1: \documentclass{elsart}
2:
3: % Use the option doublespacing or reviewcopy to obtain double line spacing
4: % \documentclass[doublespacing]{elsart}
5:
6: % if you use PostScript figures in your article
7: % use the graphics package for simple commands
8: % \usepackage{graphics}
9: % or use the graphicx package for more complicated commands
10: % \usepackage{graphicx}
11: % or use the epsfig package if you prefer to use the old commands
12: % \usepackage{epsfig}
13:
14: % The amssymb package provides various useful mathematical symbols
15: \usepackage{amssymb}
16: \usepackage{graphicx}
17: %\usepackage{graphics}
18: \usepackage{hyperref}
19:
20: %\usepackage{epsf}
21: %\usepackage{showlabels}
22:
23:
24:
25: \def\crpar{\mbox{$\not\!\partial$}}
26: \def\crA{\mbox{$\not\!\!A$}}
27: \def\cra{\mbox{$\not\!a$}}
28: \def\crp{\mbox{$\not\!p$}}
29: \def\crx{\mbox{$\not\!x$}}
30: \def\tab{\mbox{}\hspace{1.0cm}}
31: \def\Tr{\mbox{Tr}\,}
32: \def\tr{\mbox{tr}\,}
33: \def\sgn{\,\mbox{sgn}\,}
34: \newcommand{\R} {\mbox{Re}\,}
35: \newcommand{\I} {\mbox{Im}\,}
36:
37:
38: \newcommand{\be}{\begin{equation}}
39: \newcommand{\ee}{\end{equation}}
40: \newcommand{\bea}{\begin{eqnarray}}
41: \newcommand{\eea}{\end{eqnarray}}
42: \newcommand{\eu}{{\rm e}}
43: \newcommand{\ii}{{\rm i}}
44: \newcommand{\de}{{\displaystyle\rm\mathstrut d}}
45: \newcommand{\Pf}{{\rm Pf}}
46:
47:
48: \begin{document}
49:
50: \begin{frontmatter}
51:
52: % Title, authors and addresses
53:
54: % use the thanksref command within \title, \author or \address for footnotes;
55: % use the corauthref command within \author for corresponding author footnotes;
56: % use the ead command for the email address,
57: % and the form \ead[url] for the home page:
58: % \title{Title\thanksref{label1}}
59: % \thanks[label1]{}
60: % \author{Name\corauthref{cor1}\thanksref{label2}}
61: % \ead{email address}
62: % \ead[url]{home page}
63: % \thanks[label2]{}
64: % \corauth[cor1]{}
65: % \address{Address\thanksref{label3}}
66: % \thanks[label3]{}
67:
68: \title{Emptiness Formation Probability for the Anisotropic XY Spin
69: Chain in a Magnetic Field}
70:
71: % use optional labels to link authors explicitly to addresses:
72: % \author[label1,label2]{}
73: % \address[label1]{}
74: % \address[label2]{}
75:
76: \author {Alexander G. Abanov, Fabio Franchini}
77: % \email{Alexandre.Abanov@sunysb.edu}
78: % \email{Fabio.Franchini@sunysb.edu}
79: \address{Physics \& Astronomy Department,}
80: \address{State University of New York,}
81: \address{Stony Brook, New York 11794-3800}
82:
83:
84:
85: \begin{abstract}
86: We study an asymptotic behavior of the probability of formation
87: of a ferromagnetic string (referred to as EFP) of length $n$ in
88: a ground state of the one-dimensional
89: anisotropic XY model in a transversal magnetic field as
90: $n\to \infty$.
91: We find that it is exponential everywhere in the phase diagram
92: of the XY model except at the critical lines where the spectrum
93: is gapless.
94: One of those lines corresponds to the isotropic XY model where
95: EFP decays in a Gaussian way, as was shown in
96: Ref.\cite{shiroishi}.
97: The other lines are at the critical value of the magnetic field.
98: There, we show that EFP is still exponential but acquires a
99: non-trivial power-law prefactor with a universal exponent.
100: \end{abstract}
101:
102: \begin{keyword}
103: % keywords here, in the form: keyword \sep keyword
104: Emptiness Formation Probability, Integrable Models, XY Spin Chain, Toeplitz Determinants, Fisher-Hartwig Conjecture
105: % PACS codes here, in the form: \PACS code \sep code
106: \PACS 75.10.Pq \sep 02.30.Ik \sep 02.30.Tb
107: \end{keyword}
108: \end{frontmatter}
109:
110: % main text
111: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
112: \section{Introduction}
113: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
114:
115:
116: Emptiness Formation Probability (EFP) is a special correlation function
117: which plays an important role in the theory of integrable models
118: \cite{korepin93}.
119: In the case of spin-$\frac{1}{2}$ chains at zero temperature it is
120: defined as
121: \be
122: P(n) \equiv
123: \left\langle \prod_{j=1}^n {1 - \sigma_j^z \over 2} \right\rangle
124: \label{EFPDef}
125: \ee
126: where $\sigma_i^{x,y,z}$ are Pauli matrices, the integer $j$ labels
127: the sites of the 1d lattice, and the average in (\ref{EFPDef}) is
128: performed over the ground state of the Hamiltonian of the spin chain.
129: EFP measures the probability that $n$ consecutive spin sites are
130: all found aligned downward in the ground state.
131: The asymptotic behavior of $P(n)$ at large $n$ is, probably, the most
132: interesting one from the physical point of view, as it is believed that
133: it measures the tendency of the system to a phase separation.
134:
135: There has been considerable progress in finding this asymptotic behavior
136: in the case of the integrable XXZ spin chain in critical regime
137: \cite{korepin93,korepin94,shiroishi,stroganov1,KMST_0-2002,KMST_gen-2002,KLNS-2002}.
138: At finite temperature the asymptotic behavior of EFP is exponential
139: \cite{korepin93} $P(n)\sim e^{-\gamma n}$, while at zero temperature in
140: critical regime it is Gaussian $P(n)\sim e^{-\alpha n^{2}}$.
141: The qualitative argument in favor of a Gaussian decay of EFP at zero
142: temperature was given in Ref. \cite{abanovkor} in the framework of a
143: field theoretical approach.
144: There, it was argued that an optimal fluctuation of the field
145: corresponding to EFP in the critical model will have (very roughly)
146: the form of a round droplet in space-time with area $\sim n^{2}$ and
147: will have an action proportional to $n^{2}$.
148: The asymptotic form of $P(n)$ is defined by the action of this optimal
149: configuration and is proportional to $P(n)\sim e^{-\alpha n^{2}}$,
150: which explains the Gaussian decay of EFP at zero temperature.
151: Similarly, at finite temperature $T$ in the limit $n\gg 1/T$ the droplet
152: becomes rectangular (one dimension $n$ is replaced by the inverse
153: temperature $1/T$) and the action cost is proportional to $n$,
154: giving $P(n)\sim e^{-\gamma n}$.
155: This argument is tied to the criticality of the theory.
156: Indeed, for noncritical theory one should not expect the same scaling of
157: time and space dimensions of an optimal configuration.
158: It is, therefore, very interesting to see how the asymptotic behavior
159: of EFP is related to the criticality of a theory.
160:
161: In this paper we examine the relation between an asymptotic behavior
162: of EFP and criticality using the example of the Anisotropic XY spin-1/2
163: chain in a transverse magnetic field $h$
164: \be
165: H = \sum_{j} \left[
166: \left( {1 + \gamma \over 2} \right) \sigma_j^x \sigma_{j+1}^x +
167: \left( {1 - \gamma \over 2} \right) \sigma_j^y \sigma_{j+1}^y \right]
168: - h \sum_{j} \sigma_j^z
169: \label{spinham}
170: \ee
171: This model has been solved in \cite{LSM-1961} in the case of zero
172: magnetic field and in \cite{mccoy} in the presence of a magnetic field.
173: The phase diagram is depicted in Fig. \ref{phasespace}.
174: The phase diagram of the XY model has obvious symmetries
175: $\gamma\to -\gamma$ and $h\to -h$.
176: However, the latter one is broken by EFP itself.
177: Therefore, we show only the part of the diagram corresponding to
178: $\gamma \ge 0$.
179: The phase diagram has both critical and non-critical regimes.
180: The first critical regime -- the Isotropic ($\gamma=0$) XY model --
181: is labeled $\Omega_0$.
182: There is another critical line $h=1$ which is labeled $\Omega_+$ where
183: excitations are gapless (and similarly line $\Omega_-$ at $h=-1$).
184: The rest of the phase diagram is non-critical with finite gap for
185: excitations and is divided into domains $\Sigma_+$, $\Sigma_0$, and
186: $\Sigma_-$ for $h>1$, $-1<h<1$, and $h<-1$ respectively.
187: Fig. \ref{phasespace} includes the line $\gamma=1$ ($\Gamma_I$)
188: corresponding to an Ising model in transverse magnetic field and the
189: line $\gamma^2+h^2=1$ ($\Gamma_E$) on which the wave function of the
190: ground state is factorized into a product of single spin states
191: \cite{shrock}.
192:
193: \begin{figure}
194: % \resizebox {\columnwidth}{!}{ \includegraphics{PhaseSpace} }
195: % \includegraphics[scale=0.5]{PhaseDiagramF}
196: \includegraphics[width=\columnwidth]{PhaseDiagramF}
197: \caption{Phase diagram of the XY Model. }
198: \label{phasespace}
199: \end{figure}
200:
201:
202:
203:
204: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
205: \section{The Emptiness Formation Probability in the XY Model}
206: \label{XYModel}
207: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
208:
209: It was shown in Refs. \cite{LSM-1961,mccoy} that spin-spin correlation
210: functions in the XY model can be represented as determinants of some
211: Toeplitz matrices so that one can find their asymptotic behavior using
212: known theorems from the theory of Toeplitz matrices.
213: This approach has been applied to EFP for the isotropic ($\gamma=0$) XY
214: model in magnetic field in Ref. \cite{shiroishi}.
215: In this section we (slightly) generalize the derivation of Ref.
216: \cite{shiroishi} to the case of $\gamma\neq 0$ and represent EFP as a
217: determinant of a Toeplitz matrix.
218:
219: We follow the standard prescription \cite{LSM-1961} and reformulate the
220: Hamiltonian (\ref{spinham}) in terms of spinless fermions $\psi_i$ by
221: means of a Jordan-Wigner transformation:
222: \bea
223: \sigma_j^+ & = &
224: \psi_j^\dagger \eu^{\ii \pi \sum_{k<i} \psi_k^\dagger \psi_k} =
225: \psi_j^\dagger \prod_{k<j}
226: \left( 2 \psi_k^\dagger \psi_k - 1 \right), \\
227: \sigma_j^z & = & 2 \psi_j^\dagger \psi_j - 1,
228: \eea
229: where, as usual, $\sigma^\pm = (\sigma^x \pm i \sigma^y)/2$.
230: In Fourier components $\psi_j = \sum_q \psi_q \eu^{\ii q j}$ we have:
231: \be
232: H = \sum_q \left[
233: 2 \left( \cos \! q - h \right) \psi_q^\dagger \psi_q
234: + \ii \gamma \sin \! q \: \psi_q^\dagger \psi_{-q}^\dagger
235: - \ii \gamma \sin \! q \: \psi_{-q} \psi_q \right].
236: \label{spinlessham}
237: \ee
238: The Bogoliubov transformation
239: \be
240: \chi_{q} = \cos \! {\vartheta_q \over 2} \: \psi_{q}
241: + \ii \sin \! {\vartheta_q \over 2} \: \psi_{-q}^\dagger
242: \label{bogtrans}
243: \ee
244: with ``rotation angle'' $\vartheta_q$
245: \be
246: e^{i\vartheta_q} = \frac{1}{\varepsilon_q}(\cos q-h+i\gamma\sin q),
247: \ee
248: brings the Hamiltonian (\ref{spinlessham}) to a diagonal form
249: $\sum_q \varepsilon_q \chi_q^\dagger\chi_q$ with the quasiparticle spectrum
250: \be
251: \varepsilon_q = \sqrt{ \left( \cos q - h \right)^2
252: + \gamma^2 \sin^2 q}.
253: \label{spectrum}
254: \ee
255: %\bea
256: % \sin \vartheta_q &=& \frac{\gamma\sin q}{\varepsilon_q},
257: % \nonumber \\
258: % \cos \vartheta_q &=& \frac{\cos q-h}{\varepsilon_q},
259: % \nonumber \\
260: % \varepsilon_q &=& \sqrt{ \left( \cos q - h \right)^2
261: % + \gamma^2 \sin^2 q}.
262: % \label{spectrum}
263: %\eea
264: The fermionic correlation functions are also easy to obtain from
265: (\ref{spinlessham}).
266: In the thermodynamic limit they read \cite{LSM-1961,mccoy}
267: \bea
268: F_{jk} &\equiv& \ii\langle \psi_j \psi_k \rangle
269: = - \ii\langle \psi_j^\dagger \psi_k^\dagger \rangle
270: = \int_{-\pi}^\pi {\de q \over 2\pi}\; \frac{1}{2}\sin\vartheta_q
271: \eu^{\ii q (j-k)}
272: \label{F}
273: \\
274: G_{jk} &\equiv& \langle \psi_j \psi_k^\dagger \rangle
275: = \int_{-\pi}^\pi {\de q \over 2 \pi}\; \frac{1+\cos\vartheta_q}{2}
276: \eu^{\ii q (j-k)} .
277: \label{G}
278: \eea
279:
280: We express EFP as the fermionic correlator \cite{shiroishi}
281: \be
282: P(n) = \langle \prod_{j=1}^n \psi_j \psi_j^\dagger \rangle.
283: \label{expect}
284: \ee
285: and using Wick's theorem on the r.h.s of (\ref{expect}), we obtain
286: \be
287: P(n) = \Pf( {\bf M} ) \equiv \sqrt{\det( {\bf M} )},
288: \ee
289: where $\Pf( {\bf M} )$ is the {\it Pfaffian} of the $2n \times 2n$
290: skew-symmetric matrix ${\bf M}$ of correlation functions
291: \be
292: {\bf M} = \pmatrix{ {-\ii\bf F} &
293: {\bf G} \cr
294: - {\bf G} &
295: \ii{\bf F} \cr
296: }.
297: \ee
298: Here ${\bf F}$ and ${\bf G}$ are $n\times n$ matrices with matrix
299: elements given by $F_{jk}$ and $G_{jk}$ from (\ref{F},\ref{G})
300: respectively.
301:
302: We perform a unitary transformation
303: \be
304: {\bf M'} = {\bf U M U}^{\dagger}
305: = \pmatrix { 0 & {\bf S_n} \cr -{\bf S_n}^\dagger & 0 \cr}, \qquad
306: {\bf U} = {1 \over \sqrt{2}} \pmatrix { {\bf I} & {\bf -I} \cr
307: {\bf I} & {\bf I} \cr},
308: \ee
309: where ${\bf I}$ is the $n\times n$ unit matrix and ${\bf S_n}={\bf
310: G}+\ii{\bf F}$ and ${\bf S_n}^\dagger={\bf G}-\ii{\bf F}$. This allows
311: us to calculate the determinant of ${\bf M}$ as
312: \be
313: \det ( {\bf M} ) = \det ( {\bf M'} ) =
314: \det ( {\bf S_n} ) \cdot \det ( {\bf S_n}^\dagger )
315: = \left|\det ( {\bf S_n} ) \right|^2.
316: \label{detM}
317: \ee
318: The matrix ${\bf S_n}$ is a $n\times n$ Toeplitz matrix (its matrix
319: elements depend only on the difference of raw and column
320: indices \cite{basor}).
321: The generating function $\sigma(q)$ of a Toeplitz matrix is defined by
322: \be
323: ({\bf S_n})_{jk}=\int_{-\pi}^\pi {d q \over 2 \pi}\; \sigma (q)
324: \eu^{\ii q(j-k)}
325: \label{Tg}
326: \ee
327: and can be found from (\ref{F},\ref{G}) as
328: \be
329: \sigma(q) = {1 \over 2}\left( 1 + \eu^{\ii\vartheta_q} \right)
330: = \frac{1}{2}
331: +\frac{\cos q -h +i\gamma\sin q}{2\sqrt{(\cos q-h)^2+\gamma^2\sin^2q}}.
332: \label{genfunc}
333: \ee
334:
335: Thus, the problem of calculation of EFP
336: \be
337: P(n) = \left|\det ( {\bf S_n} ) \right|,
338: \label{PnX}
339: \ee
340: is reduced (exactly) to the calculation of the determinant of the
341: $n\times n$ Toeplitz matrix ${\bf S_n}$ defined by the generating
342: function (\ref{Tg},\ref{genfunc}).
343: The representation (\ref{PnX}) is exact and valid for any $n$.
344: In our study we are interested in finding an asymptotic behavior at
345: large $n \to \infty$.
346:
347: In the following section we apply known theorems on the asymptotic
348: behavior of Toeplitz determinants to extract the corresponding
349: asymptotes of $P(n)$ at $n\to \infty$.
350: The reader who is interested in results can skip the following
351: section and go directly to Sec.\ref{Conclusions} where we present
352: and discuss our results.
353:
354:
355:
356:
357: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
358: \section{The results for the asymptotic behavior of EFP}
359: \label{results}
360: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
361:
362: The asymptotic behavior of EFP at $n\to\infty$ is exactly related to the
363: asymptotic behavior of the determinant of the corresponding Toeplitz
364: matrix (\ref{Tg},\ref{genfunc},\ref{PnX}) and can be extracted from known
365: theorems and conjectures in the theory of Toeplitz matrices.
366: These types of calculations have been done first in
367: \cite{LSM-1961,mccoy} for spin-spin correlation functions.
368: It is well known that the asymptotic behavior of the determinant of a
369: Toeplitz matrix as the size of the matrix goes to infinity strongly
370: depends upon the zeros of the generating function of the matrix and
371: upon its phase.
372:
373: In the case when the generating function $\sigma(q)$ has only isolated
374: singularities (zeros and phase discontinuities) the most general
375: known result for the asymptotic behavior of a
376: Toeplitz determinant is called {\it generalized Fisher-Hartwig conjecture}
377: (gFH) and is due to Basor and Tracy \cite{basor}.
378: The result can be written
379: as
380: \be
381: P(n) \sim \sum_{i} E_{i} n^{-\lambda_{i}} e^{-\beta_{i} n},
382: \label{FHsymb}
383: \ee
384: where the decay rates $\beta_{i}$, the exponents $\lambda_{i}$, and the
385: coefficients $E_{i}$ are known functionals of $\sigma(q)$ \cite{basor}.
386: The original {\it Fisher-Hartwig conjecture} (FH)
387: \cite{FisherHartwig-1968} consists of just one term of the sum in
388: (\ref{FHsymb}). We apply gFH to the generating function
389: (\ref{genfunc}) and extract the asymptotic behavior of $P(n)$.
390: All necessary theorems on the asymptotic behavior of Toeplitz
391: determinants can be found in
392: Ref.\cite{wu,FisherHartwig-1968,widomsing,basor,Ehrhardt-1997,widomsupp}.
393: For the most recent review of gFH see also Ref.\cite{Ehrhardt-2001}.
394:
395:
396: In some parts of the phase diagram we have to rely on ``conjectures'',
397: i.e., results which are not proven yet.
398: Therefore, it is prudent to look for some independent ways of checking
399: those results.
400: We did some numeric calculations on these Toeplitz determinants and the
401: results of these calculations support the analytic results presented
402: below \cite{AF}.
403: Also, we used some limiting cases where EFP or an asymptotic behavior
404: of Toeplitz determinant can be obtained in an independent way to
405: confirm our conclusions.
406: E.g., on the line $\Gamma_E$ EFP can be calculated exactly and the
407: asymptotic behavior of the result is the same as the one obtained
408: using FH for the whole region $\Sigma_0$ of the phase diagram.
409:
410: In the rest of this section we present the results for the asymptotic
411: behavior of $P(n)$ in different regions of the phase diagram of the XY
412: model (Fig. \ref{phasespace}).
413: In this letter we concentrate on the calculation of $\lambda$ and
414: $\beta(h,\gamma)$ leaving our findings for the coefficient
415: $E(h,\gamma)$ as well as the results of our numeric calculations for a
416: future, more extended publication \cite{AF}.
417:
418:
419:
420: \subsection{The non-critical theory ($\Sigma_\pm$ and $\Sigma_0$)}
421: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
422:
423:
424: %%%%%%%%%%%%%%%%%%
425: \subsubsection{The region $\Sigma_-$}
426:
427:
428: In this region ($\gamma\neq 0$,$h<-1$) the generating function
429: (\ref{genfunc}) is nonzero for all $q$ and the FH conjecture is
430: reduced to a (rigorously proven) {\it Strong Szeg\"o Limit Theorem}.
431: It gives
432: \be
433: P(n)\sim E(h,\gamma) e^{-\beta(h,\gamma) n}
434: \label{pnc}
435: \ee
436: with
437: \bea
438: \beta(h, \gamma) & = &
439: - \int_{-\pi}^\pi {\de q \over 2\pi}\; \log \left| \sigma(q) \right|
440: \nonumber \\ & = &
441: - \int_0^\pi {\de q \over 2\pi}\;
442: \log \left[ {1 \over 2} \left( 1 + {\cos q -h \over \sqrt{
443: \left( \cos q - h \right)^2 + \gamma^2 \sin^2 q} } \right) \right].
444: \label{betagh}
445: \eea
446: The integral in (\ref{betagh}) is convergent for all $h$ and all
447: $\gamma\neq 0$ and $\beta(h,\gamma)$ is a continuous function of its
448: parameters.
449: In fact, it turns out (see below) that this expression for
450: $\beta(h,\gamma)$ describes correctly the decay rate of EFP everywhere
451: in the phase diagram except for $\gamma=0$.
452: However, both pre-exponential factors and the mathematical status of
453: the results are quite different in different parts of the phase diagram
454: (see below).
455:
456: In Figure \ref{betagraph}, $\beta(h,\gamma)$ is plotted as a function
457: of $h$ at several values of $\gamma$.
458: One can see that $\beta(h,\gamma)$ is continuous but has a weak
459: (logarithmic) singularity at $h=\pm 1$.
460: This is obviously one of the effects of the criticality of the model on
461: the asymptotic behavior of EFP.
462:
463: \begin{figure}
464: % \resizebox {\columnwidth}{!}{ \includegraphics{BetaGraph} }
465: % \includegraphics[scale=1.2]{BetaGraph}
466: \includegraphics[width=\columnwidth]{BetaGraph}
467:
468: \caption{Plot of the decay factor $\beta$ as a function of the
469: parameters $\gamma$ and $h$. The function diverges for $\gamma = 0$
470: and is continuous for $h = \pm 1$ (although has a weak singularity at $h=\pm 1$).}
471:
472: \label{betagraph}
473: \end{figure}
474:
475:
476: %%%%%%%%%%%%%%%%%%
477: \subsubsection{The region $\Sigma_0$}
478:
479:
480: In $\Sigma_0$ ($\gamma\neq 0$,$-1<h<1$) the generating function
481: $\sigma(q)$ vanishes at $q = \pm \pi$ and acquires an overall phase of
482: $\pi$ as $q$ goes from $-\pi$ to $\pi$.
483: The Szeg\"o theorem is not applicable anymore.
484: One has to rely on the FH conjecture (which, in fact is proven for this
485: case -- the case of one singularity \cite{Ehrhardt-1997}) to derive the
486: result, which is identical to (\ref{pnc},\ref{betagh}).
487: The expression for the coefficient $E(h,\gamma)$ is different from the
488: one in $\Sigma_-$ but its form is beyond the scope of this paper.
489:
490:
491: %%%%%%%%%%%%%%%%%%
492: \subsubsection{The region $\Sigma_+$}
493:
494:
495: In $\Sigma_+$ ($\gamma\neq 0$, $h>1$) the generating function $\sigma(q)$
496: vanishes at the points $q = 0$ and
497: $q = \pm \pi$, and its phase has jumps at these points.
498: In this case an application of FH leads to some ambiguity, e.g., one
499: obtains two values for $\beta(h,\gamma)$: $\beta_1=\beta$ and
500: $\beta_2=\beta+\pi i$ with $\beta$ from (\ref{betagh}).
501: This ambiguity is resolved in the use of the gFH which gives EFP as a
502: sum of two terms so that both values of $\beta$ are used.
503:
504: We obtain
505: \be
506: P(n) \sim {\it E}(h,\gamma)
507: \left[ 1 + (-1)^n A(h,\gamma) \right] \: \eu^{- n \beta(h, \gamma)}.
508: \label{oscbehavior}
509: \ee
510: Thus, in this region we again have an exponential asymptotic decay of
511: EFP but, the constant prefactor is different for even and odd $n$.
512:
513:
514:
515:
516: %%%%%%%%%%%%%%%%%%
517: \subsubsection{The line $\Gamma_2$: an exact calculation}
518:
519:
520:
521: Before proceeding to the analysis of EFP on the critical lines let us
522: check our results (\ref{pnc},\ref{betagh}) on the special line\footnote{We
523: are grateful to Fabian Essler who suggested us to check our results
524: on this special line and pointed out the reference \cite{shrock} to
525: us.} in the phase diagram defined by
526: \be
527: h^2 + \gamma^2 = 1.
528: \label{trajectory}
529: \ee
530: It was shown in Ref.\cite{shrock} that on this line the ground state
531: is a product of single spin states and is given by
532: \be
533: \vert G \rangle = \prod_j \vert \theta, j \rangle = \prod_j
534: \left[ \cos \left( {\theta \over 2} \right) | \uparrow, j \rangle +
535: (-1)^j \sin \left( {\theta \over 2} \right) | \downarrow, j \rangle
536: \right],
537: \label{groundstate}
538: \ee
539:
540: where $| \uparrow, j \rangle$ is an up-spin state at the lattice
541: site $j$ etc.
542: One can directly check that the state
543: (\ref{groundstate}) is an eigenstate of (\ref{spinham}) if the value
544: of the parameter $\theta$ is
545: \be
546: \cos^2 \theta = \frac{1-\gamma}{1 + \gamma}
547: \ee
548: and (\ref{trajectory}) is satisfied. It is also easy to
549: show \cite{shrock} that this state is, in fact, the ground state of
550: (\ref{spinham}).
551:
552: The probability of formation of a ferromagnetic string in the state
553: (\ref{groundstate}) is obviously
554: \be
555: P(n) = \sin^{2n} \left( {\theta \over 2} \right)
556: = \left( {1 \over 2} - {1 \over 2}
557: \sqrt{ 1 - \gamma \over {1 +\gamma} }
558: \right)^n
559: \label{pnexact}
560: \ee
561: which is an exact result on the line (\ref{trajectory}).
562: The value of $\beta(h,\gamma)$ which immediately follows from this
563: exact result is
564: \be
565: \beta(h=\sqrt{1-\gamma^2},\gamma) =
566: - \log\left( {1 \over 2} - {1 \over 2}
567: \sqrt{ 1 - \gamma \over {1 +\gamma} } \right)
568: \label{betaexact}
569: \ee
570: Indeed, one can easily check that the integral in the general formula
571: for the decay rate (\ref{betagh}) under the condition
572: (\ref{trajectory}) gives precisely (\ref{betaexact}).
573: In fact, the Toeplitz matrix generated by (\ref{genfunc}) becomes
574: triangular on the line (\ref{trajectory}) with diagonal matrix
575: element $(S_n)_{jj} = \sin^2(\theta/2)$ and the determinant of
576: ${\bf S_n}$ is exactly (\ref{pnexact}).
577:
578:
579:
580: \subsection{The critical line $\Omega_0$ ($\gamma = 0$)}
581: \label{gammazero}
582: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
583:
584: The case $\gamma = 0$, corresponding to the Isotropic XY Model, has
585: been extensively studied in Ref.\cite{shiroishi}.
586: For $\gamma=0$ the generating functions (\ref{genfunc}) is reduced
587: to the one found in \cite{shiroishi}.
588:
589: For $h < 1$, the generating function $\sigma(q)$ has a limited
590: support: to find the asymptotic behavior of the determinant of the
591: Toeplitz matrix one can apply Widom's Theorem \cite{widomsupp} and
592: obtain \cite{shiroishi}:
593: \be
594: P(n) \sim
595: 2^{5 \over 24} \eu^{3 \zeta'(-1)} (1-h)^{- {1 \over 8}}
596: n^{- {1 \over 4}} \left( {1 + h \over 2} \right)^{n^2 \over 2}
597: \label{pnsh}
598: \ee
599: and we see that in this case, EFP decay as a Gaussian with an
600: additional power-law pre-factor.\footnote{Notice, that the formula
601: (\ref{pnsh}) was in fact obtained in \cite{dysmehta} in the context
602: of unitary random matrices (which are known to be equivalent to free
603: fermions).}
604:
605: For $|h| > 1$, the theory in not critical anymore and the ground state
606: is completely polarized in $z$ direction, giving a trivial EFP $P(n)=0$
607: for $h>1$ and $P(n)=1$ for $h<-1$.
608:
609:
610:
611: \subsection{The critical lines $\Omega_\pm$($h = \pm 1$)}
612: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
613:
614:
615: %%%%%%%%%%%%%%%%%%
616: \subsubsection{The critical line $\Omega_+$($h = 1$)}
617: \label{ssOmegap}
618:
619: For $h=1$ the generating function $\sigma(q)$ vanishes at $q = \pm \pi$
620: and its phase has jumps at $q = 0,\pi$.
621: The gFH in this case produces many terms of the form (\ref{FHsymb}).
622: However, in contrast to the $\Sigma_+$ region, all terms are suppressed
623: by power law factors of $n$ with respect to the leading one.
624: The leading term has $\beta$ from (\ref{betagh}) and critical exponent
625: $\lambda = {1\over 16}$:
626: \be
627: P(n) \sim E(\gamma) n^{- {1 \over 16} } \: \eu^{-n \beta(1, \gamma)}.
628: \nonumber
629: \ee
630: The next subleading term (faster decay at $n\to \infty$) is given by
631: \be
632: E'(\gamma) (-1)^n n^{- {9 \over 16} } \: \eu^{-n \beta(1, \gamma)}.
633: \nonumber
634: \ee
635: Although inclusion of the latter (subleading) term is somewhat beyond
636: even the generalized FH, we write the sum of these two terms as a
637: conjecture for EFP at $h=1$ and obtain
638: \be
639: P(n) \sim {\it E}(\gamma) \:
640: n^{- {1 \over 16} } \left[ 1 + (-1)^n A(\gamma)/n^{1 \over 2}
641: \right] \: \eu^{- n \beta(1, \gamma)}.
642: \label{pnomp}
643: \ee
644: As this result relies on this unproven conjecture, we present our numeric
645: results for this case in Figure \ref{DetPlot}.
646: Indeed, we see that the form (\ref{pnomp}) is in good agreement with numerics.
647: We conclude that at $h=1$ EFP decays exponentially at $n\to\infty$ but
648: with an additional power law pre-factor and a damped oscillatory component.
649:
650: \begin{figure}
651: % \resizebox {\columnwidth}{!}{ \includegraphics{BetaGraph} }
652: % \includegraphics[scale=1.2]{DetPlot}
653: \includegraphics[width=\columnwidth]{DetPlot}
654:
655: \caption{Results of the numeric calculation of the Toeplitz determinant
656: is shown as $P(n) e^{\beta n} n^{1/16}$ vs $n$ at $\gamma =1$, $h=1$.
657: The value of $\beta=\log 2 +2G/\pi$ with Catalan's constant $G$ is obtained
658: from (\ref{betagh}).
659: The solid line is the analytic result $E(1+(-1)^n A/n^{1 \over 2})$ with
660: $E=0.8065...$, $A=0.2399...$ obtained by fitting at large $n$. To make the
661: plot more readable we show only every 11th point (for $n=1,12,23,\ldots$).}
662:
663: \label{DetPlot}
664: \end{figure}
665:
666: {\it Remark.} It is curious to notice that $1/16$ and $9/16$ remind of the scaling
667: dimensions of spins $\sigma^x$ and $\sigma^y$.\footnote{See Ref.\cite{mccoy}
668: where it was shown that power laws are $1/4$ and $9/4$ for $\sigma^x$ and
669: $\sigma^y$ correlators respectively.}
670: It looks as if EFP operator (\ref{EFPDef}) among other things has inserted
671: square roots of spins at the ends of the string.
672:
673:
674:
675: %%%%%%%%%%%%%%%%%%
676: \subsubsection{The critical line $\Omega_-$($h = -1$)}
677:
678: For $h=-1$ the generating function $\sigma(q)$ does not vanish but has a
679: phase discontinuity at $q=\pm \pi$.
680: We apply FH to obtain
681: \be
682: P(n) \sim {\it E}(\gamma) \:
683: n^{- {1 \over 16} } \: \eu^{- n \beta(-1, \gamma)}.
684: \ee
685: and stretching gFH the same way as we did for $h=+1$ we include a subleading term
686: \be
687: P(n) \sim {\it E}(\gamma) \:
688: n^{- {1 \over 16} } \left[ 1 + A(\gamma)/n^{1 \over 2}
689: \right] \: \eu^{- n \beta(-1, \gamma)}.
690: \label{pnomm}
691: \ee
692: This form is also in good agreement with our numeric results.
693:
694:
695:
696: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
697: \section{Discussion and Conclusions}
698: \label{Conclusions}
699: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
700:
701: An asymptotic behavior of Emptiness Formation Probability $P(n)$ as
702: $n\to\infty$ for the Anisotropic XY model in a transverse magnetic field
703: as a function of the anisotropy $\gamma$ and the magnetic field $h$
704: has been studied.
705: We summarize our results in Table \ref{table1}.
706: All these results are new except for the isotropic case $\Omega_0$
707: ($\gamma=0$) which was studied in Ref.\cite{shiroishi}.
708:
709: \begin{table}
710: \caption{Asymptotic behavior of EFP in the different regimes.
711: The exponential decay rate $\beta$ is given by Eq. (\ref{betagh})
712: for all regimes.
713: The coefficients $E,A$ are functions of $h$ and $\gamma$ as appropriate.}
714: \centering
715: \label{table1}
716: \begin{tabular}{|c|c|c|l|l|l|}
717: \hline
718: \multicolumn{6}{|c|}{\bfseries EFP for an Anisotropic XY model} \\
719: \hline
720: Region & $\gamma$, $h$ & criticality & $P(n)$ & Eq.\# & Theorem \\
721: \hline
722: \hline
723: $\Sigma_-$ & $h<-1$ & No & $ E \eu^{-n \beta}$ & \ref{pnc} & Szeg\"o \\
724: \hline
725: $\Sigma_0$ & $-1<h<1$ & No & $ E \eu^{-n \beta }$ & \ref{pnc} & FH \\
726: \hline
727: $\Sigma_+$ & $h>1$ & No &
728: $ {\it E} \left[ 1 + (-1)^n A \right] \: \eu^{-n \beta}$ & \ref{oscbehavior} & gFH \\
729: \hline
730: $\Gamma_2$ & $\gamma^2+h^2=1$ & No & $ E \eu^{-n \beta}$ & \ref{pnexact} & Exact \\
731: \hline
732: \hline
733: $\Omega_+$ & $h=1$ & Yes &
734: $ {\it E} \: n^{- {1 / 16} } \left[ 1 + (-1)^n A/\sqrt{n} \: \right] \: \eu^{- n \beta}$ & \ref{pnomp} & gFH \\
735: \hline
736: $\Omega_-$ & $h=-1$ & Yes & $ {\it E} \: n^{- {1 / 16} } \left[ 1 + A/\sqrt{n} \: \right] \: \eu^{- n \beta}$ & \ref{pnomm} & gFH \\
737: \hline
738: $\Omega_0$ & $\gamma=0$,\, $-1<h<1$ & Yes & $ E n^{-1/4}e^{-n^2 \alpha}$ & \ref{pnsh} & Widom \\
739: \hline
740: \end{tabular}
741: \end{table}
742:
743: Our main motivation was to study the relation between the criticality of
744: the theory and the asymptotics of EFP.
745: Let us look at the results on the critical lines $\Omega_0$ and $\Omega_\pm$.
746: The Gaussian behavior on $\Omega_0$ is in accord with the qualitative argument
747: of Ref.\cite{abanovkor}.
748: However, on $\Omega_\pm$ the decay of EFP is exponential instead of Gaussian,
749: and apparently contradicts the qualitative picture of Ref.\cite{abanovkor}
750: (which is reproduced in the introduction to this paper for the sake of completeness).
751: The reason for this disagreement is that at $h=\pm 1$ the model is critical for the
752: {\it quasiparticles} $\chi$ defined in (\ref{bogtrans}).
753: However, we are interested in EFP for the ``original'' Jordan-Wigner fermions $\psi$,
754: and this correlator has clearly a different expression in terms of $\chi$ than the
755: simple one (\ref{expect}) which holds at $\gamma =0$.
756: From the technical point of view, the difference is that in the qualitative argument
757: in favor of a Gaussian decay of EFP for critical systems there is an implicit
758: assumption that the density of fermions (or magnetization) is related in a local way
759: to the field responsible for critical degrees of freedom (free boson field $\phi$).
760: This assumption is not valid on the lines $h=\pm 1$.
761: Although the theory is critical on those lines and can be described by some
762: free field $\phi$, the relation between the magnetization and this field is
763: highly nonlocal and one can not apply the simple argument of \cite{abanovkor}
764: to the XY model at $h=\pm 1$.
765:
766: Although EFP at the critical magnetic field does not show a Gaussian behavior,
767: there is an important difference between the asymptotic behavior
768: of EFP on and off the critical lines.
769: Namely, a power-law pre-factor appears on all critical lines.
770: For the XY model it is universal (constant on a given critical line) and takes
771: values $\lambda=1/4$ for $\gamma=0$ \cite{shiroishi} and $\lambda =1/16$ on the
772: lines $h=\pm 1$. It would be interesting to understand which operators determine
773: these particular values of ``scaling dimensions'' of EFP
774: (see the remark at the end of section \ref{ssOmegap}).
775:
776: In some parts of the phase diagram we used the so-called {\it generalized
777: Fisher-Hartwig conjecture} \cite{basor} which is not proven yet.
778: However, our numeric calculations support our analytical results, see, e.g.,
779: Figure \ref{DetPlot}.
780: We note that to the best of our knowledge this is the first physically
781: motivated example where the original Fisher-Hartwig conjecture fails and its
782: extended version is necessary.
783: We also suggested that the gFH could be used to find the subleading corrections
784: to the asymptotic behavior, as we did for $h = \pm 1$ in (\ref{pnomp},\ref{pnomm}).
785: From the physical point of view, the oscillations of EFP that we obtain for
786: $h \ge 1$ from the gFH are related to a spontaneous $Z_2$ symmetry breaking which
787: occurs at $h=1$.
788:
789: In conclusion, we notice that it is straightforward to generalize our results for
790: nonzero temperature.
791: The only modification is that at $T\neq 0$ the thermal correlation functions must
792: be used instead of (\ref{F},\ref{G}).
793: Then, the generating function (\ref{genfunc}) is non-singular everywhere and we
794: have an exponential decay of $P(n)$ in the whole phase diagram according to the
795: standard Szeg\"o Theorem.
796:
797:
798:
799: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
800: \section{Acknowledgments}
801: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
802:
803: We greatly benefited from multiple discussions with F. Essler, V.E. Korepin,
804: and B. McCoy.
805: AGA would like to acknowledge the financial support from Alfred P. Sloan foundation.
806:
807: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
808: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
809: \begin{thebibliography}{00}
810:
811: % \bibitem{label}
812: % Text of bibliographic item
813:
814: % notes:
815: % \bibitem{label} \note
816:
817: % subbibitems:
818: % \begin{subbibitems}{label}
819: % \bibitem{label1}
820: % \bibitem{label2}
821: % If there is a note, it should come last:
822: % \bibitem{label3} \note
823: % \end{subbibitems}
824:
825: \bibitem{shiroishi}
826: M. Shiroishi, M. Takahashi, and Y. Nishiyama, J. Phys. Soc. Jap.
827: {\bf 70}, 3535 (2001).
828: \\ {\it Emptiness Formation Probability for the One-Dimensional
829: Isotropic XY Model.}
830:
831: \bibitem{korepin93}
832: V.E. Korepin, N.M. Bogoliubov, and A.G. Izergin, {\it Quantum
833: Inverse Scattering Mehod and Correlation Functions}, Cambridge
834: University Press, Cambridge, UK, 1993.
835:
836: \bibitem{korepin94}
837: V.E. Korepin, A.G. Izergin, F.H.L. Essler, and D.B. Uglov, Phys.
838: Lett. {\bf A 190}, 182 (1994).
839: \\ {\it Correlation functions of the spin-1/2 XXX antiferromagnet.}
840:
841: \bibitem{stroganov1}
842: Yu. Stroganov, J. Phys. A-- Math. Gen. {\bf 34}, L179 (2001).
843: \\ {\it The Importance of being Odd.}
844: \\
845: A. V. Razumov, and Yu. G. Stroganov, J. Phys. A-- Math. Gen. {\bf
846: 34}, 3185 (2001).
847: \\ {\it Spin chains and combinatorics.}
848:
849: \bibitem{KMST_0-2002}
850: N. Kitanine, J.M. Maillet, N.A. Slavnov and V. Terras, J.
851: Nucl. Phys. {\bf B642}, 433-455 (2002).
852: \\ {\it Correlation functions of the XXZ spin-$\frac{1}{2}$
853: Heisenberg chain at free fermion point from their multiple integral
854: represenations}
855: \\
856: N. Kitanine, J.M. Maillet, N.A. Slavnov and V. Terras, J.
857: Phys. A: Math. Gen. {\bf 35}, L385-L388 (2002).
858: \\ {\it Emptiness formation probability of the XXZ spin-$\frac{1}{2}$
859: Heisenberg chain at $\Delta=\frac{1}{2}$.}
860:
861: \bibitem{KMST_gen-2002}
862: N. Kitanine, J.M. Maillet, N.A. Slavnov, and V. Terras, J. Phys. A:
863: Math. Gen. \textbf{35}, L753-L758 (2002).
864: \\ {\it Large distance asymptotic behaviour of the emptiness
865: formation probability of the XXZ spin-$\frac{1}{2}$ Heisenberg chain}
866:
867: \bibitem{KLNS-2002}
868: V.E. Korepin, S. Lukyanov, Y. Nishiyama and M. Shiroishi,
869: to be published in Phys. Lett. A,
870: arXiv:cond-mat/0210140.
871: \\ {\it Asymptotic Behavior of the Emptiness Formation Probability in
872: the Critical Phase of XXZ Spin Chain.}
873:
874: \bibitem{abanovkor}
875: A.G. Abanov and V.E. Korepin, Nucl. Phys. {\bf B 647}, 565,
876: (2002).
877: \\ {\it On the probability of ferromagnetic strings in
878: antiferromagnetic spin chains.}
879:
880: \bibitem{LSM-1961}
881: E. Lieb, T. Schultz, and D. Mattis, Ann. of Phys. \textbf{16}, 407-466
882: (1961)
883: \\ {\it Two Soluble Models of an Antiferromagnetic Chain}
884:
885: \bibitem{mccoy}
886: % E. Barouch, B.M. McCoy, and M. Dresden, Phys. Rev. {\bf A 2}, 1075
887: % (1970).
888: %\\ {\it Statistical Mechanics of the XY Model. I}.
889:
890: E. Barouch, and B.M. McCoy, Phys. Rev. {\bf A 3}, 786 (1971).
891: \\ {\it Statistical Mechanics of the XY Model. II. Spin-Correlation
892: Functions}.
893:
894: % E. Barouch, and B.M. McCoy, Phys. Rev. {\bf A 3}, 2137 (1971).
895: %\\ {\it Statistical Mechanics of the XY Model. III}.
896:
897: % B.M. McCoy, E. Barouch, and D.B. Abraham, Phys. Rev. {\bf A 4}, 2331
898: % (1971).
899: %\\ {\it Statistical Mechanics of the XY Model. IV. Time-Dependent
900: %Spin-Correlation Functions}.
901:
902: \bibitem{shrock}
903: J. Kurmann, H. Thomas, and G. M\"uller, Physica A {\bf 112}, 235 (1982).
904: \\ {\it Antiferromagnetic Long-Range Order in the Anisotropic Quantum Spin Chain.}
905: \\
906: G. M\"uller, and R.E. Shrock, Phys. Rev. {\bf B 32}, 5845 (1985).
907: \\ {\it Implications of direct-product ground states in the
908: one-dimensional quantum XYZ and XY spin chains.}
909:
910: \bibitem{wu}
911: T.T. Wu, Phys. Rev. {\bf 149}, 380 (1966).
912: \\ {\it Theory of Toeplitz Determinants and the Spin Correlations of the Two-Dimensional Ising Model. I}
913:
914: \bibitem{FisherHartwig-1968}
915: M.E. Fisher and R.E. Hartwig, Adv. Chem. Phys. {\bf 15}, 333 (1968).
916: \\ {\it Toeplitz determinants, some applications, theorems and conjectures.}
917:
918: \bibitem{widomsing}
919: H. Widom, Amer. J. Math. {\bf 95}, 333 (1973).
920: \\ {\it Toeplitz Determinants with Singular Generating Functions.}
921:
922: \bibitem{basor}
923: E.L. Basor, and C.A. Tracy, Phys. {\bf A 177}, 167 (1991).
924: \\ {\it The Fisher-Hartwig conjecture and generalizations.}
925: \\
926: E.L. Basor, and K.E. Morrison, Lin. Alg. App. {\bf 202}, 129 (1994).
927: \\ {\it The Fisher-Hartwig Conjecture and Toeplitz Eigenvalues.}
928:
929: \bibitem{widomsupp}
930: H. Widom, Ind. Univ. Math. J. {\bf 21}, 277 (1971).
931: \\ {\it The Strong Szeg\"o Limit Theorem for Circular Arcs.}
932:
933: \bibitem{Ehrhardt-1997}
934: T. Ehrhardt and B. Silbermann, J. Funct. Anal. {\bf 148}, 229-256 (1997).
935: \\ {\it Toeplitz Determinants with One Fisher-Hartwig Singularity.}
936:
937: \bibitem{Ehrhardt-2001}
938: T. Ehrhardt, Operator Th: Advances and App. {\bf 124}, 217-241 (2001).
939: \\ {\it A status report on the asymptotic behavior of Toeplitz determinants with Fisher-Hartwig singularities.}
940:
941: %\bibitem{Hirschman}
942: % I.I. Hirschman, Jr., Amer. J. Math. {\bf 88}, 577 (1966).
943: %\\ {\it The Strong Szeg\"o Limit Theorem for Toeplitz Determiants.}
944:
945:
946: \bibitem{AF}
947: A.G. Abanov and F. Franchini, to be published.
948:
949:
950: \bibitem{dysmehta}
951: J.des Cloizeaux and M.L. Mehta, J. Math. Phys. {\bf 14}, 1648 (1973).
952: \\ {\it Asymptotic Behavior of Spacing Distributions for Eigenvalues of Random Matrices.}
953: \\
954: F. Dyson, Commun. Math. Phys. {\bf 47}, 171 (1976).
955: \\ {\it Fredholm Determinants and Inverse Scattering Problems.}
956:
957:
958:
959:
960: \end{thebibliography}
961: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
962:
963:
964:
965: \end{document}
966:
967: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
968: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
969: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
970: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
971: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
972: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
973: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
974: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
975: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
976:
977:
978: \bibitem{bethe}
979: H. Bethe, Zeitschrift f\"ur Physik, {\bf 71}, 205 (1931).
980: \\ {\it Zur Theorie der Metalle: I. Eingenwerte and Eigenfunktionen
981: der linearen Atomkette.}
982:
983: \bibitem{yangyang}
984: C.N. Yang, and C.P. Yang, Phys. Rev. {\bf 150}, 327, (1966).
985: \\ {\it One-Dimensional Chain of Anisotropic Spin-Spin Interactions.
986: II. Properties of the Ground-State Energy Per Lattice Site for an
987: Infinite System.}
988:
989: \bibitem{jimbo}
990: M. Jimbo, and T. Miwa, {\it Algebraic analysis of solvable lattice
991: models.}, no 85, Providence : American Mathematical Society, 1995.
992:
993: \bibitem{korepin2002}
994: H.E. Boos, V.E. Korepin, Y. Nishiyama, and M. Shiroishi, J.Phys.
995: {\bf A 35}, 4443 (2002).
996: \\ {\it Quantum correlations and number theory.}
997:
998: \bibitem{sachdev}
999: S. Sachdev, {\it Quantum Phase Transitions.}, Cambridge University
1000: Press, Cambridge, UK, 1999.
1001:
1002: \bibitem{mccoywu}
1003: B.M. McCoy, and T.T. Wu, {\it The Two Dimensional Ising Model.},
1004: Harvard University Press, Cambridge, USA, 1973.
1005:
1006: \bibitem{todakubo}
1007: M. Toda, R. Kubo, and N. Saito., {\it Statistical physics.},
1008: vol. 2, Springer-Verlag, New York, USA, 1992.
1009:
1010: \bibitem{mehta2}
1011: K.L. Mehta, {\it Matrix Theory - Selected Topics and Useful
1012: Results.}, Les Editions de Physique, Les Ulis Cedex, France, 1977.
1013:
1014:
1015:
1016:
1017: