1:
2: \documentclass[preprintnumbers,amsmath,two column]{revtex4}
3: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
4: \usepackage{graphicx}
5: \usepackage{dcolumn}
6: \usepackage{bm}
7:
8: \begin{document}
9:
10: \title{Ground-state properties of the two-site Hubbard-Holstein model: an exact
11: solution}
12: \author{Yu-Yu Zhang$^{1}$, Tao Liu$^{2}$, Qing-Hu Chen$^{1,3,\dag}$, Xiaoguang Wang$%
13: ^{1}$, Kelin Wang$^{2,4}$}
14:
15: \address{$^{1}$Department of Physics, Zhejiang University,
16: Hangzhou 310027, P. R. China.\\
17: $^{2}$Department of Physics, Southwest University of Science and
18: Technology, Mianyang 621010, P. R. China.\\
19: $^{3}$Center for Statistical and Theoretical Condensed Matter
20: Physics, Zhejiang Normal University, Jinhua 321004, P. R. China \\
21: $^{4}$Department of Mordern Physics, University of Science and
22: Technology of China, Hefei 230026, P. R. China.}
23:
24: \date{\today}
25:
26: \begin{abstract}
27: We revisit the two-site Hubbard-Holstein model by using extended
28: phonon coherent states. The nontrivial singlet bipolaron is studied
29: exactly in the whole coupling regime. The ground-state (GS) energy
30: and the double occupancy probability are calculated. The linear
31: entropy is exploited successfully to quantify bipartite entanglement
32: between electrons and their environment phonons, displaying a
33: maximum entanglement of the singlet-bipolaron in strong coupling
34: regime. A dramatic drop in the crossover regime is observed in the
35: GS fidelity and its susceptibility. The bipolaron properties is also
36: characterized classically by correlation functions. It is found that
37: the crossover from a two-site to single-site bipolaron is more
38: abrupt and shifts to a larger electron-phonon coupling strength as
39: electron-electron Coulomb repulsion increases.
40: \end{abstract}
41:
42: \maketitle
43:
44:
45: \section{introduction}
46:
47: It has been demonstrated for more than one decade that the polaronic
48: effects observed in High-$T_{c}$ superconducting
49: cuprates~\cite{ASAlexandrov} and the colossal magnorestive
50: manganites~\cite{AJMilis} are relevant to the electron-phonon (e-ph)
51: coupling in these systems. The famous Holstein molecular crystal
52: model~\cite{THolstein}, where electrons are coupled with local
53: phonons, has then revived in recent years. More recently, to include
54: the electron-electron (e-e) Coulomb repulsion interactions, the
55: Hubbard-Holstein (HH) model~\cite{Fehske} for strongly correlated
56: electron systems has made significant progress in understanding
57: many-body aspects. As is well known, an exact solution to this model
58: in the thermodynamic limit is impossible, and many approximate
59: approaches are then employed. The two-site HH model, where two
60: electrons hop between two adjacent lattice sites, can be solved
61: analytically\cite{Berciu}. It is not only a prototype of the HH
62: model but also helpful for the better understanding to the polaron
63: and bipolaron behavior in infinite lattices.
64:
65: Recently, an contemporary alternative characterization of the ground
66: state (GS) properties has been focused on quantum information tools
67: in terms of quantum
68: entanglement~\cite{Scott,William,Vidal,Gu,chenqh} and
69: fidelity~\cite{Quan,Zanardi,Vezzani,Min,Shu,Buon,Chen}. These
70: studies will establish somewhat interesting understanding from the
71: field of quantum information theory to condensed matter
72: physics~\cite{Dunning,Norman}. Usually, it is hard to calculate
73: these quantities due to the lack of knowledge on the exact GS wave
74: function.
75:
76: The two-site HH model has been previously addressed by means of different
77: methods, such as variational method \cite{ANDas,Acquarone}, perturbation
78: theory~\cite{JC}, and numerical diagonalization\cite{Ranninger,EVLde}. In
79: spite of these efforts there remains poor convergence in the intermediate
80: coupling regime. A reliable treatment in the whole coupling regime is still
81: needed. Recently, Berciu derived all Greens's function analytically for the
82: two-site HH model in terms continued fractions\cite{Berciu}.
83:
84: In this work, by using extended bosonic coherent states
85: ~\cite{chen,Han,liu}, we develop a new exact technique to deal with
86: the two-site HH model. The wave function is proposed explicitly, by
87: which many quantities can be calculated directly. The paper is
88: organized as follows. In Sec II we introduce the model and describe
89: the approach. In Sec III we calculate the linear entropy , the GS
90: fidelity and its susceptibility to study crossover properties from
91: quantum information perspective. The static correlation function is
92: also evaluated to analyze the GS properties. The main conclusions
93: are given in the last section.
94:
95: \section{ Model Hamiltonian and exact solution}
96:
97: The Hamiltonian of the two-site HH model takes the form
98: \begin{eqnarray} \label{hamiltonian}
99: H&=&\sum_{i,\sigma}\varepsilon n_{i \sigma}
100: -\sum_{\sigma}t(c_{1\sigma}^{\dagger}c_{2\sigma}+c_{2\sigma}^{\dagger}c_{1%
101: \sigma}) \nonumber \\
102: &+&U\sum_{i}n_{i\uparrow}n_{i\downarrow}+Vn_{1}n_{2}
103: +g_{1}\omega_{0}\sum_{i,\sigma}n_{i,\sigma}(b_{i}+b_{i}^{\dagger}) \nonumber
104: \\
105: &+&g_{2}\omega_{0}\sum_{i,\sigma}n_{i,\sigma}
106: (b_{i+\delta}+b_{i+\delta}^{\dagger})+\omega_{0}\sum_{i}b_{i}^{\dagger}b_{i},
107: \end{eqnarray}
108: where $i(=1$ or $2$) denotes the label of sites. $i+\delta=2$ for $i=1$ and
109: vice versa. $c_{i\sigma}(c_{i\sigma}^{\dagger})$ is the annihilation
110: (creation) operator for the electrons and $n_{i,\sigma}(=c_{i\sigma}^{%
111: \dagger}c_{i\sigma})$ is the corresponding number operator at site $i$ with
112: spin $\sigma$, besides $n_{i}=n_{i\uparrow}+n_{i\downarrow}$. $\varepsilon$
113: is the unperturbed site potential. $t$ is the usual hopping integral. $U$
114: and $V$ denote the on-site and inter-site Coulomb repulsion between
115: electrons respectively. $g_{1}$ and $g_{2}$ denote the on-site and
116: inter-site e-ph coupling parameters. $b_{i}(b_{i}^{\dagger})$ is the
117: annihilation (creation) operator for phonons corresponding to interatomic
118: vibrations at site $i$. $\omega_{0}$ is the phonon frequency and is set unit
119: for convenience in the following.
120:
121: Introducing new phonon operators $a=(b_{1}+b_{2})/\sqrt{2}$ and $%
122: d=(b_{1}-b_{2})/\sqrt{2}$, the Hamiltonian ~(\ref{hamiltonian}) can be
123: written into two independent parts ($H=H_{d}+H_{a}$),
124: \begin{eqnarray} \label{hamiltonian d}
125: H_{d}&=&d^{\dagger}d+\sum_{i,\sigma}\varepsilon n_{i
126: \sigma}-\sum_{\sigma}t(c_{1\sigma}^{\dagger}c_{2\sigma}+c_{2\sigma}^{%
127: \dagger}c_{1\sigma}) \nonumber \\
128: &+&U\sum_{i}n_{i\uparrow}n_{i\downarrow}+Vn_{1}n_{2} \nonumber \\
129: &+&g_{-}(n_{1}-n_{2})(d+d^{\dagger})-n^{2}g_{+}^{2}
130: \end{eqnarray}
131: and
132: \begin{equation}
133: H_{a} = \tilde{a}^{\dagger}\tilde{a},
134: \end{equation}
135: where $\tilde{a}^{\dagger}=a^{\dagger}+ng_{+}$, $\tilde{a}=a+ng_{+}$, $%
136: g_{+}=(g_{1}+g_{2})/\sqrt{2}$, and $g_{-}=(g_{1}-g_{2})/\sqrt{2}$. $H_{a}$
137: describes a shifted oscillator and represents lowering of energy, which is a
138: constant motion. $H_{d}$ represents an effective e-ph coupling system which
139: phonons are coupled linearly with the electrons.
140:
141: In the effective Hamiltonian $H_{d}$, among four different
142: bipolaronic states, such as singlet bipolaronic states, singlet
143: Anderson bipolaronic states, singlet bipolaronic states
144: (antibonding), and triplet states, the singlet bipolaronic state is
145: the most nontrivial to any approaches. We will focus on this state
146: in this paper.
147:
148: For three singlet normalized electronic states $c_{1\uparrow}^{\dagger}c_{1%
149: \downarrow}^{\dagger}|0\rangle_{e}$, $c_{2\uparrow}^{\dagger}c_{2%
150: \downarrow}^{\dagger}|0\rangle_{e}$ and $\frac{1}{\sqrt{2}}%
151: (c_{1\uparrow}^{\dagger}c_{2\downarrow}^{\dagger}-c_{1\downarrow}^{%
152: \dagger}c_{2\uparrow}^{\dagger})|0\rangle_{e}$, the singlet bipolaronic
153: state wave function $|\psi\rangle$ can be expressed as
154: \begin{eqnarray} \label{SB}
155: |\psi\rangle&=& |\varphi_{1}\rangle
156: c_{1\uparrow}^{\dagger}c_{1\downarrow}^{\dagger}|0\rangle_{e}+
157: |\varphi_{2}\rangle c_{2\uparrow}^{\dagger}c_{2\downarrow}^{\dagger}
158: |0\rangle_{e} \nonumber \\
159: &+&|\varphi_{3}\rangle\frac{1}{\sqrt{2}}(c_{1\uparrow}^{\dagger}c_{2%
160: \downarrow}^{\dagger}-c_{1\downarrow}^{\dagger}c_{2\uparrow}^{\dagger})|0%
161: \rangle_{e}
162: \end{eqnarray}
163: where $|\varphi_{1}\rangle$, $|\varphi_{2}\rangle$ and $|\varphi_{3}\rangle$
164: correspond to phonon states. Inserting it into a schr\"{o}dinger equation
165: for the effective Hamiltonian in Eq. (\ref{hamiltonian d}), three equations
166: are derived by comparing the coefficients of $c_{1\uparrow}^{\dagger}c_{1%
167: \downarrow}^{\dagger}|0\rangle_{e}$, $c_{2\uparrow}^{\dagger}c_{2%
168: \downarrow}^{\dagger}|0\rangle_{e}$ and $\frac{1}{\sqrt{2}}%
169: (c_{1\uparrow}^{\dagger}c_{2\downarrow}^{\dagger}-c_{1\downarrow}^{%
170: \dagger}c_{2\uparrow}^{\dagger})|0\rangle_{e}$,
171: \begin{equation} \label{eq2}
172: [A^{\dagger}A+2\varepsilon-4(g_{1}^{2}+g_{2}^{2})+U]|\varphi_{1}\rangle-%
173: \sqrt{2}t|\varphi_{3}\rangle=E|\varphi_{1}\rangle
174: \end{equation}
175: \begin{equation} \label{eq3}
176: [B^{\dagger}B+2\varepsilon-4(g_{1}^{2}+g_{2}^{2})+U]|\varphi_{2}\rangle-%
177: \sqrt{2}t|\varphi_{3}\rangle=E|\varphi_{2}\rangle
178: \end{equation}
179: \begin{equation} \label{eq4}
180: [d^{\dagger}d+2\varepsilon-2(g_{1}+g_{2})^{2}+V]|\varphi_{3}\rangle-\sqrt{2}%
181: t(|\varphi_{1}\rangle+|\varphi_{2}\rangle)=E|\varphi_{3}\rangle
182: \end{equation}
183: where we have used two displacement transformation $A^{\dagger}=d^{%
184: \dagger}+2g_{-}$ and $B^{\dagger}=d^{\dagger}-2g_{-}$. Note that the linear
185: term for the phonon operator $d(d^{+})$ is removed, and two new free bosonic
186: field with operator $A(A^{+})$ and $B(B^{+})$ appear. In the next step, we
187: naturally choose the basis in terms of these new operator, instead of $%
188: d(d^{+})$, by which the phonon states $|\varphi_{1}\rangle$ and $%
189: |\varphi_{2}\rangle$ can be expanded in complete basis
190: ${|n\rangle_{A}}$ and ${|n\rangle_{B}}$ respectively, where
191: $|n\rangle_{A}$ and $|n\rangle_{B}$ are Fock states of the new
192: bosonic operators
193: \begin{eqnarray} \label{function1}
194: &|\varphi_{1}\rangle&=\sum_{n=0}^{N_{tr}}c_{n}|n\rangle_{A} \nonumber \\
195: &=&\sum_{n=0}^{N_{tr}}c_{n}\frac{1}{\sqrt{n!}}(d^{%
196: \dagger}+2g_{-})^{n}e^{-2g_{-}d^{\dagger}-2g_{-}^{2}}|0\rangle_{ph}
197: \end{eqnarray}
198:
199: \begin{eqnarray} \label{function2}
200: &|\varphi_{2}\rangle&=\sum_{n=0}^{N_{tr}}d_{n}|n\rangle_{B} \nonumber \\
201: &=&\sum_{n=0}^{N_{tr}}d_{n}\frac{1}{\sqrt{n!}}(d^{%
202: \dagger}-2g_{-})^{n}e^{2g_{-}d^{\dagger}-2g_{-}^{2}}|0\rangle_{ph}
203: \end{eqnarray}
204:
205: As we know that the vacuum state $\left| 0\right\rangle _{A(B)}$ is
206: just a bosonic coherent-state in $d(d^{+})$ with an eigenvalue
207: $2g(-2g_{-})$\cite {chen,Han,liu}. So this new basis is
208: overcomplete, and actually does not involve any truncation in the
209: Fock space of $d(d^{+})$, which highlights the present approach. It
210: is also clear that many-body correlations for bosons
211: are essentially included in extended coherent states (\ref{function1}) and (%
212: \ref{function2}). As usual, the phonon state $|\varphi_{3}\rangle$ is
213: expanded in a complete basis ${|n\rangle}$, which is the Fock state of $%
214: d(d^{+})$
215: \begin{equation} \label{function3}
216: |\varphi_{3}\rangle=\sum_{n=0}^{N_{tr}}f_{n}|n\rangle
217: \end{equation}
218:
219: Substituting these phonon states~(\ref{function1}),~(\ref{function2}), and~(%
220: \ref{function3}) into Eqs.~(\ref{eq2}),(~\ref{eq3}) and ~(\ref{eq4})
221: and left multiplying state $_{A}\langle m|$, $_{B}\langle m|$,
222: $\langle m|$, respectively, the equations can be written as
223: $3N_{tr}\times3N_{tr}$ matrix, and the $m^{\prime}$th row is written
224: as
225: \begin{equation} \label{eq5}
226: [m+2\varepsilon-4(g_{1}^{2}+g_{2}^{2})+U]c_{m}-\sqrt{2}t\sum_{n}{_{A}\langle
227: m}|n\rangle f_{n}=Ec_{m}
228: \end{equation}
229: \begin{equation} \label{eq6}
230: [m+2\varepsilon-4(g_{1}^{2}+g_{2}^{2})+U]d_{m}-\sqrt{2}t\sum_{n}{_{B}\langle
231: m}|n\rangle f_{n}=Ed_{m}
232: \end{equation}
233:
234: \begin{eqnarray} \label{eq7}
235: &[m&+2\varepsilon-2(g_{1}+g_{2})^{2}+V]f_{m} -\sqrt{2}t(\sum_{n}c_{n}\langle
236: m|n\rangle_{A} \nonumber \\
237: &+&\sum_{n}d_{n}\langle m|n\rangle_{B})=Ef_{m}
238: \end{eqnarray}
239:
240:
241: where $ _{A}\langle m|n\rangle =\langle
242: m|n\rangle_{B}=(-1)^{n}D_{mn}$, and $_{B}\langle m|n\rangle
243: =\langle m|n\rangle_{A}=(-1)^{m}D_{mn} $, with $
244: D_{mn}=e^{-2g_{-}^{2}}\sum_{i=0}^{{min}[m,n]}(-1)^{-i}\frac{\sqrt{
245: m!n!}(2g_{-})^{m+n-2i}}{(m-i)!(n-i)!i!}$.
246:
247:
248: The eigenvalues $E$ and eigenvectors with coefficients ${c_{n}}$,
249: ${d_{n}}$
250: and $f_{n}$ can be exactly solved by diagonalizing the above $%
251: 3N_{tr}\times3N_{tr}$ matrix numerically.
252:
253: To obtain the true exact results, in principle, the truncated number
254: $N_{tr}$ should be taken to infinity. Fortunately, finite terms of
255: the singlet state in Eq.(4) are sufficient to give very accurate
256: results in the whole parameter range. It should be noted that in the
257: exact diagonalization in Fock space of original phonon state $d$
258: \cite{EVLde}, a considerable large phonon number is needed to give
259: reasonably good results. We believe that we have exactly solved this
260: model numerically.
261:
262: \begin{figure}[tbp]
263: \includegraphics[scale=0.68]{MLF.eps}
264: \caption{(a) GS energy $E_{0}/t$ of exact solution (solid line)
265: and MLFS transformation (dashed line) vs $\lambda$ for $U=0,3,6$
266: and $10$. The red line represents the twice polaron GS energy,
267: which then separates the polaron and
268: bipolaron regime. (b) The probability of the two-site bipolarons $P_2$ for $%
269: U=0$, $3$, $6$, and $10$. The other parameters are chosen: $V=0$, $t=2.0$, $%
270: \varepsilon=0$.}
271: \label{SB}
272: \end{figure}
273:
274: To show the effectiveness of the present approach, we first calculate the GS
275: energy. Fig. 1(a) presents the GS energy $E_{0}/t$ as a function of the
276: on-site effective coupling strength $\lambda=g_{1}^{2}/t$ by setting $%
277: g_{2}=0 $ conveniently. The results for the energy by the variational method
278: based on the modified Lang-Firsov transformation with a squeezing phonon
279: state transformations (MLFS)~\cite{ANDas} are also list. It is observed that
280: the present results are lower than the MLFS results \cite{ANDas}, especially
281: in the intermediate coupling regime. Comparing with the Fig. 1(a) in Ref.
282: \cite{Berciu}, we find that the present results for the GS energy are
283: consistent with those from the lowest pole of a Green's function.
284:
285:
286: \section{ground state properties}
287:
288: \subsection{Crossover from two-site to single-site bipolarons}
289:
290: As shown in Fig. 1(a) that there is two regimes distinctly. In the Ref. \cite
291: {Berciu}, the energy vs $\lambda$ curves in two regimes are fitted by two
292: functions, and the abrupt drop of the first-order derivative signals the
293: crossover regime from two-site bipolarons to single-site bipolarons. We will
294: propose a quantitative criterion. Because the exact wave function in the
295: present technique is explicitly given, we can calculate the probability of
296: the system that two electrons are in two sites by Eq. (4) directly.
297: \begin{equation}
298: P_2=Re\langle\varphi_{3}|\varphi_{3}\rangle
299: \end{equation}
300: The probability of the two-site bipolarons $P_2$ is shown in Fig.
301: 1(b). At the weak coupling, the two electrons prefer to stay in two
302: sites. As the coupling strength increases, the two electrons tend to
303: occupy the single site. Strictly speaking, at weak (strong)
304: coupling, it is not pure two-site (single-site) bipolarons, only
305: dominated two-site (single-site) bipolarons. The crossover regime is
306: wide for weak on-site Coulomb repulsion $U$. As $U$ increases, the
307: crossover shifts to larger $\lambda$ and becomes more sharp.
308: However, the jump of the $P_2$ is not observed, indicating unlikely
309: a phase transition.
310:
311: \begin{figure}[tbp]
312: \includegraphics[width=7.8cm]{vp.eps}
313: \caption{GS energy $E_{0}/t$ and the probability of the two-site bipolarons $%
314: P_2$ in the inset vs $\lambda$ for $V=0,2,4,6,8,10$ by choosing $V=0.2$, $%
315: t=2.0$, $\varepsilon=0.8$.}
316: \label{p2}
317: \end{figure}
318: The inter-site Coulomb repulsion $V$ may favor the formation of dominated
319: single-site bipoalrons. To show this effect, we calculate the GS energy and
320: the probability of the two-site bipolarons for several values of $V$ for
321: fixed $U$, which are displayed in Fig. 2. It is clear that the crossover
322: shifts to smaller $\lambda$ with increasing $V$, and becomes more smooth.
323:
324: \begin{figure}[tbp]
325: \includegraphics[scale=0.7]{excite.eps}
326: \caption{GS energy and the former five excited energies $E/t$ (a) and the
327: corresponding probability $P_2$ (b) against $\lambda$ for $U=6$ by choosing $%
328: V=0.2$, $t=2.0$, $\varepsilon=0.8$.}
329: \label{excite}
330: \end{figure}
331:
332: In our approach, we can also get the excited states and corresponding
333: energies. Both the GS and a few low excited state bipolaron energies for $%
334: U=6 $ are presented in Fig.~\ref{excite}(a). Interesting, we find
335: that the curves for the GS and the 1st excited state energies, the
336: 2nd and 3rd excited state energies, 4th and 5th excited state
337: energies almost merge into a single line in the strong coupling
338: regime. The corresponding two eigenstates are almost degenerated.
339: Actually they are different. The difference is too small to be
340: observed. The merged point shift to larger $\lambda$ for higher
341: excited states. To show the bipolaron behavior in these states, we
342: also calculate the probability of the two-site bipolarons $P_2$ as a
343: function of $\lambda$, which are shown in Fig.\ref{excite}(b). The
344: $P_2$ vs $\lambda$ curves also merge into a single line exactly at
345: the same point as the energy vs $\lambda$ curves. In the GS and 1st
346: excited state, the value of $P_2$ decreases monotonously as
347: $\lambda$. However, in the other high excited states, a
348: non-monotonous behavior of the $P_2$ is observed. In a fixed value of $%
349: \lambda$, the probability of the two-site bipolarons $P_2$ determines the
350: positive repulsion energy for the double occupancy, the energy gained for
351: the two electrons in the same site, and the energy gained for the hopping of
352: electrons from one site to other site. To increase the energy in higher
353: excited states, the competition of these energies contributes the
354: complicated behavior shown in Fig.\ref{excite}(b).
355:
356: \subsection{Linear entropy}
357:
358: To investigate the crossover from two-site to single-site
359: bipolarons in quantum information science, we attempt to study
360: quantum entanglement between electrons and their surrounding
361: phonons by means of the linear entropy $E_{l}$~\cite{Yang,Wang},
362: which is an alternative measurement to indicate the entanglement
363: of a two-site and single-site bipolaron. It is defined as
364: \begin{equation}
365: E_{l}=1-{Tr}\rho^{2},
366: \end{equation}
367: where $\rho=Tr_{ph}(|\psi\rangle \langle\psi|)$ is the reduced density
368: matrix of electrons by taking partial trace over the phonon degrees of the
369: freedom.
370:
371: The normalized GS wave function of the singlet bipolaronic state in Eq.(4)
372: is described as
373: \begin{equation}
374: |\psi\rangle=|\varphi_{1}\rangle|1\rangle+|\varphi_{2}\rangle|2\rangle+|%
375: \varphi_{3}\rangle|3\rangle,
376: \end{equation}
377: where the phonon states satisfy $\langle\varphi_{1}|\varphi_{1}\rangle+%
378: \langle\varphi_{2}|\varphi_{2}\rangle+\langle\varphi_{3}|\varphi_{3}%
379: \rangle=1 $. And the quantum states $|1\rangle$, $|2\rangle$ and $|3\rangle$
380: represent the normalized and orthogonalized singlet electronic states $%
381: c_{1\uparrow}^{\dagger}c_{1\downarrow}^{\dagger}|0\rangle_{e}$ $%
382: c_{2\uparrow}^{\dagger}c_{2\downarrow}^{\dagger}|0\rangle_{e}$ and $\frac{1}{%
383: \sqrt{2}}(c_{1\uparrow}^{\dagger}c_{2\downarrow}^{\dagger}-c_{1\downarrow}^{%
384: \dagger}c_{2\uparrow}^{\dagger})|0\rangle_{e}$ respectively.
385:
386: The reduced density matrix $\rho$ can be derived by taking partial trace
387: over the phonon degrees of the freedom $\rho=\sum^{3}_{i,j=1}{%
388: \langle\varphi_{j}}|\varphi_{i}\rangle|i\rangle\langle j|$. Therefore the
389: linear entropy can be derived simply as
390: \begin{eqnarray}
391: E_{l}&=&1-[\langle\varphi_{1}|\varphi_{1}\rangle^{2}+\langle\varphi_{2}|%
392: \varphi_{2}\rangle^{2}+\langle\varphi_{3}|\varphi_{3}\rangle^{2} \nonumber
393: \\
394: &+&2Re(\langle\varphi_{1}|\varphi_{2}\rangle^{2}+\langle\varphi_{1}|%
395: \varphi_{3}\rangle^{2}+\langle\varphi_{2}|\varphi_{3}\rangle^{2})]
396: \end{eqnarray}
397:
398: As plotted in Fig.\ref{entangl1}, the linear entropy $E_{l}$
399: increases smoothly with the e-ph coupling parameters $\lambda$ for
400: different e-e Coulomb repulsions $U$. In general, in weak e-ph
401: coupling region, two electrons tend to occupy two sites, and form
402: so-called two-site bipolaron, resulting a low degree of the quantum
403: entanglement in the original coupling strength. For a fixed $U$, as
404: $\lambda$ increases two electrons become to tightly interact with
405: the same lattice. So the entire charge of the electrons and entire
406: deformations of the lattices are restricted on one site, leading to
407: the formation of a single-site bipolaron. However, the two-site
408: bipolaron is known as that each electron just interacts with phonons
409: of its own site and then the linear entropy is expected to be much
410: smaller than that of the singlet-site bipolaron. As shown in
411: Fig.\ref{entangl1} the linear entropy reaches its maximum
412: $E_{l}=0.5$ in the strong coupling region, which displays that the
413: single-site bipolarons are maximally entangled. Obviously the
414: crossover point shifts to a large value of the coupling strength
415: $\lambda$ as $U$ increases. The similar crossover point has been
416: demonstrated by the probability of the two-site bipolarons presented
417: in Fig. 1(b). Therefore we can say that in the presence of quantum
418: correlations the two-site and singlet-site bipolarons are
419: effectively characterized to a high (low) degree of bipartite
420: quantum entanglement.
421:
422: \begin{figure}[tbp]
423: \includegraphics[width=6cm]{ent}
424: \caption{Linear entropy $E_{l}$ vs $\lambda$ for $U=0,3,6,10$ by choosing $%
425: V=0.2$, $t=2.0$, $\varepsilon=0.8$. The maximum $E_l =0.5$
426: corresponds to the single-site bipolaron in strong coupling
427: region.} \label{entangl1}
428: \end{figure}
429:
430: \subsection{ ground state fidelity}
431:
432: More recently, another temporary effective quantum information
433: tools, i.e., the fidelity has been put forward to analyze
434: complicated interacting systems from the perspective of the GS wave
435: functions~\cite{Shu,Chen}, first identifying polaronic crossover
436: behaviors in bosonic system~\cite{Norman}. A simple expression of
437: the fidelity $F(g_{1},g_{1}+\delta g_{1})$ is given just by the
438: modulus of the overlap of two ground states corresponding to two
439: different coupling parameters $g_{1}$ and $g_{1}+\delta g_{1}$,
440: where $g_{1}$ is a tiny perturbation parameter~\cite{Zanardi}. It is
441: expected to signal a peak at the critical point. The general GS
442: fidelity is written as
443:
444: \begin{eqnarray}
445: F&=&\langle\psi(g_{1})|\psi(g_{1}+\delta g_{1})\rangle \nonumber\\
446: &=&|\langle\varphi_{1}(g_{1})|\varphi_{1}(g_{1}+\delta
447: g_{1})\rangle+ \langle\varphi_{2}(g_{1})|\varphi_{2}(g_{1}+\delta
448: g_{1})\rangle \nonumber\\
449: &+&\langle\varphi_{3}(g_{1})|\varphi_{3}(g_{1}+\delta
450: g_{1})\rangle|
451: \end{eqnarray}
452: where $|\psi(g_{1})\rangle$ and $|\psi(g_{1}+\delta g_{1})\rangle$
453: are two normalized GS corresponding to neighboring Hamiltonian
454: parameters. In our calculation $\delta g_{1}=10^{-4}$ is used. While
455: the fidelity susceptibility $S(g_{1})$ is regarded as a more
456: effective tool to detect the singularity at the crossover regime,
457: which reads~\cite{Cozzini}
458: \begin{equation}
459: S(g_{1})=\lim_{\delta g_{1}\longrightarrow
460: 0}[1-F(g_{1},g_{1}+\delta g_{1})]/(\delta g_{1}^{2})
461: \end{equation}
462:
463: \begin{figure}[tbp]
464: \includegraphics[width=9cm]{fid}
465: \caption{GS fidelity $F$ and its susceptibility $S$ versus $\lambda$ for $%
466: U=0,3,6,10$ by choosing $V=0.2$, $t=2.0$, $\varepsilon=0.8$. A sharper drop
467: is shown at the crossover for larger $U=10$(solid line).}
468: \label{fidelity}
469: \end{figure}
470:
471: The GS fidelity $F$ and its susceptibility $S$ as a functions of
472: effective coupling strength $\lambda$ ($=g_{1}^{2}/t$)
473: for different e-e Coulomb repulsion $U$ are displayed in Fig. \ref{fidelity}%
474: (a) and (b). A narrow dramatic drop is observed in the vicinity of
475: the bipolaron transition point as a consequence of the dramatic
476: change of the structure of the GS wave function, implying that the
477: crossover from the two-site to singlet-site bipolaron occurs. As
478: shown in Fig. \ref{fidelity}%
479: (a), the values of the GS fidelity are approximated to $1$. As is
480: known, for the quantum phase transition system the GS wave
481: functions from different sides of the level-crossing point are
482: almost orthogonal~\cite{Shu}. However, there is absent of the
483: level-crossing and then values of peaks of the GS fidelity are
484: around to $1$ rather than $0$ at the crossover point. Further
485: evidence for this crossover is given by the susceptibility $S$,
486: which shows a cusp
487: structure in the intermediate coupling regime in Fig. \ref{fidelity}%
488: (b). Obviously, the crossover becomes more abrupt and the critical
489: value of the e-ph coupling $\lambda$ where the peak appears becomes
490: larger as the Coulomb repulsion $U$ increases from $0$ to $10$. So
491: it is illustrated that the transition behavior from two-site to
492: singlet-site bipolaron can be effectively detected by a singularity
493: of the GS fidelity and its susceptibility.
494:
495:
496: \subsection{static correlation function}
497:
498: Since the bipolaron crossover transition has been indeed known
499: from the quantum information perspective. Naturally, we seek
500: classically to discuss this crossover
501: by the means of the static on-site and inter-site correlation functions $%
502: \langle n_1u_1\rangle $ and $\langle n_1u_2\rangle $, which reveal
503: the spatial extent of lattice deformations induced by electrons
504: respectively. The $u_i$ denotes the lattice deformations on site
505: $i$ produced by the electrons and $n_i$ is the number operator of
506: the electrons. The GS correlation functions are written as
507: \begin{equation}
508: \langle n_1u_{1,2}\rangle =\langle n_1[\pm (d+d^{\dagger
509: })/2-2gn_{1,2}]\rangle \label{corre}
510: \end{equation}
511: The positive (negative) sign and $n_1$ ($n_2$) are associated with $\langle
512: n_1u_1\rangle $ and $\langle n_1u_2\rangle $ respectively. By using the
513: exact GS wave function of the singlet bipolaronic state obtained above, the
514: correlation functions Eq.(\ref{corre}) can be expressed as follows
515: \begin{eqnarray}
516: &&\langle n_1u_2\rangle= -2g_{1}\sum_{m=0}^Nf_m^2- \nonumber \\
517: &&\sum_{m=1}^N\sqrt{m}(c_m^{*}c_{m-1}+0.5f_m^{*}f_{m-1}) \nonumber \\
518: &&-\sum_{m=0}^{N-1}\sqrt{m+1}(c_m^{*}c_{m+1}+0.5f_m^{*}f_{m+1})
519: \end{eqnarray}
520: \begin{eqnarray}
521: \langle n_1u_1\rangle &=&-2g_{1}\sum_{m=0}^N(4c_m^2+f_m^2) \nonumber \\
522: &&+\sum_{m=1}^N\sqrt{m}(c_m^{*}c_{m-1}+0.5f_m^{*}f_{m-1}) \nonumber \\
523: &&+\sum_{m=0}^{N-1}\sqrt{m+1}(c_m^{*}c_{m+1}+0.5f_m^{*}f_{m+1}).
524: \end{eqnarray}
525:
526: \begin{figure}[tbp]
527: \includegraphics[width=8.3cm]{corr}
528: \caption{Correlation function versus $\lambda$ (a)$-\langle
529: n_{1}u_{1}\rangle $, (b)$-\langle n_{1}u_{2}\rangle$, (c)$%
530: -\langle(n_{1}-n_{2})(u_{1}-u_{2})\rangle/g_{1}$ for $U=0,3,6,10$
531: by choosing $V=0.2$, $t=2.0$, $\varepsilon=0.8$. A more abrupt
532: crossover occurs for larger $U=10$ (solid line).}
533: \label{correlation}
534: \end{figure}
535:
536: The functions $-\langle n_{1}u_{1}\rangle$ and $-\langle
537: n_{1}u_{2}\rangle$ against the e-ph coupling strength $\lambda$
538: are plotted in Figs.~\ref {correlation} (a) and (b). One can
539: observe that $-\langle n_{1}u_{1}\rangle$ and $-\langle
540: n_{1}u_{2}\rangle$ increase monotonically with $\lambda$ in weak
541: coupling regime. When $\lambda$ exceeds a critical value, $
542: -\langle n_{1}u_{1}\rangle$ and $-\langle n_{1}u_{2}\rangle$ show
543: different behaviors. It is clearly shown that two electrons are
544: localized on one site in strong coupling region, resulting the
545: singlet-site bipolaron. It also demonstrates that the e-e
546: interaction $U$ affects the magnitude of inter-site (on-site)
547: deformations $\langle n_{1}u_{2}\rangle$ ($-\langle
548: n_{1}u_{1}\rangle$) and the transition point where $-\langle
549: n_{1}u_{2}\rangle$ reaches maximum shifts to a larger $\lambda$ as
550: $U$ increases.
551:
552: The nature of the bipolaron crossover can also be investigated by
553: the correlation function $\langle (n_1-n_2)(u_1-u_2)\rangle
554: /g_{1}$. Figs.\ref{correlation}(c) shows that the two-site
555: bipolaron regime of the coupling strength $\lambda$ is wider for
556: larger $U$ in the weak e-ph coupling regime. This crossover
557: behaviors are consistent with the above analysis on the GS
558: entanglement and fidelity.
559:
560: \section{summary and discussion}
561:
562: In this present work we have solved exactly the two-site
563: Hubbard-Holstein model by the extended phonon coherent states
564: approach. The GS energy is lower than previous results and the
565: crossover regime is revealed clearly by the electrons occupancy
566: probability. With the exact GS wave function, we study the crossover
567: properties from quantum information perspective based on the linear
568: entropy, the GS fidelity and its susceptibility. The single-site
569: bipolarons are maximally entangled and the bipolaron crossover
570: transition is more abrupt as e-e interaction $U$ increases. The
571: similar bipolaron crossover behaviors are also observed classically
572: in the static correlation function. The present study may provide
573: some insights into the more complicated Hubbard-Holstein systems in
574: a infinite chain.
575:
576: \section{Acknowledgements}
577:
578: This work was supported by National Natural Science Foundation of China,
579: PCSIRT (Grant No. IRT0754) in University in China, National Basic Research
580: Program of China (Grant No. 2009CB929104).
581:
582: $^{\dag}$Corresponding author
583:
584: \begin{thebibliography}{99}
585: \bibitem{ASAlexandrov} A. S. Alexandrov, and N. F. Mott, polarons and
586: Bipolarons, World Scientific, Singapore (1995); A. Damascelli, Z. Hussain,
587: and Z. X. Shen, Rev. Mod. Phy. 75, 473 (2003).
588:
589: \bibitem{AJMilis} A. J. Millis, Nature (London) 392,147(1998); D. M.
590: Edwards, Adv. Phys. \textbf{51}, 1259 (2003).
591:
592: \bibitem{THolstein} T. Holstein, Ann. Phy. (NY) \textbf{8}, 325 (1959).
593:
594: \bibitem{Fehske} H. Fehske, G. Wellein, J. Loos, and A. R. Bishop, Phys.
595: Rev. B \textbf{77} , 085117 (2008).
596:
597: \bibitem{Berciu} M. Berciu, Phys. Rev. B \textbf{75}, 081101(R) (2007).
598:
599: \bibitem{Scott} S. Hill, and W. K. Wootters, Phys. Rev. Lett. \textbf{78},
600: 26 (1997).
601:
602: \bibitem{William} W. K. Wootters, Phys. Rev. Lett. \textbf{80}, 10 (1998).
603:
604: \bibitem{Vidal} G. Vidal, J. I. Latorre, E.Rico, and A.Kitaev, Phys. Rev.
605: Lett. \textbf{90}, 227902 (2003).
606:
607: \bibitem{Gu} S. J. Gu, S. S. Deng, Y. Q. Li, and H. Q. Lin, Phys. Rev.
608: Lett. \textbf{93}, 086402 (2004).
609:
610: \bibitem{chenqh} Q. H. Chen, Y. Y. Zhang, T. Liu, and K. L. Wang, Phys. Rev.
611: A (in press); see also arXiv: 0809.4385.
612: \bibitem{Quan} H. T. Quan, Z. Song, X. F. Liu, P. Zanardi, and C. P. Sun,
613: Phys. Rev. Lett. \textbf{96}, 140604 (2006).
614:
615: \bibitem{Vezzani} P. Buonsante, and A. Vezzani, Phys. Rev. Lett.
616: \textbf{98}, 110601 (2007).
617:
618: \bibitem{Min} M. F. Yang, arxiv:quant-ph/07074574
619:
620: \bibitem{Shu} S. Chen, L. Wang, Y. J. Hao, and Y. P. Wang, Phys. Rev. A.
621: \textbf{77}, 032111 (2008) .
622:
623: \bibitem{Zanardi} P. Zanardi, and N. Paunkovi\'{c}, Phys. Rev. E. \textbf{74%
624: }, 0331123 (2006).
625:
626: \bibitem{Buon} P. Buonsante, and A. Vezzani, Phys. Rev. Lett. \textbf{98},
627: 110601 (2007).
628:
629: \bibitem{Chen} S. Chen, L. Wang, S. J. Gu, and Y. Wang, Phys. Rev. E.
630: \textbf{76}, 061108 (2007).
631:
632: \bibitem{Dunning} C. Dunning, J. Links, and H. Q. Zhou, Phys. Rev. Lett.
633: \textbf{94} , 227002 (2005).
634:
635: \bibitem{Norman} N. Oelkers, and J. Links, arxiv:cond-mat/0611510v2.
636:
637: \bibitem{ANDas} A. N. Das, and P. Choudhury, Phys. Rev. B \textbf{49}, 18
638: (1994).
639:
640: \bibitem{Acquarone} M. Acquaroneet al., Phys. Rev. B \textbf{58}, 7626
641: (1998).
642:
643: \bibitem{JC} J. Chatterjee, and A. N. Das, arXiv:cond-mat/0210607.
644:
645: \bibitem{Ranninger} J. Ranninger and U. Thibblin, Phys. Rev. B \textbf{45},
646: 7730 (1992).
647:
648: \bibitem{EVLde} E. V. L. de Mello, and J. Ranninger, Phys. Rev. B \textbf{55%
649: }, 14872 (1997).
650:
651: \bibitem{chen} Q. H. Chen et al., Phys. Rev. B \textbf{53}, 11296(1996).
652:
653: \bibitem{Han} R. S. Han, Z. J. Lin, and K. L. Wang, Phys. Rev. B \textbf{65}%
654: , 174303(2002)
655:
656: \bibitem{liu} K. L. Wang, T. Liu, and M. Feng, Euro. Phys. J. B \textbf{54}%
657: , 283(2006).
658:
659: \bibitem{Yang} Y. Zhao, P. Zanardi, and G.H Chen, Phys. Rev. B. \textbf{70}%
660: , 195113 (2004).
661: \bibitem{Wang} X. Wang, and B. Sanders, J. Phys. A. \textbf{38}, 67 (2005).
662:
663: \bibitem{Cozzini} M. Cozzini, P. Giorda, and P. Zanardi, Phys. Rev. B.
664: \textbf{75}, 014439 (2007).
665: \end{thebibliography}
666:
667: \end{document}
668: