1: % SJGu: Aug. 7; GSTian: Aug. 10; HQLin: Aug. 14
2: %
3: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
4: % TeX file of the article: %
5: % %
6: % %
7: % Title:
8: % authors:
9: % %
10: % Journal Objective: Physics Review A %
11: % %
12: % Subject: New submission %
13: % %
14: % address: %
15: % %
16: % %
17: % e-mail: %
18: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
19:
20:
21: \documentclass[twocolumn,showpacs,preprintnumbers,amsmath,amssymb]{revtex4}
22: %\documentclass[preprint,showpacs,preprintnumbers,amsmath,amssymb]{revtex4}
23: \usepackage{graphicx}% Include figure files
24: \usepackage{dcolumn}% Align table columns on decimal point
25: \usepackage{bm}% bold math
26:
27: %\nofiles
28:
29: \begin{document}
30:
31: %\preprint{APS/123-QED}
32:
33: \title{The ground state entanglement in the $XXZ$ model}
34:
35: \author{Shi-Jian Gu$^{1}$}
36: \author{Guang-Shan Tian$^{1,2}$}
37: \author{Hai-Qing Lin$^{1}$}
38: \affiliation{$^1$Department of Physics, The Chinese University of
39: Hong Kong, Hong Kong, China} \affiliation{$^2$School of Physics,
40: Peking University, Beijing 100871, China}
41:
42: \begin{abstract}
43: In this paper, we investigate spin entanglement in the $XXZ$ model
44: defined on a $d$-dimensional bipartite lattice. The concurrence, a
45: measure of the entanglement between two spins, is analyzed. We
46: prove rigorously that the ground state concurrence reaches maximum
47: at the isotropic point. For dimensionality $d \ge 2$, the
48: concurrence develops a cusp at the isotropic point and we
49: attribute it to the existence of magnetic long-range order.
50: \end{abstract}
51: \pacs{03.67.Mn, 03.65.Ud, 05.70.Jk, 75.10.Jm}
52:
53: % 03.67.Mn Entanglement production, characterization,
54: % and manipulation (see also 03.65.Ud Entanglement and quantum nonlocality;
55: % for entanglement in Bose-Einstein condensates, see 03.75.Gg)
56: % 03.65.Ud Entanglement and quantum nonlocality
57: % (e.g. EPR paradox, Bell's inequalities, GHZ states, etc.)
58: % (for entanglement production in quantum information, see 03.67.Mn;
59: % for entanglement in Bose-Einstein condensates, see 03.75.Gg)
60: % 05.70.Jk Critical point phenomena
61: % 75.10.Jm Quantized spin models
62:
63: \maketitle
64:
65: Entanglement, as the exhibition of pure quantum correlations
66: between separate systems, has become one of the trademarks of the
67: quantum mechanics for its nonlocal
68: connotations\cite{ABinstein35,MANielsenb}. Recently, many
69: physicists have made great efforts to understand the quantum
70: entanglement in the ground states of some many-body spin models
71: \cite{AOsterloh2002,TJOsbornee,IBose02,GVidal2003,SJGu03,JVidal04,LAWu04,MFYang04,YChen04}.
72: One expects that a thorough investigation on the entanglement in
73: these systems will provide new insight into the quantum phase
74: transition in these systems \cite{Sachdev}. For example, Osterloh
75: {\it et al}\cite{AOsterloh2002} studied the concurrence, a measure
76: of entanglement of two qubits\cite{Hill}, between two spins
77: located on a pair of nearest-neighbor sites in the
78: transverse-field Ising model\cite{TJOsbornee}. They found that
79: this quantity shows singularity and obeys the scaling law in the
80: vicinity of the quantum phase transition point of the system. On
81: the other hand, for other models, such as the antiferromagnetic
82: $XXZ$ chain, the concurrence behaves in a completely different
83: way\cite{SJGu03}. As shown by Ref. \cite{SJGu03}, the concurrence
84: is a continuous function of the anisotropic parameter and reaches
85: its maximum at the transition point. Therefore, in both cases, one
86: observes that the concurrence itself manifests interesting
87: behaviors at the quantum phase transition points. However, we
88: should emphasize that, such behaviors alone may not always signal
89: a quantum phase transition, as pointed out by the authors of
90: Refs.~\cite{LAWu04,MFYang04}.
91:
92: In Ref.~\cite{SJGu03}, we studied extensively nearest-neighbor
93: spin entanglement in the antiferromagnetic $XXZ$ chain. By
94: applying results derived from the Bethe ansatz solution of the
95: model, we showed clearly that the concurrence between two spins
96: located on a pair of nearest-neighbor sites in the system is a
97: continuous function of the anisotropic coupling parameter and
98: becomes maximal at the isotropic Heisenberg point. In this paper,
99: we continue our discussions on this issue. Our main purpose is to
100: show that some fundamental properties of the $XXZ$ model, such as
101: non-degeneracy and concavity of the ground state energy of the
102: system at the phase transition point, commands strongly on the
103: behavior of the concurrence. Therefore, we expect that the same
104: scenario will appear in a wide class of localized spin models,
105: such as the spin ladder model and, in particular, the $XXZ$ model
106: in higher dimensions~\cite{OFSyljuasenm}. It is well known that,
107: as far as the above-mentioned properties are concerned, the ground
108: states of these models are akin to the antiferromagnetic $XXZ$
109: chain.
110:
111: This paper contains two parts. In the first part, based on some
112: well-known facts about the antiferromagnetic spin models, we prove
113: rigorously that, when the antiferromagnetic $XXZ$ model is defined
114: on a $d$-dimensional {\it finite} bipartite lattice, the
115: concurrence between two spins located on a pair of
116: nearest-neighbor sites is an analytical function of the
117: anisotropic parameter and takes on its maximum at the Heisenberg
118: isotropic point. Then, in the second part of this paper, justified
119: by the existence of magnetic long-range order (LRO) in the $XXZ$
120: model, we use the spin-wave theory to show that a cusp-like
121: behavior of the concurrence develops in the thermodynamic limit
122: when the dimensionality of the lattice $d\ge 2$.
123:
124: To begin with, we first introduce several notations. On a finite
125: $d$-dimensional simple cubic lattice $\Lambda$ with
126: $N_\Lambda=L^d$ sites, the Hamiltonian of the antiferromagnetic
127: $XXZ$ model is
128: \begin{equation}
129: \hat{H}_{XXZ} = \sum_{\langle{\bf ij}\rangle}
130: \left(\hat{S}_{\bf i}^x \hat{S}_{\bf j}^x
131: + \hat{S}_{\bf i}^y S_{\bf j}^y + \Delta \hat{S}_{\bf i}^z
132: \hat{S}_{\bf j}^z\right),
133: \end{equation}
134: where $\hat{S}_{\bf i}^x,\>\hat{S}_{\bf i}^y$ and $\hat{S}_{\bf
135: i}^z$ are spin-1/2 operators at site $\bf i$ and
136: $\Delta=J_z/J_{x}\; (J_x=J_y)$ is a dimensionless parameter
137: characterizing the anisotropy of the model. The sum in the
138: Hamiltonian is over all pairs of nearest-neighbor sites $\bf i$
139: and $\bf j$. Obviously, this Hamiltonian commutes with the total
140: spin $z$-component operator $\hat{S}^z_{\rm total}=\sum_{\bf
141: i}\hat{S}^z_{\bf i}$. Thus, each eigenstate of the Hamiltonian is
142: also an eigenstate of $\hat{S}^z_{\rm total}$. Consequently, the
143: Hilbert space of the system can be decomposed into numerous
144: subspaces $V(M)$. In each subspace, the spin number
145: $\hat{S}^z_{\rm total}=M$ is specified. It is well known that, on
146: a finite simple cubic lattice $\Lambda$, the ground state of the
147: $XXZ$ model is nondegenerate in any admissible subspace $V(M)$
148: \cite{Lieb,Affleck}. In particular, its global ground state
149: $\Psi_0(\Lambda,\>\Delta)$, which coincides with the ground state
150: of the model in the subspace $V(M=0)$ \cite{Affleck}, is also
151: nondegenerate. Therefore, all the physical quantities, such as the
152: ground state energy $E_0(\Lambda,\>\Delta)$ and the spin
153: correlation function $\langle\hat{S}_{\bf i}^z\hat{S}_{\bf
154: i}^z\rangle$ are analytical functions of the parameter $\Delta$,
155: as long as the lattice is finite.
156:
157: The conservation of $\hat{S}_{\rm total}^z$ implies also that,
158: with respect to the standard basis vectors
159: $\vert\uparrow\uparrow\rangle$,
160: $\vert\uparrow\downarrow\rangle$,
161: $\vert\downarrow\uparrow\rangle$ and
162: $\vert\downarrow\downarrow\rangle$,
163: the reduced density matrix of two spins on a pair of nearest-neighbor
164: lattice sites $\bf i$ and $\bf j$ can be put
165: into the following block-diagonal form
166: \begin{equation}
167: \hat{\rho}_{\bf ij} =
168: \left(
169: \begin{array}{llll}
170: u^+ & 0 & 0 & 0 \\
171: 0 & w_1 & z & 0 \\
172: 0 & z^* & w_2 & 0 \\
173: 0 & 0 & 0 & u^-
174: \end{array}
175: \right).
176: \end{equation}
177: As a result, the concurrence of the two spins is $C_{\bf
178: ij}=2\max\left(\vert z\vert-\sqrt{u^+
179: u^-},\>0\right)$\cite{KMOConnor2001}. In terms of the correlation
180: function $G_{\bf
181: ij}^{\alpha\alpha}=\langle\hat{S}^\alpha\hat{S}^\alpha\rangle,\>
182: \alpha=x,y,z$, it can be explicitly written as\cite{XWang2002PLA}
183: \begin{eqnarray}
184: C_{\bf ij} =2\max\left(\left| G_{\bf ij}^{xx} + G_{\bf ij}^{yy}\right|
185: - G_{\bf ij}^{zz} - \frac{1}{4},\>0 \right).
186: \label{eq:concurrence}
187: \end{eqnarray}
188: By the variational principle, one can show that all the spin
189: correlations functions $G_{\bf ij}^{\alpha\alpha}$ are negative.
190: Thus, one has
191: \begin{eqnarray}
192: C_{\bf ij}
193: & = &
194: \left(- G_{\bf ij}^{xx} - G_{\bf ij}^{yy} - \frac{1}{4}
195: - G_{\bf ij}^{zz}\right) \nonumber\\
196: & = &
197: \left(-\epsilon_{\bf ij}^0(\Lambda,\>\Delta)
198: - \frac{1}{4} + (\Delta - 1) G_{\bf ij}^{zz}\right),
199: \end{eqnarray}
200: where $\epsilon_{\bf ij}^0(\Lambda,\>\Delta)=
201: E_0(\Lambda,\>\Delta)/N_B$ ($N_B$ is the number of bonds in the
202: lattice) is the ground state energy density per bond. Furthermore,
203: since all quantities in $C_{\bf ij}$ are analytical functions of
204: the parameter $\Delta$, we are allowed to take derivatives of it
205: with respect to $\Delta$. In particular, after taking the first
206: order derivative of $C_{\bf ij}$, we obtain
207: \begin{equation}
208: \frac{\partial C_{\bf ij}}{\partial\Delta}
209: = 2\left(- \frac{\partial\epsilon_{\bf ij}^0(\Lambda,\>\Delta)}
210: {\partial\Delta} + G^{zz}_{\bf ij}
211: + \left(\Delta-1\right) \frac{\partial G^{zz}_{\bf ij}}
212: {\partial\Delta} \right).
213: \label{Concurrence Derivative}
214: \end{equation}
215: Again, due to the nondegeneracy of the global ground state
216: $\Psi_0(\Lambda,\>\Delta)$ of the $XXZ$ model on a finite lattice,
217: we can use the Hellman-Feynman theorem to calculate the derivative
218: $\partial\epsilon_{\bf ij}^0(\Lambda,\>\Delta)/\partial\Delta$,
219: which equals to $G^{zz}_{\bf ij}$. Therefore, we finally obtain
220: \begin{eqnarray}
221: \frac{\partial C_{\bf ij}}{\partial\Delta}
222: & = &
223: 2 \left(\Delta-1\right) \frac{\partial G^{zz}_{\bf ij}}
224: {\partial\Delta} \nonumber\\
225: & = &
226: 2 \left(\Delta-1\right) \frac{\partial^2
227: \epsilon_{\bf ij}^0(\Lambda,\>\Delta)}
228: {\partial\Delta^2}.
229: \label{First Derivative}
230: \end{eqnarray}
231: Immediately, one sees that $\Delta=1$ is an extreme point of the
232: concurrence.
233:
234: Next, we show that $\Delta=1$ is actually a maximal point of
235: $C_{\bf ij}$ and the concurrence does not have other extreme
236: point. In fact, both the statements are the corollaries of
237: concavity of the ground state energy $E_0(\Lambda,\>\Delta)$ of
238: the Hamiltonian $\hat{H}_{XXZ}$ with respect to the anisotropic
239: parameter $\Delta$. By the variational principle\cite{Franklin},
240: we know that, for any two parameters $\Delta_1$ and $\Delta_2$,
241: the following inequality
242: \begin{equation}
243: E_0(\Lambda,\>\lambda\Delta_1 + (1 - \lambda)\Delta_2)
244: \ge \lambda E_0(\Lambda,\>\Delta_1)
245: + (1 - \lambda) E_0(\Lambda,\>\Delta_2),
246: \label{Concavity}
247: \end{equation}
248: where $0\le\lambda\le 1$, holds true for the ground state energy
249: $E_0(\Lambda,\>\Delta)$. In particular, when
250: $E_0(\Lambda,\>\Delta)$ is differentiable with respect to
251: $\Delta$, inequality (\ref{Concavity}) is equivalent to
252: \begin{equation}
253: \frac{\partial^2 E_0(\Lambda,\>\Delta)}{\partial\Delta^2}
254: \le 0.
255: \label{Second Derivative}
256: \end{equation}
257: Consequently, we have also
258: $\partial^2\epsilon_{\bf ij}^0(\Lambda,\>\Delta)/\partial\Delta^2\le 0$.
259: Now, let us take derivative of Eq.~(\ref{First Derivative})
260: again with respect to $\Delta$. It yields
261: \begin{equation}
262: \left. \left.
263: \frac{\partial^2 C_{\bf ij}}{\partial\Delta^2}
264: \right|_{\Delta=1}
265: = 2\frac{\partial^2 \epsilon_{\bf ij}^0(\Lambda,\>\Delta)}
266: {\partial\Delta^2}\right|_{\Delta=1} \le 0.
267: \end{equation}
268: Therefore, $\Delta=1$ is indeed a maximal point of the concurrence.
269:
270: Finally, we prove that $\Delta=1$ is the unique extreme point of
271: the concurrence $C_{\bf ij}$. For that purpose, we notice that
272: inequality (\ref{Second Derivative}) is actually strict. In other
273: words, the equal sign in it can be ignored. This can be easily
274: understood by observing the following fact: As $\Delta$ increases
275: from $-\infty$ to $\infty$, quantity $\langle\hat{S}_{\bf
276: i}^z\hat{S}_{\bf i}^z\rangle =\partial\epsilon_{\bf
277: ij}^0(\Lambda,\>\Delta)/\partial\Delta$ becomes more and more
278: negative. Consequently, the product on the right-hand side of
279: Eq.~(\ref{First Derivative}) cannot be zero at any point except
280: $\Delta=1$. That completes our discussion on the general behavior
281: of the concurrence $C_{\bf ij}$ for the antiferromagnetic $XXZ$
282: model on a finite $d$-dimensional simple cubic lattice. In
283: addition, we point out that the above proof can be easily extended
284: to other cases, such as the spin ladder model at $J_\perp=0$.
285:
286: In the following, we shall discuss the behavior of the concurrence
287: in the thermodynamic limit. In Ref.~\cite{SJGu03}, by using the
288: Bethe ansatz solution of the one-dimensional $XXZ$ chain, we
289: obtained the explicit expression of the concurrence near the
290: isotropic point
291: \begin{equation}
292: C_{i,i+1} = C_0 - C_1(\Delta-1)^2, \label{1DXXZ}
293: \end{equation}
294: where $C_0$ and $C_1$ are two real constants. Therefore, the
295: concurrence of the 1D $XXZ$ chain is a differentiable function of
296: $\Delta$ in the thermodynamic limit. However, things are quite
297: different in higher dimensions. For the $XXZ$ model in higher
298: dimensions, there exists no exact solution. One either uses
299: approximate analytical approach such as the spin-wave theory or
300: numerical approach such as exact diagonalization studies of finite
301: lattice. To obtain results in the thermodynamic limit, finite size
302: scaling analysis must be performed. By using the stochastic series
303: expansion quantum Monte Carlo method for lattices up to $16 \times
304: 16$, Sandvik~\cite{Sandvik} did an extensive study on the
305: two-dimensional S=1/2 antiferromagnetic Heisenberg model. The
306: finite size results for various ground state quantities were
307: extrapolated to the thermodynamic limit using fits to polynomials
308: in $1/L$, constrained by scaling forms previously obtained from
309: renormalization-group calculations for the nonlinear sigma model
310: and chiral perturbation theory. He demonstrated that the results
311: were fully consistent with the predicted leading finite size
312: corrections. With the same scaling forms, Lin, Flynn, and
313: Betts~\cite{Lin-2DXXZ} studied the $XXZ$ model on square lattices
314: and obtained various quantities as functions of anisotropic
315: parameter $\Delta$ for the infinite system. Two conclusions from
316: pervious works \cite{Sandvik,Lin-2DXXZ,JEHirsch89,WZheng91} are
317: relevant to the present study: (i) results obtained from the
318: spin-wave theory are qualitatively correct and quite accurate,
319: usually within 3 percent as compared with exact solution on finite
320: lattices; (ii) derivatives of the ground state energy with respect
321: to the anisotropic parameter $\Delta$ are not continuous at the
322: Heisenberg point $\Delta=1$, for example, see Figure 3 in Ref.
323: \cite{Lin-2DXXZ}. This conclusion is consistent with the belief
324: that there exists antiferromagnetic long-range order (LRO) in the
325: d-dimensional $XXZ$ model for $d \ge 2$. In fact, the existence of
326: the LRO for $d \ge 3$ was rigorous proved~\cite{XXZLRO}, while for
327: $d=2$ most numerical studies support it. Based on these
328: conclusions, we apply the spin-wave theory to calculate the
329: concurrence $C_{\bf ij}$ of the $XXZ$ model. We also use exact
330: diagonalization results as complementary. As shown in the
331: following, the symmetry breaking in the thermodynamic limit, which
332: is absent in the one-dimensional case, causes the singular
333: behavior of the concurrence at the quantum phase transition point.
334:
335: Following the standard procedure, the $XXZ$ Hamiltonian is mapped
336: into a boson model via the Holstein-Primakoff (HP) transformation
337: \begin{eqnarray}
338: &&\hat{S}^+_{\bf i}
339: =\sqrt{2S}\,(1-\hat{n}_{\bf i}/2S)^{1/2}\hat{a}_{\bf i}
340: \simeq \sqrt{2S}\,(1-\hat{n}_{\bf i}/4S)\, \hat{a}_{\bf i},\nonumber \\
341: &&\hat{S}^-_{\bf i}=\sqrt{2S}\>
342: \hat{a}^\dagger_{\bf i}(1-\hat{n}_{\bf i}/2S)^{1/2}
343: \simeq \sqrt{2S}\,\hat{a}^\dagger_{\bf i}(1-\hat{n}_{\bf i}/4S), \nonumber \\
344: &&\hat{S}^z_{\bf i}=S- \hat{a}_{\bf i}^\dagger \hat{a}_{\bf i},
345: \label{HP Transformation}
346: \end{eqnarray}
347: where $\hat{a}_{\bf i}$ and $\hat{a}_{\bf i}^\dagger$ are boson
348: creation and annihilation operators at site $\bf i$ for the spin
349: deviation. In the region $\Delta >1$, the antiferromagnetic
350: ordering is in the spin-$z$ direction. Consequently, we have
351: \begin{eqnarray}
352: \hat{H}_{XXZ}/\Delta
353: & \simeq &
354: \sum_{\langle{\bf ij}\rangle} \left[-S^2
355: + S \left(\hat{a}_{\bf i}^\dagger\hat{a}_{\bf i} + \hat{a}_{\bf
356: j}^\dagger\hat{a}_{\bf j}\right) \right. \nonumber\\
357: & + & \left. x S \left(\hat{a}_{\bf i}\hat{a}_{\bf j} +
358: \hat{a}_{\bf i}^\dagger\hat{a}_{\bf j}^\dagger\right)\right],
359: \end{eqnarray}
360: where $x=1/\Delta$. Using Fourier transform, we rewrite the
361: Hamiltonian as
362: \begin{eqnarray}
363: \hat{H}_{XXZ}/\Delta = -\frac{z}{2}NS^2
364: +zS\sum_{\bf k} \hat{H}({\bf k}),
365: \end{eqnarray}
366: where $z$ is the coordination number of the lattice and
367: \begin{eqnarray}
368: \hat{H}({\bf k})= \hat{a}_{\bf k}^\dagger \hat{a}_{\bf k}
369: +\frac{x\gamma_{\bf k}}{2}\left(\hat{a}_{\bf k}\hat{a}_{-{\bf k}}
370: + \hat{a}_{-{\bf k}}^\dagger \hat{a}_{\bf k}^\dagger\right)
371: \end{eqnarray}
372: with $\gamma_{\bf k}=\frac{2}{z} \sum_{m=1}^d\cos k_m$. By
373: applying the Bogoliubov transformation
374: \begin{eqnarray}
375: &&\hat{a}_{\bf k}= u_{\bf k} \hat{c}_{\bf k}
376: - v_{\bf k} \hat{c}_{-{\bf k}}^\dagger \nonumber \\
377: && \hat{a}_{\bf k}^\dagger = -v_{\bf k}\hat{c}_{-{\bf k}}
378: + u_{\bf k}\hat{c}_{\bf k}^\dagger,
379: \end{eqnarray}
380: we diagonalize $\hat{H}({\bf k})$ and obtain
381: \begin{eqnarray}
382: \hat{H}({\bf k})
383: & = &
384: v_{\bf k}^2 - x \gamma_{\bf k}u_{\bf k} v_{\bf k} \nonumber\\
385: & + &
386: \left(u_{\bf k}^2 + v_{\bf k}^2 -2x u_{\bf k}
387: v_{\bf k}\gamma_{\bf k}\right)
388: \hat{c}_{\bf k}^\dagger\hat{c}_{\bf k},
389: \end{eqnarray}
390: where the $u_{\bf k}$ and $v_{\bf k}$ satisfy
391: the following constraint conditions
392: \begin{eqnarray}
393: & &
394: u_{\bf k}^2 - v_{\bf k}^2=1, \nonumber \\
395: & &
396: \frac{x\gamma_{\bf k}}{2}(u_{\bf k}^2
397: + v_{\bf k}^2) - u_{\bf k} v_{\bf k} = 0.
398: \end{eqnarray}
399: Finally, the ground state energy of the model in the region of
400: $\Delta>1$ can be written as
401: \begin{eqnarray}
402: E_0(\Delta>1) =-\frac{z}{2} N S^2
403: + \frac{zS}{2}\sum_{\bf k} \left(\sqrt{1
404: - x^2\gamma_{\bf k}^2} - 1\right).
405: \label{Ground-State Energy1}
406: \end{eqnarray}
407:
408: By similar approach, we can also obtain the ground state energy of
409: the $XXZ$ model in the parameter region of $0<\Delta<1$. In this
410: case, the system has antiferromagnetc order in the $XY$ plane in
411: the thermodynamic limit. As a result, the diagonalized Hamiltonian
412: has the following form
413: \begin{eqnarray}
414: \hat{H}({\bf k})
415: & = &
416: \left(1 + y\gamma_{\bf k}\right)
417: v_{\bf k}^2 - x\gamma_{\bf k} u_{\bf k} v_{\bf k} \nonumber \\
418: & + &
419: \left[\left(1 + y\gamma_{\bf k}\right)
420: \left(u_{\bf k}^2 + v_{\bf k}^2\right)
421: - 2x u_{\bf k} v_{\bf k} \gamma_{\bf k}\right]
422: \hat{c}_{\bf k}^\dagger \hat{c}_{\bf k},
423: \end{eqnarray}
424: where $x=(1+\Delta)/2$ and $y=(1-\Delta)/2$, and the corresponding
425: ground state energy is
426: \begin{eqnarray}
427: E_0
428: & = &
429: - \frac{z}{2} N S^2 +
430: \frac{zS}{2}\sum_{\bf k}
431: \left(1 + y\gamma_{\bf k}\right) \nonumber \\
432: & \times &
433: \left(\sqrt{1-x^2\gamma_{\bf k}^2/(1+y\gamma_{\bf k})^2}
434: - 1\right).
435: \label{Ground-State Energy2}
436: \end{eqnarray}
437:
438: \begin{figure}
439: \includegraphics[width=7cm]{xxz2con}
440: \caption{\label{figure_xxz2con} The concurrence $C_{\bf ij}$ of
441: the two-dimensional $XXZ$ model as a function of
442: $\Delta$($=J_z/J_{x}$). In the Figure, the doted lines are
443: obtained from the exact diagonalization for $4*4$(square) and
444: $6*6$(circle) square lattices respectively.}
445: \end{figure}
446:
447: \begin{figure}
448: \includegraphics[width=7cm]{xxz3con}
449: \caption{\label{figure_xxz3con} The concurrence $C_{\bf ij}$ of
450: the three-dimensional $XXZ$ model as a function of
451: $\Delta$($=J_z/J_{x}$).}
452: \end{figure}
453:
454: Within the spin-wave theory framework, we calculate the spin
455: correlation function $G_{\bf ij}^{zz}$ and hence the concurrence
456: $C_{\bf ij}$ of the model in two and three dimensions. Our results
457: are shown in Figs.~\ref{figure_xxz2con} and \ref{figure_xxz3con},
458: respectively. We also show results obtained from the exact
459: diagonalization of the $XXZ$ model on finite square lattices. The
460: trend as function of lattice size is clear. It is interesting to
461: see that, in both cases, the concurrences $C_{\bf ij}$ of the
462: $XXZ$ model not only have their maximal value at the critical
463: point $\Delta=1$, but also show discontinuities in their first
464: derivative with respect to $\Delta$ at the transition point. This
465: behavior is quite different from the one-dimensional case
466: (Eq.~\ref{1DXXZ}), as we expected. We attribute this difference to
467: the existence of the magnetic long-range orders in the system with
468: $d\ge 2$.
469:
470: As we have seen, the concurrence $C_{\bf ij}$ is closely related
471: to the ground state energy of the model. As a result, any
472: singularity in the ground state energy may be inherited by the
473: concurrence\cite{LAWu04}. On the other hand, on a finite
474: $d$-dimensional simple cubic lattice, the ground state of the
475: antiferromagnetic $XXZ$ model is non-degenerate for
476: $\Delta\in(-\infty,\>\infty)$\cite{Affleck}. Therefore, the ground
477: state energy $E_0(\Lambda,\>\Delta)$ as well as the concurrence
478: $C_{\bf ij}$ are analytical functions of $\Delta$, regardless of
479: the dimensionality of the lattice. However, it is no longer true
480: in the thermodynamic limit. For the one-dimensional $XXZ$ model,
481: it is well known that its ground state in both $\Delta<1$ and
482: $\Delta>1$ regions does not have magnetic long-range order.
483: Therefore, we do not expect a dramatic change in the ground state
484: energy $E_0$ taking place at $\Delta=1$. Consequently, the
485: concurrence will behave more or less like itself on a finite
486: lattice. However, in two and three dimensions, the ground state
487: energy of the system develops a cusp at the transition point in
488: the thermodynamic limit\cite{Lin-2DXXZ}. This phenomenon can be
489: understood by the picture of the first excited-energy levels
490: crossing at $\Delta=1$, required by the existence of magnetic
491: long-range order~\cite{Tian}. Therefore, a singularity inherited
492: by the concurrence at the transition point, is expected to appear,
493: as shown by our calculations.
494:
495: In summary, we have studied the ground state two-spin
496: entanglement, as measured by the concurrence, in the
497: $d$-dimensional $XXZ$ model. We gave a rigorous proof that the
498: ground state concurrence in the $XXZ$ model reaches maximum at the
499: isotropic point. We extended our previous studies in one dimension
500: \cite{SJGu03} to two and three dimensions by using the spin-wave
501: theory and exact diagonalization technique. The use of the
502: spin-wave theory is justified by the existence of magnetic
503: long-range order in the $XXZ$ model for dimensionality $d\ge2$. We
504: found that the concurrence in two- and three-dimensional $XXZ$
505: models also reaches maximum at the isotropic point $\Delta=1$.
506: Unlike the one dimension case, the concurrence shows cusp-like
507: behavior around the critical point, and its first derivative is
508: not continuous in the vicinity of the critical point.
509:
510:
511: This work was supported by a grant from the Research Grants
512: Council of the HKSAR, China (Project No. 401703). G. S. Tian
513: acknowledges financial support from the C. N. Yang Visiting
514: Fellowship program.
515:
516: Note Added: While we were preparing this paper, we received a
517: preprint from Dr. M. F. Yang\cite{MFYang04}. Some of our results
518: were also obtained by him.
519:
520: \begin{references}
521:
522: \bibitem{ABinstein35}
523: A. Einstein, B. Podolsky, and N. Rosen, Phys. Rev. {\bf 47}, 777
524: (1935).
525:
526: \bibitem{MANielsenb}
527: M. A. Nielsen and I. L. Chuang, {\it Quantum Computation and
528: Quantum information} (Cambridge University Press, Cambridge,
529: 2000).
530:
531: %The following references is QPT related.
532: \bibitem{AOsterloh2002}
533: A. Osterloh, Luigi Amico, G. Falci and Rosario Fazio, Nature {\bf
534: 416}, 608 (2002).
535:
536: \bibitem{TJOsbornee}
537: T. J. Osborne and M.A. Nielsen, Phys. Rev. A {\bf 66},
538: 032110(2002)
539:
540: \bibitem{IBose02}
541: I. Bose and E. Chattopadhyay, Phys. Rev. A 66, 062320 (2002).
542:
543: \bibitem{GVidal2003}
544: G. Vidal, J. I. Latorre, E. Rico, and A. Kitaev, Phys. Rev. Lett.
545: {\bf 90}, 227902 (2003); J. I. Latorre, E. Rico, and G. Vidal,
546: quant-ph/0304098 (2003).
547:
548: \bibitem{SJGu03}
549: S. J. Gu, H. Q. Lin, and Y. Q. Li, Phys. Rev. A {\bf 68}, 042330
550: (2003).
551:
552: \bibitem{JVidal04}
553: J. Vidal, G. Palacios, and R. Mosseri, Phys. Rev. A {\bf 69},
554: 022107 (2004); J. Vidal, R. Mosseri, J. Dukelsky, Phys. Rev. A 69,
555: 054101 (2004).
556:
557: \bibitem{LAWu04}
558: L. A. Wu, M. S. Sarandy, and D. A. Lidar, arXiv: quant-ph/0407056
559:
560: \bibitem{MFYang04}
561: M. F. Yang, arXiv: quant-ph/0407226.
562:
563: \bibitem{YChen04}
564: Y. Chen, P. Zanardi, Z. D. Wang, and F. C. Zhang, arXiv:
565: quant-ph/0407228.
566:
567: % A book on quantum phase transition
568: \bibitem{Sachdev}
569: S. Sachdev, {\it Quantum Phase Transitions}, (Cambridge University
570: Press, Cambridge, UK, 2000).
571:
572: \bibitem{Hill} S. Hill and W. K. Wootters, Phys. Rev. Lett. {\bf 78},
573: 5022 (1997); W. K. Wootters, Phys. Rev. Lett. {\bf 80}, 2245 (1998).
574:
575: % A reference on two-dimensional XXZ model
576: \bibitem{OFSyljuasenm}
577: Olav F. Sylju{\aa}sen, arXiv:quant-ph/0312101
578:
579: \bibitem{Lieb} E. Lieb and D. Mattis, J. Math. Phys. {\bf 3},
580: 749 (1962).
581:
582: \bibitem{Affleck} I. Affleck and E. Lieb, Lett. Math. Phys.
583: {\bf 12}, 57 (1986).
584:
585: %Other related reference on concurrence in spin model
586: \bibitem{KMOConnor2001}
587: K. M. O'Connor and W. K. Wootters, Phys. Rev. A {\bf 63}, 052302
588: (2001).
589:
590: \bibitem{XWang2002PLA}
591: X. Wang, and P. Zanardi, Phys. Lett. A {\bf 301}, 1(2002).
592:
593: \bibitem{Franklin} J. Franklin, {\it Matrix Theory} (Prentice Hall,
594: New Jersey, 1968).
595:
596: %spin wave theory and XXZ models
597: \bibitem{Sandvik}
598: A. W. Sandvik, Phys. Rev. B {\bf 56}, 11678 (1997).
599:
600: \bibitem{Lin-2DXXZ}
601: H. Q. Lin, J. S. Flynn, and D. D. Betts, Phys. Rev. B {\bf 64},
602: 214411 (2001).
603:
604: \bibitem{JEHirsch89}
605: J. E. Hirsch, and S. Tang, Phys. Rev. B {\bf 40}, 4769 (1989).
606:
607: \bibitem{WZheng91}
608: Zheng Weihong, J. Oittmaa, and C. J. Hamer, Phys. Rev. B {\bf 43}, 8321 (1991).
609:
610: \bibitem{XXZLRO}
611: F. J. Dyson, E. H. Lieb, and B. Simon, J. Stat. Phys. {\bf 18},
612: 335 (1978); E. J. Neves and J. F. Perez, Phys. Lett. A{\bf 114},
613: 331 (1986); T. Kennedy, E. Lieb, and B. S. Shastry, J. Stat. Phys.
614: {\bf 53}, 1019 (1988).
615:
616: \bibitem{Tian} G. S. Tian and H. Q. Lin, Phys. Rev. B {\bf 67},
617: 245105 (2003).
618:
619:
620: \end{references}
621:
622: \end{document}
623:
624:
625:
626: %The above proof can be easily extended to other models.
627: %For examples, for an antiferromagnetic Heisenberg
628: %chain with the next-nearest-neighbor interaction, whose
629: %Hamiltonian reads
630: %\begin{eqnarray}
631: %\hat{H}(J)
632: %&=&\sum_{j=1}^{L}\left(\hat{S}_j\hat{S}_{j+1}
633: %+ J \hat{S}_j\hat{S}_{j+2}\right),
634: %\label{eq:Hamiltonianj1j2},
635: %\end{eqnarray}
636: %one can also prove that the concurrence of the nearest neighbors
637: %reaches a maximum at $J=0$ using the same strategy. Though the
638: %maximum point $J=0$ is not the critical point of the quantum phase
639: %transition\cite{SRWhite96}.
640:
641: % A reference on J1J2 model
642: \bibitem{SRWhite96}
643: S. R. White, and I. Affleck, Phys. Rev. B {54}, 9862 (1996).
644:
645:
646: %The above proof can be easily extended to other cases,
647: %such as the spin ladder model at $J_\perp=0$ and the so-called $J_1-J_2$
648: %model at $J_2=0$. However, we would like to point out that,
649: %in the latter case, although the concurrence $C_{\bf ij}$ has a maximum
650: %at $J_2=0$, one has not found any recognizable phase transition
651: %there\cite{SRWhite96}. That gives us an example, which shows
652: %that an extreme point of the concurrence may not be always
653: %a quantum phase transition point of the system.
654:
655: