gap-lrc}(a).
331: This is in agreement with results for $J_2=0$ from quantum Monte Carlo
332: (QMC) simulations \cite{sandvik}
333: and large $S$ analysis \cite{canals}. I find that
334: the TSDMRG ground state energy $-0.6011$ at $J_2=0.$ is not in very good
335: agreement with the QMC result $-0.6699$ of Ref.\cite{sandvik}.
336: Despite this discrepancy, the TSDMRG is nevertheless
337: able to reproduce the low-energy behavior of the ordered phase. This is not
338: in fact surprising. In the Resonating valence bond picture, the N\'eel
339: state and its low energy excitations can be written as linear combination
340: of a tensor product of dimers. The TSDMRG variational solution of
341: Hamiltonian(\ref{hamiltonian}) which is a linear combination of
342: the wave functions $\Psi$ has exactly this form.
343: A similar analysis also applies for region III. $C_{l_u}$ is dominant
344: in region II. But as seen in Fig.\ref{src}, $C_{l_u}$ has a minimum at
345: $J_2=1$ and then increases. It becomes very close to $C_r$ when $J_2$
346: enters region III.
347: Hence magnetic energy can be gained again through the rungs. Since
348: the structure along the legs is not modified from Fig.\ref{ph}(d),
349: the resulting wave vector will be $Q=({\pi/2,\pi})$. This is seen in
350: Fig.\ref{gap-lrc} in the behavior of the spin-spin correlation function
351: $C_l(i)$ along the legs. $C_l(i)$ displays a period of $4$.
352: The correlations between the rungs (not shown) oscillate with $q_y=\pi$.
353: Fig.\ref{pd-2D} presents a sketch of the different ground state phases
354: of the HMCL as function of $J_2$. I note that in Ref.\cite{sfb}, a
355: very similar phase diagram was suggested; the only difference with
356: the TSDMRG phase diagram is the wave vector of the N\'eel phase
357: between the plaquette and crossed-dimer phases.
358:
359:
360: \begin{figure}
361: \includegraphics[width=3. in, height=1. in]{pd_2D.eps}
362: \caption{ Ground-state phases of the 2D checkerboard model as function
363: of $J_2$. Note that the phase boundaries are rough estimates taken from the
364: phases of the two-leg ladder. }
365: \vspace{0.5cm}
366: \label{pd-2D}
367: \end{figure}
368:
369: In summary, I have shown that the TSDMRG method can
370: reliably be used to study the disordered phases with short correlation
371: lengths of isotropic 2D models. In these phases, the system is
372: a collection of dimers or plaquettes. This makes the two-leg
373: ladder a very good starting point for a variational calculation. I showed
374: that the basic physics of 2D systems could already be read through
375: short-range correlations of the two-leg ladders.
376: This variational calculation is able to predict reasonably
377: magnetically ordered phases as well. In this work, I did not
378: discuss the question of low-lying singlet excitations within the gap.
379: Targeting them will lead
380: to large truncation errors and the calculations will become impractical.
381: These excitations are naturally truncated out when they are not needed
382: to form a target state. Finally, The same method could be applied
383: to the Sutherland-Shastry, $J_1-J_2$ or the Kagom\'e models in 2D.
384: It could also be applied to the pyrochlore lattice,
385: provided that the Hamiltonian could be written in some form
386: involving 1D subsystems with a large gap.
387:
388:
389: \begin{acknowledgments}
390: This work started during a visit at the Weizmann Institute. The author
391: thanks E. Altman for hospitality. This work was supported by the NSF
392: Grant No. DMR-0426775.
393: \end{acknowledgments}
394:
395:
396: \begin{thebibliography}{99}
397: \bibitem{lhuillier} G. Misguich and C. Lhuillier in "Frustrated Spin Systems"
398: Ed. H.T. Diep, World Scientific (2004).
399: \bibitem{lieb} E. H. Lieb and P. Schupp, Phys. Rev. Lett. {\bf 83}, 5362
400: (1999).
401: \bibitem{palmer}S. E. Palmer and J. T. Chalker, Phys. Rev. {\bf B 64}, 094412
402: (2001).
403: \bibitem{canals} B. Canals, Phys. Rev. {\bf B 65}, 184408 (2002).
404: \bibitem{fouet} J.-B. Fouet, M. Mambrini, P. Sindzingre, and C. Lhuillier,
405: Phys. Rev. {\bf B 67} 054411 (2003).
406: \bibitem{berg} E. Berg, E. Altman, and A. Auerbach, Phys. Rev. Lett. {\bf 90},
407: 147204 (2003).
408: \bibitem{sfb} O. A. Starykh, A. Furusaki, and L. Balents, Phys. Rev. {\bf B 72},
409: 094416 (2005).
410: \bibitem{moukouri-TSDMRG} S. Moukouri and L.G. Caron, Phys. Rev. {\bf B 67},
411: 092405 (2003).
412: \bibitem{moukouri-TSDMRG2} S. Moukouri, Phys. Rev. {\bf B 70}, 014403 (2004).
413: \bibitem{moukouri-TSDMRG3} S. Moukouri, J. Stat. Mech. P02002 (2006).
414: \bibitem{read-sachdev} N. Read and S. Sachdev, Phys. Rev. Lett. {\bf 66},
415: 1773 (1991).
416: \bibitem{aklt} I. Affleck, T. Kennedy, E. H. Lieb, and H. Tasaka,
417: Comm. Math. Phys. {\bf 115}, 477 (1988).
418: \bibitem{acm} M. Azzouz, Liang Chen, and S. Moukouri, Phys. Rev. {\bf B 50},
419: 6233 (1994).
420: \bibitem{sandvik} A. W. Sandvik, Phys. Rev. {\bf B 56} 11 678 (1997).
421:
422: %\bibitem{haldane1} F. D. M. Haldane, Phys. Lett. {\bf 93A}, 464 (1983);
423: % Phys. Rev. Lett. {\bf 50}, 1153 (1983).
424: %\bibitem{dombre} T. Dombre and N. Read, Phys. Rev. {\bf B 38}, 7181 (1988).
425: %\bibitem{stone} E. Fradkin and M. Stone, Phys. Rev. {\bf B 38}, 7215 (1988).
426: %\bibitem{zee} X.-G. Wen and A. Zee, Phys. Rev. Lett. {\bf 61}, 1025 (1988).
427: %\bibitem{haldane2} F. D. M. Haldane, Phys. Rev. Lett. {\bf 61}, 1029 (1988).
428: %\bibitem{dls} F. J. Dyson, E. H. Lieb, and B. Simon, J. Stat. Phys. {\bf 18},
429: % 335 (1978).
430: %\bibitem{neves} E. J. Neves and J. F. Peres, Phys. Lett. {\bf 114A},
431: % 331 (1986).
432: %\bibitem{young} J. D. Reger and A. P. Young, Phys. Rev. {\bf B 37}, 5978 (1988).
433: %\bibitem{affleck} I. Affleck, Phys. Rev. {\bf B 37}, 5186 (1988).
434: %\bibitem{lsm} E. Lieb, T. Schultz, and D. Mattis, Ann. Phys. (N.Y.) {\bf 16},
435: % 407 (1961).
436: %\bibitem{read} N. Read and S. Sachdev, Phys. Rev. Lett {\bf 62}, 1694 (1989).
437: %\bibitem {fradkin} E. Fradkin in "Field Theories of Condensed Matter Systems",
438: %Ed. D. Pines, West View Press (1991).
439: %\bibitem{white} S. R. White, Phys. Rev. Lett. {\bf 69}, 2863 (1992). Phys.
440: % Rev. {\bf B 48}, 10 345 (1993).
441: %\bibitem{wilson} K. G. Wilson, Rev. Mod. Phys. {\bf 47}, 773 (1975).
442: %\bibitem{sorensen} E. S. Sorensen and I. Affleck, Phys. Rev. Lett. {\bf 71},
443: % 1633 (1993).
444: %\bibitem{alvarez} J.V. Alvarez and S. Moukouri, Int. J. Mod. Phys.
445: % {\bf C 16}, 843 (2005).
446: %\bibitem{tsvelik} A. A. Nersesyan and A. M. Tsvelik
447: % Phys. Rev. B {\bf 67}, 024422 (2003); A. M. Tsvelik
448: %Phys. Rev. B {\bf 70}, 134412 (2004).
449: %\bibitem{starykh} O. A. Starykh and L. Balents Phys. Rev. Lett. {\bf 93},
450: %127202 (2004).
451: %\bibitem{sindzingre} A. Sindzingre, Phys. Rev. B {\bf 69}, 094418 (2004).
452: %\bibitem{reviewHTC} J.Orenstein and A.J. Millis, Science {\bf 288}, 468 (2000).
453: %\bibitem{reviewOC} C. Bourbonnais and D. J\'erome in "Advance in
454: %Synthetic Metals" Eds. P. Bernier, S. Lefrant and G. Bidan
455: % (Elsevier, New York), 206 (1999).
456: %\bibitem{reviewIOC} J.W. Allen, Sol. St. Comm. {\bf 123}, 469 (2002).
457: %\bibitem{anderson} P.W. Anderson, Science {\bf 288}, 480 (2000).
458: %\bibitem{kivelson0} S. Kivelson, Synthetic Metals {\bf 125}, 99 (2002).
459: %\bibitem{sachdev} S. Sachdev "Quantum Phase Transitions", Cambridge University
460: % Press (1999).
461: %\bibitem{anderson} P.W. Anderson, Phys. Rev. Lett. {\bf 64}, 1839 (1990);
462: % {\bf 65}, 2306 (1990).
463: %\bibitem{bourbonnais} C. Bourbonnais and L.G. Caron, Int. J. Mod. Phys.
464: % {\bf B 5}, 1033 (1991).
465: %\bibitem{shankar} R. Shankar, Rev. Mod. Phys. {\bf 66}, 129 (1994).
466: %\bibitem{carpentier} A. Vishwanath and D. Carpentier, Phys. Rev. Lett.
467: %{\bf 86}, 676 (2001).
468: %\bibitem{kivelson} V.J Emery, E. Fradkin, S.A. Kivelson, and T.C. Lubensky,
469: % Phys. Rev. Lett. {\bf 85}, 2160 (2000).
470: %\bibitem{kane} R. Mukhopadhyay, C.L. Kane, and T.C. Lubensky, Phys. Rev.
471: % {\bf B 64}, 045120 (2001).
472: %\bibitem{laughlin} R.B. Laughlin, Adv. Phys. {\bf 47}, 943 (1998).
473: %\bibitem{moukouri-KB} S. Moukouri. Phys. Lett. A {\bf 325} 177.
474: %\bibitem{moukouri-triang} S. Moukouri and J.V. Alvarez cond-mat, Phys. Lett.
475: %{\bf 344A}, 387 (2005).
476: %\bibitem{kato}T. Kato, Prog. Teor. Phys. {\bf 4}, 514 (1949); {\bf 5}, 95
477: % (1950).
478: %\bibitem{bloch} C. Bloch, Nucl. Phys. {\bf 6}, 329 (1958).
479: %\bibitem{parola} A. Parola, S. Sorella and Q.F. Zhong, Phys. Rev. Lett.
480: % {\bf 71}, 4393 (1993).
481: %\bibitem{sandvik} A. Sandvik, Phys. Rev. Lett. {\bf 83}, 3069 (1999).
482: %\bibitem{coldea} R. Coldea {\it et. al.}, Phys. Rev. Lett. {\bf 86},
483: % 1335 (2001); Phys. Rev. {\bf B 68}, 134424 (2003).
484: %\bibitem{wang} X. Wang, N. Zhu, C. Chen, Phys. Rev. {\bf B 66} 172405 (2002).
485: %\bibitem{affleck} I. Affleck and S. Qin, J. Phys. A {\bf 32}, 7815 (1999).
486: %\bibitem{villain} J. Villain, R. Bidaux, J.-P. Carton, and R. Conte,
487: %J. Phys (France) {\bf 41}, 1263 (1980).
488:
489: \end{thebibliography}
490:
491: \end{document}
492: