1: %\documentclass[aps,preprint,superscriptaddress,showpacs]{revtex4}
2:
3: %\documentclass[aps,twocolumn,superscriptaddress,showpacs]{revtex4}
4: %\documentclass[aps,twocolumn,groupedaddress,showpacs]{revtex4}
5:
6: \documentclass[prl,twocolumn,showpacs,amssymb,amsfonts,amsmath,groupedaddress,floatfix,nofootinbib]{revtex4}
7:
8: \usepackage{epsf,latexsym,graphicx,graphics,psfrag}
9: \usepackage{times}
10: \usepackage{amsfonts,epsfig}
11:
12: \newcommand{\bra}[1]{\langle #1 |}
13: \newcommand{\ket}[1]{| #1 \rangle}
14: \newcommand{\braket}[2]{\left \langle #1 | #2 \right\rangle}
15:
16: \newcommand\R{{\mathrm {I\!R}}}
17: \newcommand\N{{\mathrm {I\!N}}}
18: \newcommand\h{{\cal H}}
19: \newcommand{\qed}{$\hfill \Box$}
20:
21: \newcommand{\ra}{{\rightarrow}}
22: \newcommand{\be}{\begin{equation}}
23: \newcommand{\ee}{\end{equation}}
24: \newcommand{\ba}{\begin{eqnarray}}
25: \newcommand{\ea}{\end{eqnarray}}
26:
27: \newcommand{\ignore}[1]{}
28:
29: \def\CC{{\rm\kern.24em \vrule width.04em height1.46ex depth-.07ex
30: \kern-.30em C}}
31: \def\P{{\rm I\kern-.25em P}}
32:
33: \def\RR{{\rm
34: \vrule width.04em height1.58ex depth-.0ex
35: \kern-.04em R}}
36:
37: \def\bbbone{{\mathchoice {\rm 1\mskip-4mu l} {\rm 1\mskip-4mu l}
38: {\rm 1\mskip-4.5mu l} {\rm 1\mskip-5mu l}}}
39:
40: \def\bbbc{{\mathchoice {\setbox0=\hbox{$\displaystyle\rm C$}\hbox{\hbox
41: to0pt{\kern0.4\wd0\vrule height0.9\ht0\hss}\box0}}
42: {\setbox0=\hbox{$\textstyle\rm C$}\hbox{\hbox
43: to0pt{\kern0.4\wd0\vrule height0.9\ht0\hss}\box0}}
44: {\setbox0=\hbox{$\scriptstyle\rm C$}\hbox{\hbox
45: to0pt{\kern0.4\wd0\vrule height0.9\ht0\hss}\box0}}
46: {\setbox0=\hbox{$\scriptscriptstyle\rm C$}\hbox{\hbox
47: to0pt{\kern0.4\wd0\vrule height0.9\ht0\hss}\box0}}}}
48:
49: \def\bbbz{{\mathchoice {\hbox{$\sf\textstyle Z\kern-0.4em Z$}}
50: {\hbox{$\sf\textstyle Z\kern-0.4em Z$}} {\hbox{$\sf\scriptstyle
51: Z\kern-0.3em Z$}} {\hbox{$\sf\scriptscriptstyle Z\kern-0.2em
52: Z$}}}}
53:
54: \newcommand{\putfig}[2]{$$\leavevmode\hbox{\epsfxsize=#2 cm
55: \epsffile{#1.eps}}$$}
56: \newcommand{\insertfig}[2]{\leavevmode \vcenter{\hbox{\epsfxsize=#2 cm
57: \epsffile{#1.eps}}}}
58:
59: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
60:
61:
62:
63: \begin{document}
64: \title{Ground state overlap and quantum phase transitions}
65: \author{Paolo Zanardi}
66: \author{Nikola Paunkovi\'c\footnote{Current address: SQIG, Instituto de Telecomunica\c{c}\~oes and Instituto Superior T\'ecnico, P-1049-001 Lisbon, Portugal .}}
67: \affiliation{ Institute for Scientific Interchange (ISI), Villa Gualino, Viale Settimio Severo 65, I-10133
68: Torino, Italy }
69: \begin{abstract}
70: We present a characterization of quantum phase transitions in terms of the the overlap function between two
71: ground states obtained for two different values of external parameters. On the examples of the Dicke and $XY$
72: models, we show that the regions of criticality of a system are marked by the extremal points of the overlap and
73: functions closely related to it. Further, we discuss the connections between this approach and the Anderson
74: orthogonality catastrophe as well as with the dynamical study of the Loschmidt echo for critical systems.
75:
76: \end{abstract}
77: \pacs{} \maketitle
78:
79: \section{Introduction} Quantum phase transitions (QPT)
80: \cite{sachdev} have drawn a considerable interest within various
81: fields of physics in the recent years. They are studied in
82: condensed matter physics because they provide valuable information
83: about the novel type of finite-temperature states of matter that
84: emerge in the vicinity of QPT \cite{sachdev}. %Also, the physics of
85: %QPT is closely connected to the study of high-$T_c$
86: %superconductors \cite{senthil}, quantum Hall systems, fractional
87: %quantum hall liquids etc. \cite{vojta}.
88: %In the context of quantum
89: %information science QPT are studied as they promise to provide
90: %future mechanisms for quantum computation \cite{QI}.
91: Unlike the ordinary phase transitions, driven by thermal fluctuations, QPT occur at zero temperature and are
92: driven by purely quantum fluctuations. In the parameter space, the points of non-analyticity of the ground
93: state energy density are referred to as critical points and define the QPT. In these points one typically
94: witnesses the divergence of the length associated to the two-point correlation function of some relevant quantum
95: field. An alternative way of characterizing QPT is by the vanishing, in the thermodynamical limit, of the energy
96: gap between the ground and the first excited state in the critical points.
97: %As they occur at the zero temperature, QPT
98: %can be only driven by Heisenberg uncertainty principle.
99: %This way
100: %of looking at QPT provoked
101: Recently, a huge interest was raised in the attempt of characterizating QPT in terms of the notions and tools
102: of quantum information \cite{qis}. More specifically QPT have been studied by analyzing scaling, asymptotic
103: behavior and extremal points of various entanglement measures \cite{osterloh, vidal, zanardi, others,lidar}.
104: %of a system: it is believed that the points of extremality of the
105: %well chosen entanglement function could capture the regions of
106: %criticality thus revealing QPT \cite{zanardi, others,lidar}.
107: More recently, the connection between geometric Berry phases and QPT in
108: the case of the $XY$ model has been also studied \cite{carollo}.
109:
110: In this paper, we aim to provide yet another characterization of
111: the regions of criticality that define QPT. We shall show how
112: critical points can be individuated by studying a surprisingly simple
113: quantity: the overlap i.e., the scalar product, between two ground states corresponding to two slightly
114: different values of the parameters. The physical intuition behind this approach should be obvious: QPT mark the
115: separation between regions of the parameter space which correspond to ground states having deeply different
116: structural properties e.g., order parameters. This difference is here quantified by the simplest Hilbert-space
117: geometrical quantity i.e., the overlap. Note that the square modulus of the overlap is nothing but the fidelity,
118: widely used in quantum information as a function that provides the criterion for distinguishability between
119: quantum states \cite{qis}. Therefore, it is a natural candidate for a study of macroscopic distinguishability
120: between quantum states that define different macroscopic states of matter (different phases). When applied to
121: cases of many-body systems containing many degrees of freedom, the overlap (or, fidelity) might seem to be too
122: coarse quantity, and not bearing any apparent information about the difference in order properties between
123: quantum phases, to be of any use. Nevertheless the main result of this paper is that in some cases it is indeed
124: possible to do so.
125: %It is important to stress that the point here, rather than to introduce
126: %a novel efficient numerical tools we are trying to make, is a conceptual one.
127: The critical behavior of a system undergoing QPT is reflected in the geometry of its Hilbert space: approaching
128: the QPT the overlap (distance) between neighboring ground states shows a dramatic drop (increase). We would like
129: to notice that Cejnar {\em et. al.} \cite{cejnar} discussed the overlap entropy between the eigen-bases of a
130: system's Hamiltonian and various physically relevant bases, in the context of enhanced decoherence effects in
131: the regions of criticality (see also \cite{other}).
132:
133:
134: %This gives a conceptually simple, physically motivated criterion
135: %for detecting regions of criticality.
136: %An analysis along related
137: %lines has been recently done in the context of QPT and the
138: %decoherence theory \cite{zanardi-china}.
139:
140: In the following two sections, we conduct our analysis on the
141: cases of two simple, yet physically relevant and mathematically
142: instructive, examples of the Dicke model and the $XY$
143: spin-chain model. Next, we discuss the connection between the
144: scaling and asymptotic behaviors and the so-called Anderson
145: orthogonality catastrophe \cite{anderson}. Moreover, the relation
146: with the dynamical study of decoherence and quantum criticality
147: \cite{zanardi-china} is briefly addressed.
148: Finally, in the last section conclusions are discussed.
149:
150:
151: % (\tilde{\gamma},\tilde{\lambda}) = (\gamma + \delta\gamma, \lambda + \delta\lambda)
152:
153: For a generic point in parameter space we use label $q\in\mathbb{R}^L$, where $L$ is the number of external
154: parameters determining system's Hamiltonian. As the overlap function depends on the difference between
155: parameters as well, we introduce $\tilde{q} \equiv q + \delta q$ to denote the neighboring point $\tilde{q}$ and
156: the difference $\delta q$. Following this notation, we denote the ground states by $\ket{g} \equiv \ket{g(q)}$
157: and $\ket{\tilde{q}} \equiv \ket{g(\tilde{q})}$. In general, all the functions $F(q)$ evaluated in the point
158: $\tilde{q}$ we will denote as $\tilde{F}$, while those evaluated in the critical point $q_c$, we will denote as
159: $F_c$ (note that by combining two cases, we have $\tilde{F}_c=F(q_c+\delta q)$). Then, the overlap function is
160: simply given by the scalar product $\langle g(q) | g(\tilde{q}) \rangle$ (note that all the results of this
161: paper could be easily formulated in terms of fidelity as well).
162: %As the regions of criticality are given by
163: %the values of the parameters $q$ only, we will in our study take a
164: %small, yet finite value of $\delta q$ and
165: We shall examine the behavior of
166: the overlap as a function of $q$ only, while keeping
167: $\delta q$ fixed and small.
168:
169:
170: \section{ The Dicke Model} Our first example is the Dicke
171: model. It describes a dipole interaction between a
172: single bosonic mode $\hat{a}$ and a collection of $N$ two-level
173: atoms. If for $N$ atoms we introduce the collective angular
174: momentum operators $\hat{J}_s, s\in\{\pm,z\}$, Dicke Hamiltonian
175: has the following form (we take $\hbar=1$): \be \label{Dicke
176: Hamiltonian} \hat{H}(\lambda) = \omega_0\hat{J}_z +
177: \omega\hat{a}^\dag\hat{a} +
178: \frac{\lambda}{\sqrt{2j}}\left(\hat{a}^\dag+\hat{a}\right)\left(\hat{J}_+
179: + \hat{J}_-\right). \ee Parameter $\lambda$ is the atom-field
180: coupling strength and is the one driving the QPT in this model.
181: Therefore, we have $q=\lambda$ and denote the Hamiltonian's
182: dependance on that parameter only. Parameters $\omega_0$ and
183: $\omega$ stand for the atomic level-splitting and bosonic mode
184: frequency, respectively, while $j$ describes the length of a
185: collective spin vector, and is assumed to be constant and equal to
186: $j=N/2$. In the thermodynamical limit $(N\rightarrow \infty)$,
187: which is here equivalent to $(j\rightarrow \infty)$, Dicke
188: Hamiltonian undergoes a quantum phase transition for the critical
189: value of its parameter $\lambda$ given by
190: $\lambda_c=(\omega\omega_0)/2$. When $\lambda<\lambda_c$, the
191: system is in highly unexcited {\em normal} phase, while
192: $\lambda>\lambda_c$ defines the {\em super-radiant} phase in which
193: both the field and $N$ atoms become macroscopically excited. The
194: super-radiant phase is characterized by the broken symmetry given
195: by the parity operator $\hat{\Pi} = \exp(i\pi\hat{N}),
196: \hat{N}=(\hat{a}^\dag\hat{a}+\hat{J}_z+j)$: the ground state is
197: doubly degenerate. As shown in \cite{emary}, by introducing
198: bosonic operators $\hat{b}$ through Holstein-Primakoff
199: representation \cite{primakoff}, the above Dicke Hamiltonian
200: (\ref{Dicke Hamiltonian}) can be exactly diagonalized in the
201: thermodynamical limit. In the normal phase, its form is: \be
202: \label{Dicke - Normal} \hat{H}^n(\lambda) =
203: \omega_0\hat{b}^\dag\hat{b} + \omega\hat{a}^\dag\hat{a} +
204: \lambda\left(\hat{a}^\dag+\hat{a}\right)\left(\hat{b}^\dag +
205: \hat{b}\right) - j\omega_0.\ee Its ground state is:
206: %\be\label{ground-normal}
207: $g(x,y)=\left(\frac{\varepsilon_+\varepsilon_-}{\pi^2}\right)^{\frac{1}{4}}e^{-1/2\langle
208: {\bf{R}}, A{\bf{R}}\rangle} [{\bf{R}}=(x,y)]$ where $x$ and $y$
209: are the real space coordinates associated to the modes $\hat{a}$
210: and $\hat{b},$ $A=U^{-1}MU$,
211: $M=\text{diag}[\varepsilon_-,\varepsilon_+]$ and
212: $U$ an orthogonal matrix $U=\left[ \begin{array}{cc} c & -s \\
213: s & c
214: \end{array}\right]$, ($c=\cos\gamma, s=\sin\gamma$ are given by the squeezing angle
215: $\gamma=(1/2)\arctan[4\lambda\sqrt{\omega\omega_0}/(\omega^2+\omega_0^2)]$). $\varepsilon_{\pm}$ represent the
216: fundamental collective excitations of the system and are given by:
217: $\varepsilon^2_{\pm}=\frac{1}{2}\left(\omega^2+\omega_0^2\pm\sqrt{(\omega^2-\omega_0^2)^2+16\lambda^2\omega^2\omega_0^2}\right).$
218: From the above formula, we see that $\varepsilon_-(\lambda_c)\equiv\varepsilon^c_-=0$: the system becomes
219: gapless and undergoes a QPT for $\lambda=\lambda_c$.
220:
221: The overlap, calculated between two ground states $g$ and $\tilde{g}$, is given by
222: %(\ref{ground-normal}) and the corresponding matrices $A$ and
223: %$\tilde{A}$ (note that we use the notation introduced in the
224: %previous section). By using the formula
225: %$\int\int_{-\infty}^{+\infty}dxdy\exp\{ -[ x \ y ] A \left[
226: %\begin{array}{cc} x \\ y
227: %\end{array}\right]\}=\pi/\det(A)$, we get for the overlap
228: %(fidelity) function the following expression:
229: \be \label{DickeOverlap} \langle g|\tilde{g}\rangle\!=\!2\frac{[\det\! A \det\!
230: \tilde{A}]^{\frac{1}{4}}}{[\det(A\!\!+\!\!\tilde{A})]^\frac{1}{2}}\!\!=\!\!2\frac{[\det\!
231: A]^{\frac{1}{4}}}{[\det\! \tilde{A}]^\frac{1}{4}[\det(1\!\!+\!\!\tilde{A}^{-1}\!\!A)]^\frac{1}{2}}.\ee Note that
232: the overlap is a function of both $\lambda$ and $\delta\lambda$. In the limit
233: $(\lambda\rightarrow\lambda_{c})$, with $\delta\lambda>0$ being fixed, $\det
234: A=\varepsilon_+\varepsilon_-\rightarrow 0$, while $\det \tilde{A}\geq\det
235: \tilde{A}_c=\tilde{\varepsilon}^c_+\tilde{\varepsilon}^c_->0$. The same holds for $\det(1+\tilde{A}^{-1}A)$, for
236: a sufficiently small $\delta\lambda$ (note that $\lim_{\delta\lambda\rightarrow 0}\tilde{A}^{-1}=A^{-1}$). But
237: in the present case, it is possible to show that for every, and not just small $\delta\lambda$,
238: $\det(1+\tilde{A}^{-1}A)$ does not vanish. Using the formula $\det (1+A)=1+\mbox{Tr}A+\det A$ for $2\times 2$
239: matrices, we get: $\det(1+\tilde{A}^{-1}A)\rightarrow 1+\mbox{Tr}(\tilde{A}_c^{-1}A_c)$ (note that
240: $\det(\tilde{A}^{-1}A)=[\det\tilde{A}]^{-1}\det A\rightarrow 0$). After a straightforward calculation, we obtain
241: the result: $
242: \mbox{Tr}(\tilde{A}_c^{-1}A_c)=\frac{\varepsilon^c_+}{\tilde{\varepsilon}^c_+\tilde{\varepsilon}^c_-}\left[
243: (s\tilde{c}+c\tilde{s})^2\tilde{\varepsilon}^c_+ + (s\tilde{s}-c\tilde{c})^2\tilde{\varepsilon}^c_-\right].$
244: Therefore, $\mbox{Tr}(\tilde{A}_c^{-1}A_c)>0$ for every $\lambda$ and we can conclude that $\langle
245: g|\tilde{g}\rangle\propto (\varepsilon_-)^{1/4}$ as $(\lambda\rightarrow\lambda_c)$. In Ref.~\cite{emary}, it
246: was shown that when approaching the critical point from both normal and super-radiant sides, the excitation
247: energy $\varepsilon_-$ drops as the square root of $\Delta\equiv |\lambda-\lambda_c|$, which gives us the
248: asymptotic behavior of the overlap function in the vicinity of the critical point: $\langle
249: g|\tilde{g}\rangle\propto \Delta^{1/8}$. Although we have provided here only the results for overlap function
250: for the system in the normal phase, the analogous analysis for the super-radiant phase gives us the same
251: qualitative results, as the two ground states are again the Gaussian-type states, but with translated and
252: re-scaled $x$ and $y$ axes. Therefore, we omit it here.
253:
254: % Following the above introduced notation, we define:
255: % $\varepsilon_+^c\equiv \varepsilon_+(\lambda_c)$,
256: % $\tilde{\varepsilon}_\pm^c \equiv
257: % \varepsilon_\pm(\lambda_c+\delta\lambda)$, $\tilde{A}_c^{-1}\equiv
258: % A^{-1}(\lambda_c+\delta\lambda)$ and $A_c\equiv A(\lambda_c)$.
259:
260: %______________________________________________________________________ FIGURE
261:
262: \begin{figure}[ht]
263: \includegraphics[width=6.5cm,height=3.7cm,angle=0]{d1.eps}
264: \caption{(color online) The overlap function $\langle g |
265: \tilde{g}\rangle$, equation (\ref{DickeOverlap}), as a function of
266: $\lambda<\lambda_c$, taken for the resonant case
267: $\omega_0=\omega=1$ and $\delta\lambda=10^{-6}$. Note the dramatic
268: decreasing of the function as we approach the point of
269: criticality.} \label{Dicke Overlap - Figure}
270: \end{figure}
271:
272: %______________________________________________________________________
273:
274: We conclude this section by presenting the numerical results for the overlap function in the normal phase. In
275: Fig. \ref{Dicke Overlap - Figure} we plot the overlap (\ref{DickeOverlap}) between two ground states of Dicke
276: Hamiltonian for the resonant case $\omega_0=\omega=1$ and $\delta\lambda=10^{-6}$. We see that it is almost
277: constant and equal to unity for wide range of $\lambda$, apart from the very narrow area around $\lambda_c$,
278: when it drastically drops to zero. Such behavior of the overlap function around the point of criticality can be
279: ascribed to the fact that the ground state for $\lambda=\lambda_c$ becomes completely delocalized along one of
280: two rotated axes, as opposed to the localized ground state outside of the point of criticality (see
281: \cite{emary}).
282:
283: %Finally,
284: %we note that approximate expression $\langle g|\tilde{g}\rangle
285: %\propto \exp\{-\frac{1}{16}\mbox{Tr}(A^{-1}\delta A)^2\}$, with
286: %$\delta A = \tilde{A}-A$, for the overlap, obtained
287: %from equation (\ref{Dicke Overlap}) by Taylor expansion up to
288: %second order terms, matches numerically well results obtained by
289: %the exact formula (\ref{Dicke Overlap}). In obtaining this
290: %formula, we used the identity: $\det A = \exp [\mbox{Tr}(\log
291: %A)]$.
292:
293: \section{The XY Spin Chain} In the following section, we
294: discuss the example of the one-dimensional $XY$ anisotropic
295: spin-half chain in the external magnetic field. Its Hamiltonian is
296: given by the following expression:
297:
298: \be \label{Hamiltonian} \hat{H}(\gamma, \lambda)\! =\!\! -\!\!\!\!\sum_{i = -M}^{M} \!\!\! \left(
299: \frac{1+\gamma}{2}\hat{\sigma}_{i}^x \hat{\sigma}_{i+1}^x\!\! +\!\! \frac{1-\gamma}{2} \hat{\sigma}_{i}^y
300: \hat{\sigma}_{i+1}^y\!\! +\!\! \frac{\lambda}{2}\hat{\sigma}_{i}^z\right). \!\!\ee The parameter $\gamma \in
301: \mathbb{R} $ defines the anisotropy, while $\lambda \in \mathbb{R}$ represents external magnetic field along the
302: $z$ axis, up to a factor $\frac{1}{2}$. Therefore, $q=(\gamma,\lambda)$. The operators
303: $\hat{\sigma}_{i}^{\alpha},\, \alpha \in \{ x,y,z\}$ are the usual Pauli operators. This Hamiltonian can be
304: exactly diagonalized by successively applying Jordan-Wigner, Furier and Bogoliubov transformation (see for
305: example \cite{sachdev}). This way, we obtain the following form of the Hamiltonian: $ \hat{H}(\gamma, \lambda) =
306: \sum _{k = -M}^M\Lambda_k(\hat{b}^{\dagger}_k \hat{b}_k - 1).$ The energies of one-particle excitations are
307: given by $\Lambda_k=\sqrt{\varepsilon_k^2+\gamma^2\sin^2\frac{2\pi k}{N}},$ with $\varepsilon_k=\cos\frac{2\pi
308: k}{N}-\lambda$ and $N=2M+1$ being the total number of sites (spins). One-particle excitations are given by the
309: fermionic operators $\hat{b}_k=\cos \frac{\theta_k}{2}\hat{d}_k -i\sin\frac{\theta_k}{2}\hat{d}^{\dag}_{-k},$
310: with $ \cos\theta_k = \varepsilon_k/\Lambda_k$.
311: %= \frac{\cos\frac{2\pi
312: %k}{N}-\lambda}{\sqrt{(\cos\frac{2\pi
313: %k}{N}-\lambda)^2+\gamma^2\sin^2\frac{2\pi k}{N}}}.$
314: Finally, the
315: ground state $\ket{g(\gamma, \lambda)}$, that is defined as the
316: state to be annihilated by each operator $\hat{b}_k$
317: ($\hat{b}_k\ket{g(\gamma, \lambda)}\equiv 0$), is given as a
318: tensor product of qubit-like states: \be \label{ground_state}
319: \ket{g(\gamma, \lambda)}\!\! = \!\!\bigotimes_{ k =
320: 1}^M\!\!\left(\!\! \cos\frac{\theta_k}{2} \ket{0}_k\ket{0}_{-k}
321: \!\!-i \sin\frac{\theta_k}{2}\ket{1}_k\ket{1}_{-k} \!\!\right).
322: \ee
323:
324: In its space of parameters, the family of Hamiltonians given by
325: equation (\ref{Hamiltonian}) exhibits two regions of criticality,
326: defined by the existence of gapless excitations: {\em (i)} $XX$
327: region of criticality, for $\gamma=0$ and $\lambda \in (-1,1)$;
328: {\em (ii)} $XY$ region of criticality given by the lines $\lambda
329: = \pm 1$.
330:
331: As in the previous example, let us first consider the exact
332: overlap function. From equation (\ref{ground_state}), it follows
333: that the exact overlap function between the ground states
334: $\ket{g}$ and $\ket{\tilde{q}}$ is: \ba \label{overlap-exact}
335: \langle g(q) | g(\tilde{q}) \rangle = \prod_{k=1}^M
336: \cos\frac{\theta_k - \tilde{\theta}_k}{2}, \ea where
337: $\tilde{\theta}_k=\theta_k(\tilde{q})$. Note the dependence on the
338: number of sites $N$ that is implicit in all the previous formulae
339: from this section. In Fig. \ref{XY-figure}(a), we present the
340: numerical result obtained using the above equation
341: (\ref{overlap-exact}), for $N=10^6$ spins and
342: $\delta\lambda=\delta\gamma=10^{-6}$. We observe that the regions
343: of criticality are clearly marked by a sudden drop of the value of
344: the overlap function. As before, we ascribe this type of behavior
345: to a dramatic change in the structure of the ground state of the
346: system while undergoing QPT.
347:
348: In order to investigate the overlap function more quantitatively
349: and relate its behavior to the existence of the regions of
350: criticality, we note that while the overlap depends on the values
351: of both the parameters $q$ and the difference $\delta q$, the
352: regions of criticality are defined by the values of parameters
353: only. Therefore, in the following we choose to study the functions
354: \be \label{Sl} S^{\lambda}_N(\lambda, \gamma) \equiv
355: \sum_{k=1}^M\left(\frac{\partial\theta_k}{\partial\lambda}\right)^2,
356: S^{\gamma}_N(\lambda, \gamma) \equiv
357: \sum_{k=1}^M\left(\frac{\partial\theta_k}{\partial\gamma}\right)^2,\ee
358: that define the first non-zero order of the Taylor expansion of
359: the overlap function (\ref{overlap-exact}).
360: %($\langle g(q) | g(\tilde{q})
361: %\rangle \propto 1 - \frac{1}{8} \sum_{k=1}^M
362: %\{\left(\frac{\partial\theta_k}{\partial\lambda}\right)^2\delta\lambda^2
363: %+
364: %\left(\frac{\partial\theta_k}{\partial\gamma}\right)^2\delta\gamma^2
365: %+ \left(\frac{\partial\theta_k}{\partial\lambda}\right)
366: %\left(\frac{\partial\theta_k}{\partial\gamma}\right)\delta\lambda\delta\gamma\}$).
367: Functions $S^{\lambda}_N(\lambda, \gamma)$ and
368: $S^{\gamma}_N(\lambda, \gamma)$ are natural candidates for our
369: study because they express the ``rate of change" of the ground
370: state, taken in the point $q$. They do not depend on the
371: difference $\delta q$, and although for every finite $N$ it is
372: possible to find $\delta q$ small enough so that the exact overlap
373: is arbitrarily well approximated by the expression $\exp
374: (-\frac{1}{8}S^q_N(q)\delta q^2)$, functions
375: $S^{\lambda}_N(\lambda, \gamma)$ and $S^{\gamma}_N(\lambda,
376: \gamma)$ on their own capture the behavior of the $\langle g(q) |
377: g(\tilde{q}) \rangle$ function and are enough for our current
378: study. They also allow for analytic investigation, together with
379: numerical one. In Figs. \ref{XY-figure}(b) and \ref{XY-figure}(c)
380: we present the numerical results for $S^{\lambda}_N(\lambda,
381: \gamma)$ and $S^{\gamma}_N(\lambda, \gamma)$, respectively, for
382: $N=10^6$ spins. Again, the regions of criticality could easily be
383: inferred by simply observing both plots. Note that in this case
384: the relative difference between the numerical values in the
385: regions of criticality and elsewhere is much
386: bigger than in the case of the exact overlap (see Fig.
387: \ref{XY-figure}(a)). But now, {\em both} plots are needed to
388: detect both regions of criticality. This is so because by moving
389: along $\gamma=0$, while keeping $|\lambda|<1$, we do not move
390: outside the $XX$ region of criticality and therefore do not expect
391: the qualitative change in the structure of the ground state, and
392: consequently in the behavior of $S^{\lambda}_N(\lambda, \gamma)$
393: as well. The same holds for $S^{\gamma}_N(\lambda, \gamma)$ and
394: the $XY$ region of criticality.
395:
396: %In the case of the $XY$ model, regions of criticality define the
397: %{\em continuous} QPT: they are the consequence of the existence of
398: %{\em gapless excitations} that emerge only in thermodynamical
399: %limit, when $(N \rightarrow \infty)$.
400: %Therefore,
401:
402: %______________________________________________________________________ FIGURE
403: %\begingroup
404: %\begin{widetext}
405: \begin{center}
406: \begin{figure}[tp]
407: \begin{picture}(10,0)
408: \put(-30,50){(a)}\put(8,50){\tiny{1}}\put(3,27){\tiny{0.9996}} \put(-30,-45){(b)}\put(-8,-35){\tiny{$3.5\times
409: 10^9$}}\put(3,-65){\tiny{0}} \put(-30,-133){(c)}\put(-8,-127){\tiny{$2.5\times 10^8$}}\put(3,-155){\tiny{0}}
410: \end{picture}
411: \psfrag{l}[Bc][][0.75][0]{$\lambda$} \psfrag{g}[Bc][][0.75][0]{$\gamma$} \psfrag{gg}[Bc][][0.75][0]{$$}
412: \psfrag{l}[Bc][][0.75][0]{$\lambda$}%
413: \psfrag{g}[Bc][][0.75][0]{$\gamma$} \psfrag{gg}[Bc][][0.75][0]{$$}
414: \includegraphics[width=4.5cm,height=3.2cm,angle=0]{e1}\hskip
415: 2mm
416: \includegraphics[width=4.5cm,height=3.2cm,angle=0]{sl1}\hskip
417: 2mm
418: \includegraphics[width=4.5cm,height=3.2cm,angle=0]{sg1}
419: %\includegraphics[width=13cm,height=3.7cm,angle=0]{plots/a}
420: %\includegraphics[width=4.5cm,height=3.2cm,angle=0]{plots/Exact_Overlap_(million_spins)1}\hskip
421: %-2mm
422: %\includegraphics[width=4.5cm,height=3.2cm,angle=0]{plots/Sl(million_spins)1}\hskip
423: %-2mm
424: %\includegraphics[width=4.5cm,height=3.2cm,angle=0]{plots/Sg(million_spins)1}
425: \hskip -2mm \caption{(color online) (a) The overlap function $\langle g(q) | g(\tilde{q})\rangle$, as a
426: function of $\lambda$ and $\gamma$, for $N=10^6$ and $\delta\lambda=\delta\gamma=10^{-6}$. Note the clear dips
427: of the plot in the regions of criticality. (b) $S^{\lambda}_N(\lambda, \gamma)$. (c) $S^{\gamma}_N(\lambda,
428: \gamma)$.} \label{XY-figure}
429: \end{figure}
430: \end{center}
431: %\end{widetext}
432: %\endgroup
433:
434: We first examine the scaling behavior of $S^{\lambda}_N(\lambda, \gamma)$ and $S^{\gamma}_N(\lambda, \gamma)$
435: with respect to number of spins $N$. The numerics present us with the following results. First, as expected,
436: $S^{\lambda}_N(\lambda, \gamma)$ and $S^{\gamma}_N(\lambda, \gamma)$ scale linearly with $N$ when
437: $(\gamma\rightarrow 0)$ and $(\lambda\rightarrow \pm 1)$, respectively. In the regions of criticality, we have
438: that $S^{\lambda}_N(|\lambda|=1, \gamma)\propto N^2/\gamma^2$, while on the other hand, for $(\gamma\rightarrow
439: 0)$ we still have $S^{\gamma}_N(\lambda, \gamma)\propto N$, without being a function of $\lambda$ (we note here
440: that $S^{\gamma}_N(|\lambda|<1, \gamma=0)=0$). Such different behavior is a consequence of the fact that while
441: the $XY$ region of criticality defines the second order QPT, $XX$ is the example of the third order QPT.
442:
443:
444: We have also conducted a separate analytical study, confirming the above numerical results. First, we note that
445: for every point $q$ in parameter space, and every {\em finite} $N$, partial derivatives
446: $\left(\frac{\partial\theta_k}{\partial\lambda}\right)$ and
447: $\left(\frac{\partial\theta_k}{\partial\gamma}\right)$ are continuous functions of the parameters. They can
448: become infinite only in the thermodynamical limit, when $(N \rightarrow \infty)$, and only in the {\em regions
449: of criticality}. By looking at the explicit form of derivatives (we use $x_k = \frac{2\pi}{N}k$):\be \label{Dl}
450: \left(\frac{\partial\theta_k}{\partial\lambda}\right) = \frac{\gamma(\sin x_k)}{[(\cos x_k - \lambda)^2 +
451: \gamma^2(\sin x_k)^2]},\ee
452:
453: \be \label{Dg} \left(\frac{\partial\theta_k}{\partial\gamma}\right) =
454: - \frac{|\sin x_k|(\cos x_k - \lambda)}{[(\cos x_k - \lambda)^2 +
455: \gamma^2(\sin x_k)^2]},\ee we see that only when the energy $\Lambda_k$ (the denominator of both of the above
456: expressions) gets arbitrarily small (or zero), the derivatives (\ref{Dl}) and (\ref{Dg}) can become divergent.
457: In other words, only when $\cos x_k$ gets arbitrarily close to $\lambda$, {\em and} either $\gamma$ or $\sin
458: x_k$ get close to zero. That is, in the regions of criticality. Note that we assume that for every $N \in
459: \mathbb{N}$ and $k \in \{ 1,\ldots M \}$, equation $\cos x_k = \lambda$ has no solution, which presents a
460: generic case ($\lambda$'s that allow for the solutions of this equation form a set of measure zero on the
461: $(-1,1)$ interval). Therefore, outside the regions of criticality $S^{\lambda}_N(\lambda, \gamma)$ and
462: $S^{\gamma}_N(\lambda, \gamma)$ scale linearly with $N$.
463:
464:
465: Regarding the regions of criticality, we first consider the scaling behavior of $S^{\lambda}_N(\lambda,
466: \gamma)$ in the vicinity of the $XX$ criticality. As there always exists $k_0$ such that in the $(N\rightarrow
467: \infty)$ limit $\cos x_{k_0} \rightarrow \lambda$, then for such $x_{k_0}$ and every finite $\gamma$, it follows
468: from (\ref{Dl}) that $\left(\frac{\partial\theta_{k_0}}{\partial\lambda}\right) \rightarrow (\gamma \sin
469: x_{k_0})^{-1}$, when $(N\rightarrow \infty)$. In other words, it does not scale with $N$ (note that although
470: $k_0=k_0(N)$ is a function of $N$, $\lim_{N\rightarrow\infty}\sin x_{k_0}=\sin\arccos\lambda$). As all other
471: derivatives are finite, we have that $S^{\lambda}_N(|\lambda|<1, \gamma\rightarrow 0)\propto N/\gamma^2$.
472: Similar discussion can be applied to the case of $S^{\gamma}_N(\lambda, \gamma)$ in the $XY$ region of
473: criticality. Again, there exists a qubit defined by $k_1=1$ for which $\cos x_{k_1} \rightarrow 1$ in the
474: $(N\rightarrow \infty)$ limit, so that its existence could bring about the scaling of $S^{\gamma}_N(\lambda,
475: \gamma)$ larger than linear in thermodynamical limit. Using the Taylor expansion of sine and cosine functions
476: around zero (note that in that case, $\sin x_{k_1}\rightarrow0$ as well), from equation (\ref{Dg}) we obtain
477: $\left(\frac{\partial\theta_k}{\partial\gamma}\right)\propto x_{k_1}/\gamma^2\rightarrow 0,
478: (N\rightarrow\infty)$. In other words, $S^{\gamma}_N(|\lambda|=1, \gamma)\propto N$.
479:
480: Now, we turn to more interesting cases of the relevant functions $S^{\lambda}_N(\lambda, \gamma)$ and
481: $S^{\gamma}_N(\lambda, \gamma)$, in the $XY$ and $XX$ regions of criticality, respectively. Using Taylor
482: expansions of sine and cosine functions around zero, we see that in $\lambda = \pm 1$ the derivative
483: $\left(\frac{\partial\theta_{k_1}}{\partial\lambda}\right)$ given by $k_1=1$ behaves like
484: $\left(\frac{\partial\theta_{k_1}}{\partial\lambda}\right) \propto N/(2\pi\gamma)$ as $(N\rightarrow \infty)$
485: (see equation (\ref{Dl})) and therefore $S^{\lambda}_N(|\lambda|=1, \gamma)\propto N^2/\gamma^2$. We also see
486: that the scaling factor depends on $\gamma$. Finally, from (\ref{Dg}), we also see that $S^{\gamma}_N(\lambda,
487: \gamma)\propto N$.
488:
489:
490: The alternative way to examine the signatures of QPT is to look at the asymptotic behavior of two functions
491: (\ref{Sl}) near the regions of criticality. From the numerical study we obtain that the asymptotic behavior of
492: $S^{\lambda}_N(\lambda, \gamma)$ in the vicinity of critical points $\lambda_c = \pm1$, $\gamma \in (0,1]$, is
493: given by the following formula: $S^{\lambda}_N(\lambda, \gamma)\propto
494: a(\gamma,N)/|1-\lambda|^{\alpha(\gamma,N)}$. From the study of the scaling behavior, we already know that
495: $a(\gamma,N)=a(\gamma)N^2$ and that $a(\gamma) \propto 1/\gamma^2$. Further, from numerics we have that the
496: exponent $\alpha(\gamma,N)$ is constant with respect to $\gamma$ and approaches to $\alpha=1$ as $(N\rightarrow
497: \infty)$. Such asymptotic behavior, with constant exponent $\alpha=1$ for all $\gamma \in (0,1]$ could be seen
498: as a consequence of the fact that in that range of parameters the $XY$ model belongs to the same class of
499: universality. The numerics gives also the asymptotic behavior of $S^{\gamma}_N(\lambda, \gamma)$ in the vicinity
500: of $\gamma=0$ (with $|\lambda|<1$) similar to the previous one, $ S^{\gamma}_N(\lambda, \gamma)\propto
501: b(\lambda,N)/\gamma^{\beta(\lambda,N)}$, with the exponent $\beta(\lambda,N)$ approaching to $\beta=1$ as
502: $(N\rightarrow \infty)$. But, the coefficient $b(\lambda,N)$ depends only on $N$, and as noted before, scales
503: linearly with it, $b(\lambda,N)\propto bN$.
504:
505: \section{QPT: Orthogonality catastrophe, Loschmidt echo} The above two examples represent a generic case of a
506: many-body system which exhibits continuous QPT only in the thermodynamical limit.
507: %The
508: %$XY$ model is of particular interest, as it can be explicitly
509: %diagonalized for a finite $N$, and it presents a free-fermion
510: %system whose ground state (\ref{ground_state}) is given as a
511: %simple tensor product of qubit-like states.
512: In the case of the $XY$ model, as the number of spins increases, the overlap between two different ground states
513: (\ref{ground_state}) approaches to zero, no matter how small the difference in parameters $\delta q$ is, so that
514: in thermodynamical limit each two ground states are mutually orthogonal; they live in a continuous tensor
515: product space \cite{ITP}.
516: %(mathematically, they are
517: %represented by the infinite tensor product vectors and belong to
518: %different sectors of the GNS representation \cite{ITP}).
519: Such behavior of systems having infinitely many degrees of
520: freedom, when the two physical states corresponding to two
521: arbitrarily close sets of parameters (two arbitrarily similar
522: physical situations) become orthogonal to each other, has been
523: already studied in many-body physics and is known as Anderson {\em
524: orthogonality catastrophe} \cite{anderson}. From our study of the
525: $XY$ model, we have seen that not only that every two ground
526: states become orthogonal in thermodynamical limit, but also the
527: rate of ``orthogonalization" between two ground states of large,
528: but finite system, changes qualitatively and grows faster in the
529: vicinity of the regions of criticality.
530: %This observation confirms
531: %the right choice of the study of the functions
532: %$S^{\lambda}_N(\lambda, \gamma)$ and $S^{\gamma}_N(\lambda,
533: %\gamma)$: they do not depend on the difference $\delta q$ between
534: %the sets of parameters (on how much two physical situations given
535: %by two different Hamiltonians differ from each other), but only on
536: %the value of parameters in the point $q$. The
537: %The functions $S^{\lambda}_N(\lambda, \gamma)$ and
538: %$S^{\gamma}_N(\lambda, \gamma)$: represent the ``rate of change"
539: %of the ground state as we start moving it away from the point $q$.
540: This way, the regions of criticality of an infinite system are
541: already marked by the scaling and asymptotic behavior of the
542: relevant functions of a finite-size system. Loosely speaking, the
543: regions of criticality of QPT are given as regions where the
544: orthogonality catastrophe is expressed on qualitatively greater
545: scale. Notice that recently, the occurrence of a particular
546: instance of Anderson-type orthogonality catastrophe was studied
547: for the case of a system in the vicinity of QPT
548: \cite{sachdev-catastrophe}.
549: %In the current work, our approach is
550: %different and we discuss the possible connections between QPT and
551: %the general concept of orthogonality catastrophe looking at the
552: %behavior of a system in a region of criticality through a paradigm
553: %introduced by Anderson \cite{anderson}.
554:
555: Now we would like to establish an explicit connection between the
556: sort of kinematical approach used in this paper and the dynamical
557: one of Ref. \cite{zanardi-china}. In order to do so let us
558: introduce the projected density of states function
559: $D(\omega;q,\tilde{q}) \equiv \bra{g(\tilde{q})}\delta(\omega -
560: \hat{H}(q))\ket{g(\tilde{q})}$ that describes the spread of the
561: ground state $\ket{g(\tilde{q})}$ expressed in the eigenbasis
562: obtained for the point $q$. Then, the square of the overlap can be expressed as
563: \begin{equation}
564: |\langle g(q)|g(\tilde{q})\rangle |^2= 1 -
565: \int_{E_1}^{\infty}D(\omega)d\omega
566: \end{equation}
567: ($E_1$ denotes the first
568: excited eigenvalue). We see that in regions in which the spread of
569: $\ket{g(\tilde{q})}$ with respect to $\ket{g(q)}$ is big
570: (quantified in terms of the variance of $D(\omega)$), in other
571: words in regions where two ground states differ significantly, the
572: overlap will be small. Recently, Quan {\em et al.}
573: \cite{zanardi-china} established a link between the critical
574: behavior of the environment and quantum decoherence, showing that
575: the Loschmidt echo \cite{Loschmidt,Loschmidt1} $L(q,t)$ of the environment
576: exponentially goes to zero as we approach the regions of
577: criticality. A simple algebra gives us that the Fourier transform
578: of the projected density of states is precisely the Loschmidt
579: echo, $|\int_{-\infty}^{+\infty}D(\omega)e^{-i\omega
580: t}d\omega|^2=L(q,t).$
581: We see that the kinematics of a system,
582: given by the geometry of ground states through the overlap
583: function, influences its dynamics as well:
584: the smaller the overlap i.e., broader $D(\omega)$ the faster
585: the decay of the Loschmidt echo.
586:
587: \section{Conclusions and Discussion} In this paper, by discussing
588: the examples of the Dicke and $XY$ spin-chain models, we have
589: presented a characterization of quantum phase transitions
590: based on the study of the scaling and asymptotic behaviors of the
591: overlap between two ground states taken in two
592: close points of the parameter space. Though this quantity might in general
593: not provide an efficient numerical tool it is conceptually quite appealing.
594: In fact the ground state overlap is a purely Hilbert-space geometrical quantity, whose investigation
595: does not rely on any a priori understanding of the specific kind
596: of order patterns or peculiar dynamical correlations hidden in the analyzed system.
597: %It provides with a
598: %quantitative description of the common intuition that the ground
599: %state of the system undergoing a QPT exhibits an essential change
600: %in its structure.
601: % given by the change of the symmetry group that
602: %leaves the state invariant.
603: While in the case of the Dicke Hamiltonian it was possible to
604: analyze the overlap function directly in the thermodynamical
605: limit, the case of the $XY$ model is more subtle.
606: %There, the Hilbert space of physical
607: %states in thermodynamical limit is {\em not} separable: every two
608: %ground states corresponding to two distinct parameters, regardless how close,
609: %are exactly
610: %orthogonal.
611: %this is mathematical formulation of the Orthogonality
612: %Catastrophe for the case of the $XY$ model.
613: %Yet, the study of the
614: %overlap function, as well as the scaling and asymptotic
615: %behavior of functions $S^{\lambda}_N(\lambda, \gamma)$ and
616: %$S^{\gamma}_N(\lambda, \gamma)$ that represent the "rate of
617: %change" of the ground state, shows the signature of QPT even for a
618: %finite-size system. This observation leads to a conclusion that
619: The process of ``orthogonalization" between two ground states has
620: two physically different mechanisms in the case when two states
621: belong to two different phases: one is a common decrease of the
622: overlap due to infinite number of sub-systems in thermodynamical
623: limit, the other is characteristic for the case of QPT and is due
624: to different structures of ground states in different quantum
625: phases.
626:
627: %In fact, the latter mechanism is the only one in the case
628: %of the Dicke model.
629: %There are two main paths for the future work. The first is finding
630: %other examples of systems whose regions of criticality can be
631: %inferred by the study of the overlap function, or some
632: %analogue of $S^{\lambda}_N(\lambda, \gamma)$ and
633: %$S^{\gamma}_N(\lambda, \gamma)$.
634: %We believe that the similar
635: %analysis to the one presented in this paper could be undertaken
636: %for every system whose different quantum phases are given by
637: %different symmetry groups of the ground states.
638: %Second, it would
639: %be of equal interest to find systems whose QPT are not captured by
640: %the behavior of the overlap between the ground states. One such
641: %class of candidates would be systems in which different states of
642: %matter are given by the same symmetry groups, but different
643: %quantum or topological orders \cite{wen}. We hope this would shade
644: %more light on the relationship between the structure of the ground
645: %states of a system and their different orders.
646:
647:
648:
649:
650:
651:
652:
653: %_____________________________________________________________________ ACKNOWLEDGEMENTS
654:
655: %\begin{acknowledgments}
656: \section{acknowledgments.}
657: NP is funded by the European Commission, contract No.
658: IST-2001-39215 TOPQIP. The authors greatly acknowledge the
659: discussions with A. T. Rezakhani.
660:
661:
662: %\end{acknowledgments}
663:
664: %______________________________________________________________________ BIBLIOGRAPHY
665:
666: \begin{thebibliography}{}
667:
668: \bibitem{sachdev} S.~Sachdev, {\em Quantum Phase Transitions}, Cambridge University Press (1999).
669:
670: %\bibitem{senthil} T.~Senthil,
671:
672: %\bibitem{vojta} M.~Vojta, Rep. Prog. Phys {\bf 66}, 2069 (2003).
673:
674: \bibitem{qis} For reviews, see A. Steane, Rep. Prog. Phys. {\bf 61}, 117
675: (1998); D. P. DiVincenzo and C. H. Bennett, Nature {\bf 404}, 247 (2000).
676:
677:
678: %--------- syatems efficiently simulated by classical computers ------------------
679:
680: \bibitem{osterloh} A.~Osterloh, L.~Amico, G.~Falci, and R.~Fazio, Nature 416, 608 (2002).
681: \bibitem{vidal} G.~Vidal, J.I.~Latorre, E.~Rico, and A.~Kitaev, \prl {\bf 90}, 227902 (2003).
682: %\bibitem{latorre} J.I.~Latorre,E.~Rico and G.~Vidal, Quantum Inf. Comput. {\bf 4}, 48 (2004).
683:
684: %---------- entanglement and QPT --------------------------------------------------
685:
686: \bibitem{zanardi} Y.~Chen, P.~Zanardi, Z.D.~Wang, and F.C.~Zhang, New J. Phys. (2006) quant-ph/0407228 .
687: \bibitem{others}S.-J~Gu, G.-S.~Tian, and H.-Q.~Lin, quanth-ph/0509070.
688: \bibitem{lidar} L.-A.~Wu, M.S.~Sarandy, and D.A.~Lidar, \prl {\bf 93}, 250404 (2004).
689:
690: \bibitem{carollo} A.~Carollo, and J.K.~Pachos, Phys. Rev. Lett. {\bf 95}, 157203 (2005); Shi-Liang
691: Zhu, Phys. Rev. Lett. {\bf{96}}, 077206 (2006) ; A. Hamma, quant-ph/0602091.
692:
693: \bibitem{cejnar} P. Cejnar, V. Zelevinsky, and V. V. Sokolov, Phys. Rev. E {\bf 63}, 036127 (2001).
694:
695: \bibitem{other} A. Volya, and V. Zelevinsky, Phys. Lett. B {\bf 574}, 27 (2003); V. Zelevinsky, B. A. Brown, N.
696: Fraizer, and M. Horoi, Phys. Rep. {\bf 276}, 85 (1996); P. Cejnar, and J. Jolie, Phys. Rev. E {\bf 58}, 387
697: (1998); P. Cejnar, and J. Jolie, Phys. Rev. E {\bf 61}, 6237 (2000).
698:
699:
700: %\bibitem{dicke} R. H. Dicke, Phys. Rev. {\bf 93}, 99 (1954).
701:
702:
703: \bibitem{anderson} P.W. Anderson, Phys. Rev. Lett. {\bf 18}, 1049
704: (1967).
705:
706:
707: \bibitem{zanardi-china} H.T. Quan, Z. Song, X.F. Liu, P. Zanardi and C.P.
708: Sun, Phys. Rev. Lett.
709: {\bf{96}}, 140604 (2006)
710:
711: %\bibitem{fid} Note that the square modulus of the overlap is nothing but
712: %the fidelity function widely used in quantum information \cite{qis};
713: %all the results of this paper could be the easily formulated
714: %in terms of fidelity as well.
715:
716:
717:
718: \bibitem{emary} C. Emary and T. Brandes, Phys. Rev. E, 066203
719: (2003).
720:
721: \bibitem{primakoff} T. Holstein and H. Primakoff, Phys. Rev. {\bf
722: 58} 1098 (1949).
723:
724: %\bibitem{PZ} N. Paunkovi\'{c}, P. Zanardi, in preparation.
725:
726: \bibitem{ITP} J. von Neumann, Comp. Math. {\bf 6}, 1 (1938); T Thiemann and O Winkler, Class. Quantum Grav. {\bf 18}
727: 4997 (2001).
728:
729:
730: \bibitem{sachdev-catastrophe} S. Sachdev., M. Troyer and M. Vojta, Phys. Rev. Lett. {\bf 86}, 2617 (2001)
731:
732: \bibitem{Loschmidt} P. R. Levstein, G. Usaj and H. M. Pastawski, J. Chem. Phys. 108, 2718 (1998); G. Usaj, H. M. Pastawski P. R. Levstein, Mol. Phys.95, 1229 (1998);
733: H. M. Pastawski, P. R. Levstein, G. Usaj, J. Raya and J. Hirschinger, Physica A 283, 166 (2000)
734:
735: \bibitem{Loschmidt1} Z. P. Karkuszewski, C. Jarzynski and W. H.
736: Zurek, Phys. Rev. Lett. {\bf 89}, 170405 (2002); F.M. Cucchietti,
737: D.A.R. Dalvit, J.P. Paz and W.H. Zurek, Phys. Rev. Lett. {\bf 91},
738: 210403 (2003).
739:
740:
741:
742: %\bibitem{wen} X-G.~Wen, {\em Quantum Field Theory of Many-Body Systems}, Oxford University Press (2004).
743:
744: \end{thebibliography}
745:
746: \end{document}
747: