1: \documentclass[twocolumn, showpacs]{revtex4}
2: \usepackage{graphics}
3: \usepackage{amsmath}
4: \usepackage{overpic}
5: \usepackage{hyperref}
6: \usepackage{rotating}
7: \newcommand{\ket}[1]{|#1\rangle}
8: \newcommand{\bra}[1]{\langle #1|}
9: \newcommand{\inp}[2]{\langle #1 | #2\rangle}
10:
11: \begin{document}
12: \title{Geometric Phase and Quantum Phase Transition in the Lipkin-Meshkov-Glick model}
13: \author{H. T. Cui}
14: \email{cuiht@student.dlut.edu.cn}
15: \author{K. Li, X. X. Yi} \affiliation{Department of
16: Physics, Dalian University of Technology, Dalian 116024, China}
17: \date{\today}
18: \begin{abstract}
19: The relation between the geometric phase and quantum phase
20: transition has been discussed in the Lipkin-Meshkov-Glick model. Our
21: calculation shows the ability of geometric phase of the ground state
22: to mark quantum phase transition in this model. The possibility of
23: the geometric phase or its derivatives as the universal order
24: parameter of characterizing quantum phase transitions has been also
25: discussed.
26: \end{abstract}
27: \pacs{75.10.Pq, 03.65.Vf, 05.30.Pr, 42.50.Vk}
28: \maketitle
29:
30: \section{introduction}
31: Recently the understanding of quantum phase transition
32: \cite{sachdev} has emerged from the fundamentals of quantum
33: mechanics, especially from the entanglement point of view
34: \cite{preskill}. The ground state structure have been shown to be
35: affected appreciably by the critical points, as illustrated
36: initially in the one-dimensional Ising model with transverse
37: magnetic field \cite{osterloh}. Intriguing by these pioneering
38: works, a great deal of efforts have attributed to this region
39: \cite{wu,vidal,gu}. It originates from the belief that quantum phase
40: transition should be connected to the entanglement or its
41: derivatives in many-body systems \cite{preskill}. However the
42: correspondence between the entanglement and quantum phase transition
43: in many-body systems is ambiguous \cite{yang, vidal1}, that is
44: because of the absence of the proper measurement of entanglement in
45: many-body systems. On the other hand, geometric phase \cite{berry}
46: as a measurement of the curvature of the Hilbert space, has been
47: first connected with the quantum phase transition in the
48: one-dimension spin-chain system \cite{carollo}, in which the
49: topological character of the relative geometric phase between the
50: first excited state and the ground state have been shown the ability
51: of detecting the critical points. Furthermore, Zhu extended this
52: study under the thermodynamic limit and found that the geometric
53: phase of the ground state in the one-dimensional XY model was
54: non-analytical and its derivative with the coupling constant was
55: divergent closed to critical points \cite{zhu}. Moreover the scaling
56: behavior of the geometric phase of ground state clearly
57: distinguished two different types of quantum phase transitions at
58: the critical point in the one-dimensional XY model. Consequently the
59: general theory about the relation between geometric phase of the
60: ground state and the critical point has been constructed
61: \cite{hamma}, in which the topological property of Berry's loops
62: including critical points in the parameter space has been shown the
63: ability of detecting the critical points.
64:
65: Although the theory is rapidly developing, the verifications of
66: examples only focus on the one-dimension XY spin-chain systems. The
67: reason is that XY model, which have been shown first-order phase
68: transition closed to the critical point, can be converted into
69: spinless fermionic system by Jordan-Wigner transformation, and the
70: energy spectrum can be determined exactly \cite{lieb}. However the
71: interacting between spins in this model is short-range (the
72: neighbor-nearest coupling) and the anisotropy of Heisenberg
73: interaction is indispensable for the construction of long-range
74: order \cite{lieb}. Recently the Berry phase in Dick model has been
75: examined under thermodynamics limit \cite{plastina}. The authors
76: showed that the Berry phase displayed the non-analyticity close to
77: critical point where this model exhibits a second-order phase
78: transition, and its derivative with the coupling constant was also
79: discontinued and showed a cusp closed to this point. The
80: discontinuity of Berry phase is very similar to that in XY model,
81: but we should point out that the two model belong to different
82: orders of quantum phase transitions respectively. In this paper, we
83: will show a special situation, in which the geometric phase of the
84: ground state behaves differently and itself distinguishes the
85: different quantum phase transitions.
86:
87: In particular we study the geometric phase in a system of spins with
88: a collective coupling described by the Lipkin-Meshkov-Glick (LMG)
89: model \cite{lmg}, which first introduced forty years ago in nuclear
90: physics. This model has received much attention because of his
91: apparent simplicity and popularity in the past. It provides a simple
92: description of the tunneling of bosons between two degenerate levels
93: and can thus be used to describe many physical systems such as
94: two-mode Bose-Einstein condensates \cite{cirac} or Josephson
95: junctions \cite{josephson}. More recently the entanglement in this
96: model has received great attention because of available numerical
97: calculations and plentiful phase diagram \cite{vidal2, vidal1}.
98: Under the thermodynamics limit, its phase diagram can be simply
99: established by a semiclassical approach \cite{botet}. For large
100: particle number N but finite, the situation is complicated and the
101: numerical analysis was implemented using the continuous unitary
102: transformations \cite{dusuel}. The significant difference between
103: this model and XY model is the long-range interaction, and the
104: system cannot be converted into the spinless fermionic system. Hence
105: it is of great interest to study the relation of geometric phase and
106: phase transition in this model. As will display in the remaining of
107: this paper, the geometric phase of ground state in this model
108: behaves differently and reflects faithfully the existence of the
109: critical points.
110:
111: The paper is organized as follows. In Sec. II, we first introduce
112: the LMG model, and in the limit of large N but finite, we obtain the
113: ground state analytically by the Holstein-Primakoff representation
114: and calculate the geometric phase. In order to show the university
115: of our results, in Sec. III, we study the phase transition in a more
116: complex situation. It is surprising that the geometric phase of
117: ground state detects very rigorously the critical points. Finally we
118: discuss the implications of our results and the differences from the
119: XY model.
120:
121: \section{the Lipkin-Meshkov-Glick model: biaxial case}
122: The LMG model describes a set of N spins half coupled to all others
123: with a strength independent of the position and the nature of the
124: elements and a magnetic field in the $z$ direction. The Hamiltonian
125: can be written
126: \begin{equation}\label{h1}
127: H= - \frac{1}{N}(S^2_x + \gamma S^2_y) - h S_z,
128: \end{equation}
129: in which $S_{\alpha}=\sum_{i=1}^{N}\sigma^i_{\alpha}/2 (\alpha=x, y,
130: z)$ and the $\sigma_{\alpha}$ is the Pauli operator, N is the total
131: particle number in this system. The prefactor $1/N$ is essential to
132: ensure the convergence of the free energy per spin in the
133: thermodynamic limit. For any anisotropy parameter $\gamma \in[0,1]$,
134: the Hamiltonian \eqref{h1} preserves the total spin and does not
135: couple the state having spin pointing in the direction perpendicular
136: to the field, namely
137: \begin{equation}
138: [H, \textbf{S}^2]=0, [H, \prod_{i=1}^{N}\sigma_z^i]=0.
139: \end{equation}
140: An important character that differentiates LMG model from the XY
141: model is the long-rang interaction between particles, which induces
142: a second-order phase transition at $h=1$ when $N\rightarrow\infty$
143: \cite{botet}.
144:
145: The diagonalization of Eq. \eqref{h1} can be obtained by
146: introducing the Holstein-Primakoff representation of the spin
147: operator and then truncate the resulting bosonic Hamiltonian to
148: lowest order \cite{dusuel}. Consequently we diagonalize it thanks to
149: the Bogoliubov transformation. The first thing is to perform a
150: rotation of the spin operators around the $y$ direction, that makes
151: the $z$ axis along the so-called semiclassical magnetization
152: \cite{dusuel} in which the Hamiltonian Eq. \eqref{h1} has the
153: minimal value in the semiclassical approximation. This can be done
154: as
155: \begin{equation}
156: \left( \begin{array}{c} S_x \\S_y \\S_z
157: \end{array} \right)= \left(
158: \begin{array}{ccc} \cos\theta & 0 & -\sin\theta \\ 0 & 1 & 0 \\
159: \sin\theta & 0 & \cos\theta \end{array} \right)
160: \left(\begin{array}{c} \tilde{S}_x \\ \tilde{S}_y \\
161: \tilde{S}_z \end{array} \right)
162: \end{equation}
163: in which $\theta=0$ for $h>1$, $\theta=\arccos h$ for $0<h<1$ .
164: Then the Hamiltonian Eq. \eqref{h1} is converted into
165: \begin{eqnarray}
166: H &=& - \frac{1}{N}[\sin^2\theta \tilde{S_z}^2 + \frac{\cos^2\theta
167: +\gamma}{2}(\tilde{S}^2- \tilde{S_z}^2)]\nonumber\\
168: &+&
169: \frac{\sin2\theta}{N}(\tilde{S}^{+}\tilde{S_z}+\tilde{S}^{-}\tilde{S_z}+
170: h.c.)-\frac{\cos^2\theta -
171: \gamma}{4N}(\tilde{S}^{+2}+\tilde{S}^{-2})\nonumber\\&-& h_z
172: \cos\theta\tilde{S}_z
173: -\frac{h_z}{2}\sin\theta(\tilde{S}^{+}+\tilde{S}^{-}),
174: \end{eqnarray}
175: in which $\tilde{S}^{\pm}=\tilde{S}_x \pm i\tilde{S}_y$.
176:
177: In order to obtain the geometric phase of ground state, we consider
178: the system has a rotation $\tilde{g}(\phi)$ around the new $z$
179: direction. The Hamiltonian becomes
180: \begin{equation}\label{h1phi}
181: H(\phi)=\tilde{g}(\phi)H\tilde{g}^{\dagger}(\phi).
182: \end{equation}
183: in which $\tilde{g}(\phi)=e^{i\phi\tilde{S}_z}$. Then we can use the
184: Holstein-Primakoff representation,
185: \begin{eqnarray}\label{hp}
186: \tilde{S}_z(\phi) &=& N/2 - a^{\dagger}a, \nonumber\\
187: \tilde{S}^{+}(\phi)&=&(N-a^{\dagger}a)^{1/2}a e^{i\phi},\nonumber\\
188: \tilde{S}^-(\phi)&=&a^{\dagger} e^{- i\phi}(N-a^{\dagger}a)^{1/2}
189: \end{eqnarray}
190: in which $a^{(\dagger)}$ is bosonic operator. Since the $z$ axis is
191: along the semiclassical magnetization, $a^{\dagger}a/N\ll 1$ is a
192: reasonable assumption under low-energy approximation, in which $N$
193: is large but finite. Keeping the terms of order $N, N^{1/2}, N^0$,
194: Eq. \eqref{h1phi} becomes
195: \begin{equation}
196: H(\phi)=Ne + \Delta a^{\dagger}a + \Gamma (a^{\dagger 2}e^{- 2i\phi}
197: + a^2 e^{2i \phi}),
198: \end{equation}
199: in which
200: \begin{eqnarray}
201: e&=&- \frac{1}{4}(\sin^2\theta + 2h\cos\theta),\nonumber\\
202: \Delta&=&\sin^2\theta - \frac{\gamma+
203: \cos^2\theta}{2}+h\cos\theta\nonumber\\
204: \Gamma&=&\frac{\gamma-\cos^2\theta}{4}
205: \end{eqnarray}
206: Obviously the equation above can be diagonalized by the standard
207: Bogoliubov transformation,
208: \begin{equation}\label{bt}
209: b(\phi)=\cosh x a e^{i\phi} + \sinh x a^{\dagger}e^{-i\phi}
210: \end{equation}
211: Then the Hamiltonian is
212: \begin{equation}
213: H_{diag}(\phi)=Ne+\sigma + \Delta^D b^{\dagger}(\phi)b(\phi),
214: \end{equation}
215: in which,
216: \begin{eqnarray}
217: \sigma&=&\frac{\Delta}{2}(\sqrt{1-\epsilon^2}-1)\nonumber\\
218: \Delta^D&=&\Delta\sqrt{1 - \epsilon^2}\nonumber\\
219: \epsilon&=&\frac{2\Gamma}{\Delta}=\tanh2x=
220: \begin{cases}-\frac{1-\gamma}{2h-1-\gamma}, & h>1 \\
221: -\frac{h^2-\gamma}{2-h^2-\gamma}, & 0<h<1 \end{cases}
222: \end{eqnarray}
223: It should be careful about the physical interpretation of
224: $\Delta^D$, which may not describe true gap of the system
225: \cite{dusuel}.
226:
227: Now it is time to determine the ground state $\ket{g(\phi)}$, which
228: can be obtained by applying the relation,
229: \begin{equation}
230: b(\phi)\ket{g(\phi)}=0
231: \end{equation}
232: Substituting Eq. \eqref{bt} into the equation above, one obtains the
233: ground state,
234: \begin{widetext}
235: \begin{eqnarray}
236: \ket{g(\phi)}=\frac{1}{C}\sum_{n=0}^{[N/2]}\sqrt{\frac{(2n-1)!!}{2n!!}}(-
237: \frac{e^{-i\phi}\sinh x}{e^{i\phi}\cosh x
238: })^{n-1}(-\sqrt{2}e^{-i\phi}\sinh x )\ket{2n},
239: \end{eqnarray}
240: \end{widetext}
241: in which $n!!=n(n-2)(n-4)\cdots$ and $n!!=1$ for $n\leq0$. $\ket{n}$
242: is the Fock state of bosonic operator $a^{(\dagger)}$ and the
243: normalized constant is
244: $C^2=\sum_{n=0}^{[N/2]}2\sinh^2x\frac{(2n-1)!!}{2n!!}\tanh^{2(n-1)}x$.
245: One should note that in order that the summation is convergent,
246: $|\tanh x| \leq 1$.
247:
248: \begin{figure}
249: \begin{overpic}{1}
250: \put(65, 5){$h$} \put(80, 70){$\gamma$} \put(100, 40){$\varphi_g$}
251: \end{overpic}
252: \caption{\label{g1} The geometric phase $\varphi_g$ [Arc] vs. the
253: anisotropic parameter $\gamma$ and $h$. For specification, we have
254: chosen the summation from 0 to 100 in the expression of $\varphi_g$
255: (Eq. \eqref{g}). The divergent character of $\varphi_g$ is clearly
256: displayed at $h\rightarrow1$ in this figure.}
257: \end{figure}
258: The geometric phase of the ground state, accumulated by changing
259: $\phi$ from $0$ to $\pi$, is determined by
260: $\varphi_g=-i\int_0^{\pi}d\phi\bra{g(\phi)}\partial_{\phi}\ket{g(\phi)}$.
261: The direct calculation shows
262: \begin{equation}\label{g}
263: \varphi_g=\pi[1 -
264: \frac{\sum_{n=0}^{[N/2]}2n\frac{(2n-1)!!}{2n!!}\tanh^{2(n-1)}x}
265: {\sum_{n=0}^{[N/2]}\frac{(2n-1)!!}{2n!!}\tanh^{2(n-1)}x}]
266: \end{equation}
267: which have been plotted with $\gamma$ and $h$ in Fig. \ref{g1}. It
268: is obvious that $\varphi_g$ is divergent at the point $h=1$ where
269: the LMG model has been proved to experience a second-order phase
270: transition, independent of the anisotropy $\gamma$\cite{botet}. This
271: divergency has never been explored in the previous studies
272: \cite{carollo, hamma, zhu, plastina} and shows distinguished
273: character from the XY and Dicke models. In fact, since the
274: adiabatic condition has been destroyed when $h\rightarrow 1$ because
275: of the degeneracy between the ground state and the excited state
276: and the degeneracy enhances geometric phase \cite{berry}, this
277: phenomenon is natural. However, the essential reason is the
278: collective interaction in the LMG model, which is absent in the XY
279: model and makes the long-range correlations in this system.
280:
281: \begin{figure}[t]
282: \begin{overpic}[width=4cm]{gamma=0.5.eps}
283: \put(15,60){(a)}\put(50, -5){$h$} \put(-5, 35){$\varphi_g$}
284: \end{overpic}\vspace{1em}
285: \begin{overpic}[width=4cm]{gamma1.eps}
286: \put(15,60){(b)}\put(50, -5){$h$} \put(-5, 35){$\varphi_g$}
287: \end{overpic}
288: \caption{\label{g11}$\varphi_g$ [Arc] vs. $h$ with different
289: particle number $N$. We have chosen $\gamma=0.5$ (a) and $\gamma=1$
290: (b) for this plot. The dashed and solid lines correspond
291: respectively to $N=4, 1000$.}
292: \end{figure}
293:
294: The finite size effect is also examined in this case by choosing
295: different $N$, which can be shown in the Fig. \ref{g11}(a) and (b).
296: In this figure, we only draw for $N=4, 1000$ and for higher values
297: of $N$, the curves almost do not change. From the figures we note
298: that the divergency of $\varphi_g$ seems to be insensitive to $N$.
299: However, since phase transition only happens under thermodynamic
300: limit, this phenomenon attributes to the ground-state degeneracy at
301: the critical point.
302:
303:
304: \begin{figure}
305: \begin{overpic}[width=8cm]{scale}
306: \put(50, 0){N} \put(0, 35){$\varphi_g$} \end{overpic}
307: \caption{\label{scale} The scaling behavior of $\varphi_g$ [Arc] vs.
308: $N$ with $\gamma=0.5$ when $h\rightarrow1$. The slope of the line is
309: close to $-1$. }
310: \end{figure}
311: The scaling behavior of $\varphi_g$ plotted with $N$ has also been
312: explored in Fig. \ref{scale}. It is obvious that one can obtain
313: \begin{equation}
314: \varphi_g\approx- N.
315: \end{equation}
316: Furthermore the scaling is independent of $\gamma$, which means that
317: for different $\gamma$, the phase transitions belong to the same
318: university class. We also check the scaling behavior when $\gamma=1$
319: and find that this case has the same behavior as displayed in Fig.
320: \ref{scale}. It means that the phase transition for $\gamma=1$ is
321: the same university class as that for $\gamma\neq1$. This phenomenon
322: is different from the XY model, in which the isotropic and
323: anisotropic interactions belong respectively to different university
324: classes \cite{zhu}, that comes from the collective interaction in
325: the LMG model.
326:
327: \section{uniaxial model}
328: In this part, we will discuss the phase transition in a more complex
329: system, which is the generalization of the LMG model Eq. \eqref{h1}.
330: The Hamiltonian can be written as
331: \begin{equation}\label{h2}
332: H = - \frac{1}{N}S^2_x - h_x S_x -h_z S_z,
333: \end{equation}
334: with $h_z>0$. The phase diagram of this model is obviously dependent
335: on the both parameters $h_x, h_z$, and moreover a proper order
336: parameter characterizing the transition is difficult to build.
337: Recently the correspondence between this model and a two-level boson
338: problem introduced in nuclear physics has been constructed, which
339: permits one to get an order parameter \cite{vidal3}. The phase
340: transition is then clear in this model; a first-order transition
341: occurs at $h_x=0$ with $h_z<1$ and a second-order one occurs at
342: $h_x=0$ with $h_z=1$. For $h_z>1$ or $h_x\neq0$, no transition is
343: found.
344:
345: Now we will try to determine the phase transition by calculating the
346: geometric phase of the ground state. Similar procedures as in
347: previous section can be applied for this purpose. The first step is
348: to make the system have a rotation around the $z$ direction and then
349: the Hamiltonian \ref{h2} becomes
350: $H(\phi)=g(\phi)Hg^{\dagger}(\phi)$. Next step is to introduce the
351: Holstein-Primakoff transformation Eq. \eqref{hp}. One should note
352: that the approximation $a^{\dagger}a/N \ll 1$ is invalid in this
353: case since we cannot find the semiclassical magnetization. However a
354: simple canonical transformation can be used to resolve this problem,
355: $a^{(\dagger)}e^{(-)i\phi}=b^{(\dagger)}e^{(-)i\phi}+
356: \sqrt{N}\lambda$ with $|\lambda|<1$. This transformation provides a
357: macroscopic expectation value of $S_z$ which is order of $N$, and
358: then one has $b^{\dagger}b/N\ll1$. Under the limit that $N$ is large
359: but finite, it is enough to expand $H(\phi)$ to the order $N^0$.
360: After the exhausted calculation, $H(\phi)$ is written as,
361: \begin{eqnarray}\label{h2phi}
362: H(\phi)=&&e_0(\lambda)+
363: \Omega(\lambda)(be^{i\phi}+b^{\dagger}e^{-i\phi})+\nonumber\\
364: &&\Gamma(\lambda)(b^2e^{2i\phi}+b^{\dagger
365: 2}e^{-2i\phi})+\Delta(\lambda) b^{\dagger}b,
366: \end{eqnarray}
367: in which,
368: \begin{eqnarray}
369: e_0(\lambda)&=& -N[\frac{h_z}{2}(1 -
370: 2\lambda^2)+\lambda^2(1-\lambda^2)+h_x(1-\lambda^2)]\nonumber\\&-&
371: (\frac{1}{4}-\lambda^2)- h_x\frac{\lambda(2-\lambda^2)}{8(1 -
372: \lambda^2)^{3/2}}\nonumber\\
373: \Omega(\lambda)&=&\sqrt{N} [\lambda h_z - \frac{h_x(1-
374: 2\lambda^2)}{2\sqrt{1-\lambda^2}}-\lambda(1-2\lambda^2)]\nonumber\\
375: \Gamma(\lambda)&=&-\frac{1-
376: 5\lambda^2}{4}+h_x\frac{\lambda(2-\lambda^2)}{8(1-\lambda^2)^{3/2}}\nonumber\\
377: \Delta(\lambda)&=&h_z -
378: \frac{1-7\lambda^2}{2}+h_x\frac{\lambda(4-3\lambda^2)}{4(1-\lambda^2)^{3/2}}.
379: \end{eqnarray}
380: The crucial step is to choose $\lambda_0$ properly in order that the
381: linear term (the second term in Eq. \eqref{h2phi}) is vanishing. It
382: can be realized by solving the following equation,
383: \begin{equation}
384: \lambda_0 h_z - \frac{h_x(1-
385: 2\lambda_0^2)}{2\sqrt{1-\lambda_0^2}}-\lambda_0(1-2\lambda_0^2)=0.
386: \end{equation}
387: In fact the equation above can be reduced into the biquadratic
388: equation $(h_z - y)^2(1-y^2)-h_x^2y^2=0$ with $y=1- 2\lambda_0^2$,
389: which can be solved numerically.
390:
391: Substitute $\lambda_0$ into Eq. \eqref{h2phi} and then one get the
392: quadratic Hamiltonian,
393: \begin{equation}
394: H=e_0(\lambda_0)+\Gamma(\lambda_0)(b^2e^{2i\phi}+b^{\dagger
395: 2}e^{-2i\phi}) +\Delta(\lambda_0)b^{\dagger}b,
396: \end{equation}
397: which obviously could be diagonalized by the standard Bogoliubov
398: transformation. Consequently the geometric phase of ground state can
399: be determined directly, which has the same form as Eq. \eqref{g},
400: but different definition of $x$, determined by the equation $\tanh
401: 2x=\frac{2\Gamma(\lambda_0)}{\Delta(\lambda_0)}$.
402:
403: \begin{figure}
404: \begin{overpic}{2}
405: \put(34, 10){$h_z$} \put(90, 20){$h_x$} \put(5, 55){ $-\varphi_g$}
406: \end{overpic}
407: \caption{\label{g2} The geometric phase $\varphi_g$ [Arc] vs. $h_x$
408: and $h_z$. We have chosen the summation from 0 to 100 in the
409: expression of $\varphi_g$ Eq. \eqref{g}, and for the convenience of
410: viewport, we have drawn for $-\varphi_g$ in this plot.}
411: \end{figure}
412: \begin{figure}[t]
413: \begin{overpic}[width=4cm]{hz=0.5.eps}
414: \put(10,65){(a)}\put(50, -5){$h_x$} \put(-5, 35){$\varphi_g$}
415: \end{overpic}\vspace{1em}
416: \begin{overpic}[width=4cm]{hz=1.eps}
417: \put(10,65){(b)}\put(50, -5){$h_x$} \put(-5, 35){$\varphi_g$}
418: \end{overpic}
419: \begin{overpic}[width=4cm]{hz=2.eps}
420: \put(10,65){(c)}\put(50, -5){$h_x$} \put(-5, 35){$\varphi_g$}
421: \end{overpic}
422: \caption{\label{g22}The geometric phase $\varphi_g$ [Arc] vs. $h_x$
423: with different $h_z$. We have chosen $h_z=0.5$ (a), $h_z=1$ (b) and
424: $h_z=2$ (c) for this plot. The figure (b) has been compressed
425: because of the divergency of the value $\varphi$.}
426: \end{figure}
427: A schematic demonstration of the geometric phase is presented in
428: Fig. \ref{g2}. It is obvious that there are two regions divided by
429: $h_z=1$, and the geometric phase is divergent at point $h_x=0,
430: h_z=1$. In the region $h_z<1$, geometric phase is non-analytical at
431: point $h_x=0$, which means the appearance of phase transition, and
432: in the other region, the geometric phase is the smooth function of
433: $h_z, h_x$ and no phase transition is found from the figure. This
434: phenomenon is consistent with the conclusion in Ref. \cite{vidal1},
435: but in our calculation we do not need to find a proper order
436: parameter to characterize the phase transition and the geometric
437: phase of ground state faithfully marks these transitions. A detailed
438: demonstration is also provided in Fig. \ref{g22}. From the figures,
439: one easily finds that the geometric phase $\varphi_g$ has a cusp at
440: $h_x=0$ for $h_z<1$ (see Fig. \ref{g22}(a)), which implies the
441: first-order phase transition. Furthermore $\varphi_g $ is divergent
442: at $h_x=0$ when $h_z=1$ (see Fig. \ref{g22}(b)), which is similar to
443: that in the standard LMG model (see Figs. \ref{g1}and \ref{g11}) and
444: means that there is a second-order transition. For $h_z>1$
445: $\varphi_g$ is the smooth function of $h_x, h_z$ and there is no
446: transition (see Fig. \ref{g22}(c)).
447:
448: It is surprising that the geometric phase of ground state itself can
449: differentiate phase transition without need of introducing a proper
450: order parameter, which may be difficult to find. Thus it is a
451: natural speculation that the geometric phase of ground state can
452: serve as a universal order parameter. Further discussions will be
453: presented in the next part.
454:
455: \section{conclusions and discussions}
456: The geometric phases of ground state in the LMG model and its
457: generalization have been discussed in this paper. Our calculations
458: show that the geometric phase faithfully reflects the phase
459: transition in this model; when there is a second-order phase
460: transition, the geometric phase behaves divergent (see Figs.
461: \ref{g1} and \ref{g22}(b)). However when there is a first-order
462: transition, non-analyticity of geometric phase appears at the
463: critical point (see Fig. \ref{g22}(a)). These phenomena may come
464: from the degeneracy of the ground state in the system, which usually
465: induces the mixture of different phases. The different behaviors of
466: geometric phase closed to critical points could originate from the
467: different excitations (i.e. gapped or gapless). Furthermore we find
468: that the energy of ground state is not a good parameter of marking
469: the phase transition, because the energy of ground state is
470: degenerate independent of the phase transition is second-order or
471: first-order. Recently Tian and Lin have shown that the continuous
472: quantum phase transition are actually caused by level crossing of
473: the low-lying excited states of the system \cite{tian}. This
474: conclusion also shows that the only energy of the ground state is
475: not suitable for the characterization of quantum phase transition.
476:
477: This leads to a question what physical quantity is suitable for
478: characterizing the quantum phase transition. With respect of the
479: work Ref \cite{zhu} and our calculations, it seems to provide us an
480: hint that the geometric phase of ground state or its derivatives
481: could serve as an universal order parameter to characterize
482: different phase transitions. This speculation is natural since the
483: geometric phase faithfully measure the curvature of the Hilbert
484: space (or phase space for classical mechanics) and the broken of
485: symmetry of Hilbert space must be reflected in the geometric phase.
486:
487: Another aspect of importance is the differences between our model
488: and the one-dimensional XY model. The crucial point is the
489: collective interaction in LMG model, which is absent in the XY
490: model. A main result of this interaction is that the LMG model
491: cannot be converted into the spinless fermion system. This fact
492: makes the conclusion different from that in Ref. \cite{hamma}, in
493: which the topological behavior of Berry phase can detect the
494: critical point. However, in this model, the geometric phase is
495: divergent when there is second-order transition and the detection
496: maybe invalid.
497:
498: In conclusion the relation between the geometric phase of ground
499: state and the phase transition in LMG model has been constructed in
500: this paper. Different from the results in the one-dimensional XY
501: model, the singularity of the geometric phase itself can serve as
502: the signature of critical points. Moreover we discuss the
503: possibility of the geometric phase of ground state or its
504: derivatives serving as the universal order parameter to characterize
505: the quantum phase transitions.
506:
507: This work was supported by NSF of China under grants 10305002 and
508: 60578014.
509: \begin{thebibliography}{99}
510:
511: \bibitem{sachdev}Subir Sachdev, \textit{Quantum Phase Transition}(Cambridge University Press, Cambridge, 1999).
512:
513: \bibitem{preskill} J. Preskill, J. Mot. Opt. {\bf 47}, 127 (2000).
514:
515: \bibitem{osterloh} A. Osterloh, L. Amico, G. Falci, R. Fazio, Nature, {\bf 416}, 608(2002);
516: T. J. Osborne, M. A. Nielsen, Phys. Rev. A. {\bf 66}, 032110(2002).
517:
518: \bibitem{wu}G. Vidal, J. I. Latorre, E. Rico, A. Kitaev,
519: Phys. Rev. Lett. {\bf 90}, 227902 (2003) ; L.-A. Wu, M.S. Sarandy,
520: and D.A. Lidar, Phys. Rev. Lett. {\bf 93},250404 (2004).
521:
522: \bibitem{vidal}J.I. Latorre, E. Rico and G. Vidal, Quantum Inf. Comput. {\bf 4}, 48
523: (2004).
524:
525: \bibitem{gu} S.-J Gu, G.-S. Tian, and H.-Q. Lin, quant-ph/0509070.
526:
527: \bibitem{yang} M. F. Yang, Phys. Rev. A {\bf 71}, 030302(2005).
528:
529: \bibitem{vidal1} J. Vidal, Phys. Rev. A {\bf 73}, 062318(2006).
530:
531: \bibitem{berry} M. V. Berry, Proc. R. Soc. London A {\bf 392}, 45(1984).
532:
533: \bibitem{carollo}Angelo C. M. Carollo, J. K. Pachos, Phys. Rev. Lett. {\bf 95},
534: 157203(2005);J. K. Pachos, Angelo C. M. Carollo, arXiv:
535: quant-ph/0602154.
536:
537: \bibitem{zhu}S. L. Zhu, Phys. Rev. Lett. {\bf 96}, 077206(2006).
538:
539: \bibitem{hamma}A. Hamma, arXiv: quant-ph/0602091.
540:
541: \bibitem{lieb} E. Lieb, T. Schultz, D. Mattis, Ann. Phys. {\bf 16} 407(1961).
542:
543: \bibitem{plastina} F. Plastina, G. Liberti, A. Carollo, arXiv: quant-ph/0604011.
544:
545: \bibitem{lmg} H. J. Lipkin, N. Meshkov, A. J. Glick, Nucl. Phys. {\bf
546: 62}, 188 (1965); {\bf 62}, 199 (1965); {\bf 62}, 211(1965).
547:
548: \bibitem{cirac} J. I. Cirac, M. Lewenstein, K. M{\o}lmer, P. Zoller,
549: Phys. Rev. A {\bf 57}, 1208 (1998).
550:
551: \bibitem{josephson} B. D. Josephson, Phys. Lett. {\bf 1}, 251
552: (1960); J. W. Rowell, Phys. Rev. Lett. {\bf 11}, 200 (1963).
553:
554: \bibitem{vidal2} J. Vidal, G. Palacios, and R. Mosseri, Phys. Rev. A {\bf 69}, 022107
555: (2004); J. Vidal, R. Mosseri, and J. Dukelsky, Phys. Rev. A {\bf
556: 69}, 054101 (2004); S. Dusuel and J. Vidal, Phys. Rev. Lett. {\bf
557: 93}, 237204 (2004); R. G. Unanyan, C. Ionescu, M. Fleischhauer,
558: Phys. Rev. A {\bf 72}, 022326 (2005).
559:
560: \bibitem{botet} R. Botet, R. Jullien. P. Pfeuty, Phys. Rev. Lett. {\bf
561: 49}, 478 (1982); R. Botet, R. Jullien, Phys. Rev. B {\bf 28}, 3955
562: (1982).
563:
564: \bibitem{dusuel} S. Dusuel, J. Vidal, Phys. Rev. Lett. {\bf 93},
565: 237204 (2004); Phys. Rev. B {\bf 71}, 224420 (2005).
566:
567: \bibitem{vidal3} The Appendix A in Ref. \cite{vidal1}; J. Vidal, J.
568: M. Arias, J. Dukelsky, J. Garc\'{\i}a-Ramos, Phys. Rev. C {\bf 73},
569: 054305(2006).
570:
571: \bibitem{tian} Guang-shan Tian, Hai-Qing Lin, Phys. Rev. B {\bf 67},
572: 245105(2003).
573: \end{thebibliography}
574: \end{document}
575: