1: \documentstyle[prd,aps,preprint]{revtex}
2: %\documentstyle[aps,prl,twocolum]{revtex}
3: \begin{document}
4:
5: \title{A simple variational approach to the\\ quantum Frenkel-Kontorova
6: model}
7: \author{Choon-Lin Ho$^{1,2}$ and Chung-I Chou$^3$}
8: \address{\small \sl
9: $^1$Department of Physics, Tamkang University, Tamsui 25137,
10: Taiwan\footnote{Permanent address}\\
11: $^2$Theory Division, Institute of Particles and Nuclear Studies, KEK,\\
12: Tsukuba, Ibaraki 305, Japan\\
13: $^3$Institute of Physics, Academia Sinica, Taipei 11529, Taiwan}
14:
15: %\date{}
16:
17: \maketitle
18:
19:
20: \begin{abstract}
21: We present a simple and complete variational approach to the one-dimensional
22: quantum Frenkel-Kontorova model. Dirac's time-dependent variational principle
23: is adopted together with
24: a Hatree-type many-body trial wavefunction for the atoms. The single-particle
25: state is assumed to have the Jackiw-Kerman form. We obtain an effective
26: classical Hamiltonian
27: for the system which is simple enough
28: for a complete numerical solution for the static ground state of the model.
29: Numerical results show that our simple approach captures the essence of the
30: quantum effects first observed in quantum Monte Carlo studies.
31: \end{abstract}
32:
33: \pacs{PACS numbers: 05.45.-a, 03.65.Sq, 05.30.Jp, 42.50.Dv}
34: \newpage
35:
36: {\it 1. Introduction.} The Frenkel-Kontorova (FK) model \cite{FK,FM} is a
37: simple
38: one-dimensional model used to study incommensurate structures appearing in
39: many condensed-matter systems, such as charge-density waves, magnetic spirals,
40: and adsorbed monolayers \cite{Bak}. These modulated structures arise as a
41: result of the competition between two or more length scales.
42: The FK model describes a chain of atoms connected by harmonic springs
43: subjected to an
44: external sinusoidal potential. In an important development in the study of
45: the classical FK model, Aubry \cite{Aubry} first made use of the connection
46: between the
47: FK model, the so-called ``standard map", and the Kolmogorov-Arnold-Moser
48: (KAM)
49: theorem to reveal many interesting features of the FK model. Particularly, he
50: showed that when the mean distance (also called the winding number) between
51: two successive atoms is rational, the system is always pinned. But when the
52: winding number is irrational, there exits a critical external field
53: strength below (above) which the system is unpinned (pinned). This
54: transition
55: is called by Aubry a ``transition by breaking of analyticity", and is closely
56: connected with the breakup of a KAM torus. It is very analogous to a phase
57: transition, and various critical exponents and questions of universality have
58: been extensively studied in the past.
59:
60: In recent years, the FK model has been applied to the study of transmission in
61: Josephson junction and atomic-scale friction-nanoscale tribology \cite{JT}.
62: In these cases, quantum effects are very important. Unlike the classical case,
63: study of quantum FK models is rather scanty. It was first considered in a
64: quantum Monte Carlo (QMC) analysis in \cite{BGS}. Their main observation
65: is that
66: the map appropriate to describe the quantum case is no longer the standard
67: map, but rather a map with a sawtooth shape. An explanation of this
68: phenomenon was
69: later attempted in \cite{BBC} using a mean field theory in which the
70: inclusion of the
71: contribution from quasidegenerate states is essential. But the mean field
72: computations with these states are rather involved, and the quantum map that
73: is to replace the classical standard map in different quantum regimes is not
74: clearly identified.
75: More recently, a less complicated approach was proposed in \cite{BLZ} which
76: uses the squeezed state
77: function to demonstrate that the sawtooth behavior is simply the result of
78: quantum fluctuations. In our opinion, the approach adopted in \cite{BLZ} is
79: very appealing in principle.
80: However, we believe that some difficulties in this work need to be overcome
81: before it could be considered satisfactory.
82: First, the assumed squeezed state many-body ground state is general
83: enough so
84: as to include the correlations of the positions of the atoms, expressed by the
85: covariances $G_{ij}=\langle (x_i - {\bar x}_i)(x_j - {\bar x}_j)\rangle$
86: ($i\neq j$),
87: where $x_i$ is the position of the $i$th atom, $\langle\cdots\rangle$ is the
88: expectation value in a given quantum state, and ${\bar x}_i=\langle
89: x_i\rangle$. However, to find the equilibrium state of the model, one has to
90: solve
91: a system of coupled equations of the variables $x_i$ and the $G_{ij}$. The
92: equations obtained are so complicated that the task of solving them within a
93: single
94: numerical framework is very difficult. In fact, in \cite{BLZ} a hybrid
95: numerical
96: analysis was adopted in which the equations for the $G_{ij}$ were not solved.
97: Instead,
98: the values of $G_{ij}$ were taken from QMC data. These values were then
99: treated as initial conditions in solving the equations for the atomic
100: positions
101: $x_i$'s. Technically, such hybrid analysis is not satisfactory. Second,
102: the covariance terms $G_{ij}$ ($i\neq j$) are constrained by the values
103: of the fluctuation terms $G_{ii}$ and $G_{jj}$ through the Cauchy-Schwarz
104: inequality. These constraints guarantee the boundedness from below of
105: the effective Hamiltonian \cite{BL}. But then this also calls for a proper
106: variational principle that has to take care of the interdependence of the
107: $G_{ij}$ terms.
108:
109: In this letter we shall show that
110: all the essential features observed in the QMC studies can be obtained from
111: an independent-particle picture of the many-body ground state without the
112: covariance terms. In the independent-particle picture the many-body trial
113: wavefunction are factorizable into single-particle states. One can assume the
114: single-particle
115: state to have the form of a squeeze state. For the quantum FK model,
116: a simpler and, in our
117: view, more elegant approach is to use
118: the Jackiw-Kerman (JK) function \cite{JK} as the single
119: particle state. We shall show that this simple independent-particle approach
120: produces an effective
121: classical Hamiltonian which is bounded below, admits simple numerical
122: solution of the ground state without recourse to QMC analysis, and reproduces
123: the essential features observed in QMC studies.
124:
125: {\it 2. Effective Hamiltonian.} The Hamiltonian of the quantum FK model
126: is given by
127: \begin{equation}
128: {\cal H}=\sum_i \left[\frac{{\hat p}^2_i}{2m} +
129: \frac{\gamma}{2}\left({\hat q}_{i+1}-{\hat q}_i
130: \right)^2- V\cos(l_0{\hat q}_i) \right].
131: \label{FK1}
132: \end{equation}
133: Here ${\hat q}_i$ and ${\hat p}_i$ are the position and momentum operators,
134: respectively, of the $i$th atom, $\gamma$ the elastic constant of the
135: spring, $V$ and $2\pi/l_0$ are the strength and the period of the
136: external potential. As in \cite{BBC}, it is convenient to use the
137: dimensionless variables
138: ${\hat Q}_i=l_0{\hat q}_i$, ${\hat P}_i=l_0{\hat p}_i/\sqrt{m\gamma}$,
139: and $K=Vl_0^2/\gamma$. With these new variables, we obtain
140: the following dimensionless Hamiltonian $H$
141: \begin{equation}
142: H=\sum_i \left[\frac{{\hat P}^2_i}{2} +
143: \frac{1}{2}\left({\hat Q}_{i+1}-{\hat Q}_i
144: \right)^2- K\cos({\hat Q}_i) \right].
145: \label{FK2}
146: \end{equation}
147: We have ${\cal H}=\gamma H/l_0^2$. The effective Planck constant is
148: $\tilde\hbar=\hbar l_0^2/\sqrt{m\gamma}$. For the classical FK model, the
149: Aubry transition occurs at the critical value $K_c=0.971635\cdots$.
150:
151: To study the ground state properties of the quantum FK model in (\ref{FK2}),
152: we
153: adopt here the time-dependent variational principle pioneered by Dirac
154: \cite{Dirac}. In this approach, one first constructs the effective action
155: $\Gamma=\int dt
156: ~\langle \Psi,t|i\hbar\partial_t -{\cal H}|\Psi,t\rangle$ for a given system
157: described
158: by $\cal H$ and $|\Psi,t\rangle$. Variation of $\Gamma$ is then the quantum
159: analogue of the Hamilton's principle. The time-dependent Hatree-Fock
160: approximation emerges when a specific ansatz is made for the state
161: $|\Psi,t\rangle$. We now assume the trial wavefunction of the ground state of
162: our quantum FK system to have the Hatree
163: form $|\Psi,t\rangle=\prod_i |\psi_i,t\rangle$, where the
164: normalized single-particle
165: state $|\psi_i,t\rangle$ is taken to be the JK wavefunction \cite{JK}:
166: \begin{eqnarray}
167: \langle Q_i|\psi_i,t\rangle&=&\frac{1}{(2\pi\tilde\hbar
168: G_i)^{1/4}} %\nonumber\\
169: \times \exp\Biggl\{
170: -\frac{1}{2\tilde\hbar}\left(Q_i-x_i\right)^2\Bigl[\frac{1}{2}G_i^{-1}
171: %\nonumber\\
172: -2i \Pi_i\Bigr]+\frac{i}{\tilde\hbar}p_i\left(Q_i-x_i\right)
173: \Biggr\}~.
174: \label{JK}
175: \end{eqnarray}
176: The real quantities $x_i(t)$, $p_i(t)$, $G_i(t)$ and $\Pi_i (t)$ are
177: variational parameters the variations of which at $t=\pm\infty$ are assumed to
178: vanish. The JK wavefunction can be viewed as the $Q$-representation
179: of the squeeze state \cite{TF}.
180: We prefer to use the JK form since the physical meanings of the variational
181: parameters contained in the JK wavefunction are most transparent, as we
182: shall show below. Furthermore, the JK form is in the general Gaussian form so
183: that integrations are most easily performed.
184:
185: It is not hard to
186: check that $x_i$ and $p_i$ are the expectation values of the operators
187: ${\hat Q}_i$ and ${\hat P}_i$: $x_i=\langle \Psi |{\hat Q}_i|\Psi\rangle$,
188: $p_i=\langle \Psi |{\hat P}_i|\Psi\rangle$. Also, one has
189: $\langle \Psi |({\hat Q}_i-x_i)^2|\Psi\rangle=\tilde\hbar G_i$, and
190: $\langle \Psi |i\tilde\hbar\partial_t|\Psi\rangle=\sum_i
191: (p_i\dot{x}_i-\tilde\hbar
192: G_i\dot{\Pi}_i)$, where the dot represents derivative with respect to (w.r.t.)
193: time $t$. It is now clear that
194: $\tilde\hbar G_i$ is the mean fluctuation of the position of the $i$-th atom,
195: and that $G_i>0$.
196: With these expectation values, the (rescaled) effective action $\Gamma$ for
197: the dimensionless $H$ can be worked
198: out to be $\Gamma (x,p,G,\Pi)=\int dt ~[\sum_i
199: \omega_0^{-1}~(p_i\dot{x}_i+\tilde\hbar\Pi_i
200: \dot{G}_i)-H_{eff}]$, where $\omega_0=\sqrt{\gamma/m}$ is the angular
201: frequency
202: of the spring, and $H_{eff}=\langle\Psi|H|\Psi\rangle$ is the effective
203: Hamiltonian given by
204: \begin{eqnarray}
205: H_{eff}&=&\sum_i \frac{1}{2}\left[p_i^2+\tilde\hbar\left(\frac{1}{4}G_i^{-1}
206: +4\Pi_i^2 G_i\right)\right] \nonumber\\
207: &+& \sum_i\frac{1}{2}\left(x_{i+1}-x_i\right)^2 \nonumber\\
208: &+& \sum_i\frac{\tilde\hbar}{2}\left(G_{i+1}+G_i\right) \nonumber\\
209: &-& \sum_i K\exp\left(-\frac{\tilde\hbar}{2}G_i\right)\cos x_i~.
210: \label{Heff}
211: \end{eqnarray}
212: The last term in (\ref{Heff}) can be very easily obtained from
213: $\langle\Psi|F(Q_i)|\Psi\rangle=\sum_{m=0}^\infty F^{(2m)}(x_i)({\tilde\hbar
214: G_i})^m /(2m)!!$, where $F^{(n)}(x)=\partial^n F(x)/\partial x^n$, and
215: $n!!\equiv n(n-2)(n-4)\cdots 1$. Eq.~(\ref{Heff}) is
216: bounded from below. One sees from the form of the effective action $\Gamma$
217: that $\Pi_i$ is the canonical
218: conjugate of $G_i$.
219:
220: Varying $\Gamma$ w.r.t. $x,~p,~G$ and $\Pi$ then gives the
221: equations
222: of motion in the Hatree-Fock approximation. Since we are mainly concerned
223: with
224: the static properties of the ground state of the quantum FK model, we must set
225: the time derivatives of these variables to zero. This gives the equations
226: which determine the values of variational parameters corresponding
227: to
228: the equilibrium states (which include the ground state). Equivalently, we can
229: obtain the equations for the equilibrium states by directly varying the
230: effective Hamiltonian $H_{eff}$ w.r.t. the variables.
231: Varying $H_{eff}$ w.r.t. $p_i$, $\Pi_i$, $x_i$ and $G_i$
232: give, respectively,
233: \begin{eqnarray}
234: p_i&=&0~,~~~~4\Pi_i G_i=0~,\label{e1}\\
235: x_{i+1}&-&2x_i +
236: x_{i-1}=K\exp\left(-\frac{\tilde\hbar}{2}G_i\right)\sin x_i~,\label{e2}\\
237: \frac{1}{4}G_i^{-2} &-&
238: K\exp\left(-\frac{\tilde\hbar}{2}G_i\right)\cos x_i -2=4\Pi_i^2 ~.
239: \label{e3}
240: \end{eqnarray}
241: The second equation in (\ref{e1}) implies $\Pi_i=0$ as $G_i>0$. This in turn
242: means that the right hand side of eq.(\ref{e3}) is equal
243: to zero:
244: \begin{equation}
245: \frac{1}{4}G_i^{-2} -
246: K\exp\left(-\frac{\tilde\hbar}{2}G_i\right)\cos(x_i) -2=0~.
247: \label{G}
248: \end{equation}
249: In the limit $\hbar=0$, eq.(\ref{e2}) is equivalent to the standard map.
250: We note that eq.(\ref{e2}) was also obtained in \cite{BLZ}. This is because
251: in the formulation in \cite{BLZ} the
252: covariances $G_{ij}$'s decoupled from the $x_i$ and the fluctuations $G_{ii}$
253: ($G_i$ in our case) in the variation of their Hamiltonian w.r.t $x_i$.
254: Unlike our case,
255: of course, these covariance terms do actually influence the solutions of
256: (\ref{e2}) through other equations obtained by variation of the Hamiltonian
257: w.r.t. the $G_{ii}$ and $G_{ij}$. And it is these equations that caused the
258: difficulties mentioned in the Introduction. In particular, the values of the
259: $G_{ii}$ were input from the QMC data in order to solve for the $x_i$ in
260: (\ref{e2}). Our simple approach, on the other hand, allows us to solve for
261: both the values of $x_i$ and $G_i$ coupled by eqs.(\ref{e2}) and (\ref{G})
262: consistently by a single numerical method.
263:
264: >From $p_i=\Pi_i=0$ and (\ref{Heff}), we see that the problem of finding the
265: static ground state of the quantum FK model reduces to the problem of
266: minimizing w.r.t. to $x_i$ and $G_i$ the following
267: effective potential
268: \begin{eqnarray}
269: V_{eff}&=& \sum_i\frac{1}{2}\left(x_{i+1}-x_i\right)^2 \nonumber\\
270: &+& \sum_i\frac{\tilde\hbar}{8}G_i^{-1} +
271: \sum_i\frac{\tilde\hbar}{2}\left(G_{i+1}+G_i\right) \nonumber\\
272: &-& \sum_i K\exp\left(-\frac{\tilde\hbar}{2}G_i\right)\cos x_i~.
273: \label{Veff}
274: \end{eqnarray}
275: Eq.(\ref{e2}) and
276: (\ref{G}) are just the conditions $\partial V_{eff}/\partial x_i=0$ and
277: $\partial V_{eff}/\partial G_i=0$, respectively.
278:
279: {\it 3. Numerical results.} We numerically solve for the set of variables
280: $x_i$ and
281: $G_i$ which characterize the ground state using the Newton method
282: \cite{Schell}.
283: In all our numerical computations the winding number $P/Q=610/987$, which is
284: an approximation of the golden mean winding number $(\sqrt{5}-1)/2$, is
285: used
286: with the periodic boundary condition $x_{i+Q}=x_i+2\pi P$. This winding number
287: is much more accurate than those used in previous works to approximate the
288: golden mean number, thus giving us better
289: accuracy in the computations of physical quantities related to the ground
290: state. We emphasize
291: that all values of $x_i$ and $G_i$ are determined by the same numerical method
292: consistently. In particular, we do not have to input the values of $G_i$ from
293: quantum Monte Carlo results in order to solve for the $x_i$.
294:
295: Having obtained the values of $x_i$ which give the mean positions of the
296: quantum atoms in the chain, we can compare the results with the classical
297: configuration, following \cite{BGS}, in two ways: (1) by the quantum hull
298: function,
299: which is the plot of $x_i$ (mod $2\pi$) of the atoms against their unperturbed
300: positions $2\pi i P/Q$ (mod $2\pi$); (2) by the so-called $g$-function,
301: defined by
302: \begin{equation}
303: g_i\equiv K^{-1}~(x_{i+1} - 2 x_i + x_{i-1})
304: \label{gf}
305: \end{equation}
306: versus the actual atomic positions $x_i$.
307: >From (\ref{e2}), we also have
308: \begin{equation}
309: g_i=\exp\left(-\frac{\tilde\hbar}{2}G_i\right)\sin x_i~.
310: \label{gf2}
311: \end{equation}
312: Here $G_i$ is related to $x_i$ by eq.(\ref{G}).
313: We see from this equation that quantum fluctuations $G_i$ will modify the
314: shape
315: of the classical $sine$-map. In addition to these two types of graphs, we
316: also plot the graph of $G_i$ against the unperturbed and the actual positions.
317: The formal graph was first introduced in \cite{BGS} to show the
318: strong correlation of the fluctuations of atoms' positions with their
319: unperturbed positions. We introduce the latter type of graphs here since we
320: think that such graphs provide better picture about how the quantum
321: fluctuations of the atoms are related to their actual positions.
322:
323: In Fig.~1 we show the four graphs mentioned above with different values of
324: $\tilde\hbar$ for the supercritical case $K=5$. Fig.~1(a) shows the quantum
325: hull functions. For small values of $\tilde\hbar$ the quantum hull function
326: consists of a countable set of steps discontinuities, just as in
327: the classical case: the atoms are in a
328: pinning phase. In fact, the atoms are more likely to be located near the
329: valley of the external potential well, namely, near $x_i=0$ (mod $2\pi$). As
330: the quantum effect increases, i.e., for increasing values
331: of $\tilde\hbar$, the quantum hull function gradually changes into a monotonic
332: analytic function, signifying that the system is entering the depinning phase.
333: There exists a critical value, approximately
334: $\tilde\hbar_c=6.58$ for $K=5$, above which the quantum hull function changes
335: from an nonanalytic function to an analytic one.
336: This is a quantum analogue of the Aubry transition in the classical case, and
337: can therefore be called the quantum Aubry transition.
338:
339: Next in Fig.~1(b) we show the graphs of the $g$-function. The curve defined
340: by
341: (\ref{gf2}) with $G_i$ satisfying (\ref{G}) are shown here as dashed curves
342: for different $\tilde\hbar$.
343: In the classical limit ($\tilde\hbar=0$) this curve is simply the
344: standard map ($sine$-curve). As $\tilde\hbar$ increases, the amplitude of
345: the curve decreases. For sufficiently large $\tilde\hbar$, the curve
346: resembles more closely a ``sawtooth" shape. This is first noted in QMC study
347: in \cite{BGS}. Here we see that it comes out very naturally from the equation
348: of motion (\ref{G}) and (\ref{gf2}). In the supercritical
349: case ($K=5$), when $\tilde\hbar<\tilde\hbar_c$, the positions $x_i$ of the
350: atoms cover
351: only a subset of the $g$-curves. This is in accord with the fact
352: that the atoms are in the pinning phase [{\it cf}. Fig.~1(a)]. As
353: $\tilde\hbar$ increases, the points
354: begin to spread along the $g$-curve. When $\tilde\hbar>\tilde\hbar_c$, the
355: $g$-graph is completely covered as the system has entered the depinning phase.
356:
357: Fig.~1(c) shows the quantum fluctuations $G_i$ plotted against the actual
358: atomic positions $x_i$. The dashed curves represent the curves of
359: eq.(\ref{G})
360: for different $\tilde\hbar$. For small $\tilde\hbar$, the atoms are located
361: near $x_i=0$ (mod $2\pi$) with small values of $G_i$ which means, from
362: (\ref{JK}), that the wavefunctions are highly peaked at these positions. As
363: the
364: quantum effect increases, the external potential is so modified that now the
365: atoms could be found at other positions, but with atoms at $x_i=\pi$ (mod
366: $2\pi$) having
367: the largest value of $G_i$. This indicates that wavefunctions of the atoms
368: near the
369: top of the potential are more extended with smaller amplitudes. Again, when
370: $\tilde\hbar > \tilde\hbar_c$, the curves of (\ref{G}) are completely covered
371: by the solutions $x_i$. To compare with the results in \cite{BGS}, we plot
372: the
373: values of $G_i$ against the unperturbed positions in Fig.~1(d). One sees that
374: the values of $G_i$ are strongly
375: correlated
376: with the unperturbed positions, as first noted in \cite{BGS}. For
377: $\tilde\hbar <\tilde\hbar_c$ the graphs
378: consists of steps discontinuities, and for $\tilde\hbar>\tilde\hbar_c$ the
379: graphs are continuous. This is correlated with the graphs of the quantum
380: hull function in Fig.~1(a), since from (\ref{G}) any fixed value of $x_i$
381: correspond to a fixed value of $G_i$.
382:
383: Next we show in Fig.~2 the corresponding graphs for the case $K=1.5$.
384: This represents the situation which is slightly over the critical classical
385: case. The general trends of the behavior of the graphs are the same as
386: those in Fig.~1. As expected, quantum Aubry transition takes place at a
387: smaller $\tilde\hbar_c=1.17$. We note here that the shape of the
388: $g$-function at large $\tilde\hbar$ in this case is intermediate between a
389: $sine$ and a sawtooth map.
390:
391: We have also checked the subcritical cases with $K<K_c$. The
392: classical system is already in the depinning phase in this regime. Quantum
393: fluctuations only enhance the trend of depinning. The $g$-function is
394: found to be
395: closer to a $sine$-shape with smaller amplitude for higher $\tilde\hbar$.
396: This is consistent with the QMC results \cite{BGS}.
397:
398: Finally, we note here that, while we have reproduced the essential features
399: first observed in the QMC studies of the quantum FK model, there is also
400: slight discrepancy between the
401: results of these two approaches. The difference is that,
402: for a fixed value of $K$, the QMC results \cite{BGS} indicated that the
403: sawtooth shape of the $g$-function appeared at
404: a lower value of $\tilde\hbar$, and that the atoms began to spread along the
405: $g$-curve also at a smaller $\tilde\hbar$. For example, at $K=5$ the
406: QMC
407: results showed that the above situation already appeared at $\tilde\hbar=0.2$,
408: while our results (cf. Fig.~1(b)) indicate that at $K=5$ and at a higher
409: $\tilde\hbar=2$ the system is still closer to the classical case.
410: We believe this could be explained as follows. First, our
411: independent-particle
412: wavefunction is only the lowest order approximation of the many-body
413: wavefunction of the quantum FK system. A more accurate description of the
414: system will require a better assumption of the wavefunction than that assumed
415: here. This presumably may require the inclusion of the effects of the
416: covariance terms as advocated in \cite{BLZ}, but with a more appropriate
417: variational principle to circumvent the difficulties already mentioned in the
418: Introduction. Second, our results are obtained at zero temperature, while
419: those in the QMC analysis were obtained, by the nature of the method itself, at
420: small but finite temperatures (temperature T=0.0067 as given in \cite{BGS}).
421: It
422: is natural that thermal fluctuations will cause the atoms to spread away from
423: their zero-temperature positions.
424:
425: {\it 4. Summary.} In conclusion, we have presented a simple and complete
426: variational approach
427: to the quantum FK model based on
428: a Hatree-type many-body trial wavefunction
429: of the JK form.
430: The effective Hamiltonian obtained is bounded below,
431: and is simple enough
432: for a complete numerical solution for the static ground state of the model
433: in various quantum regimes.
434: Numerical results show that our simple approach captures the essence of the
435: quantum effects first observed in QMC studies. The map appropriate for the
436: quantum FK model is well described by eq.(\ref{gf2}) and (\ref{G}).
437: In contrast to previous approaches, we do not require the existence of the
438: complicated quasidegenerate states, or the partial help from quantum Monte
439: Carlo data in order to obtain these results.
440:
441: \vskip 1 truecm
442: \centerline{\bf Acknowledgments}
443:
444: This work is supported in part by the
445: R.O.C Grant NSC 89-2112-M-032-004. Part of the work was done while one of us
446: (CLH) was visiting the Theory Division at KEK (Japan) under the auspices of
447: the
448: exchange program between KEK (Japan) and the National Center for Theoretical
449: Sciences (Taiwan). He would like to thank the
450: staff and members of the theory group of KEK for their hospitality and
451: support. After the paper was submitted, we were kindly informed by Prof. B.
452: Hu that he and W.M. Zhang in their previous incomplete work \cite{Hu} had
453: obtained
454: a Hamiltonian similar to our eq.(\ref{Heff}), but that they had not studied
455: the ground state properties.
456:
457: \begin{references}
458: \bibitem{FK} Y.I. Frenkel and T. Kontorova, Zh. Eksp. Teor. Fiz.{\bf 8}, 1340
459: (1938); Sov. Phys.-JETP {\bf 13}, 1 (1938).
460: \bibitem{FM} F. C. Frank and J. H. van der Merwe, Proc. R. Soc. London,
461: Ser. {\bf A} {\bf 198}, 205 (1949).
462: \bibitem{Bak} P. Bak, Rep. Prog. Phys. {\bf 45}, 587 (1982).
463: \bibitem{Aubry} S. Aubry, in {\it Solitons and Condensed Matter Physics},
464: ed. A.R. Bishop and T. Schneider, (Springer-Verlag, Berlin, 1978);
465: J. Phys. (Paris) {\bf 44}, 147(1983); Physica {\bf 7D}, 240
466: (1983); {\em ibid}, {\bf 8D} 381 (1983).
467: \bibitem{JT} M. G. Rozman, M. Urbakh, and J. Klafter, Phys. Rev. Lett.
468: {\bf 77} 683 (1996); Phys. Rev. {\bf E} {\bf 54}, 6485 (1996);
469: M. Weiss and F.-J. Elmer, Phys. Rev. {\bf B} {\bf 53}, 7539 (1996);
470: T. Gyalog and H. Thomas, Europhys. Lett. {\bf 37}, 195 (1996).
471: \bibitem{BGS} F. Borgonovi, I. Guarneri and D. Shepelyansky,
472: Phys. Rev. Lett. {\bf 63}, 2010 (1989); Z. Phys.{\bf B 79}, 133 (1990);
473: F. Borgonovi, {\it Ph.D} dissertation, Universit\'a Degli Studi di
474: Pavia, 1989 (unpublished).
475: \bibitem{BBC} G. P. Berman, E. N. Bulgakov and D. K. Campbell, Phys. Rev.
476: {\bf B 49}, 8212 (1994).
477: \bibitem{BLZ} B. Hu, B. Li and W.-M. Zhang, Phys. Rev. {\bf E 58}, 4068
478: (1998); B. Hu and B. Li, Quantum Frenkel-Kontorova model (to appear in
479: Physica A).
480: \bibitem{BL} B. Hu and B. Li, private communications.
481: \bibitem{JK} R. Jackiw and A. Kerman, Phys. Lett. {\bf A 71}, 158 (1979).
482: \bibitem{Dirac} P.A.M. Dirac, Proc. Cambridge Phil. Soc. {\bf 26}, 376 (1930).
483: \bibitem{TF} Y. Tsue and Y. Fujiwara, Prog. Theor. Phys. {\bf 86}, 443 (1991).
484: \bibitem{Schell} H.J. Schellnhuber, H. Urbschat and A. Block, {\sl Phys. Rev.}
485: {\bf A33}, 2856 (1986);
486: H.J. Schellnhuber, H. Urbschat and J. Wilbrink, {\sl Z. Phys.} {\bf B80},
487: 305 (1990).
488: \bibitem{Hu} B. Hu, ``The Frenkel-Kontorova Model: classical generalizations
489: and quantum glimpses" in {\it Proceedings of the 4th Drexel Conference on
490: Quantum Nonintegrability}, eds. D.H. Feng and B.L. Hu, 1997. In a brief
491: section of this paper a Hamiltonian very similar to our eq.(\ref{Heff}) (but
492: with a fixed $\tilde\hbar=1$) was presented based on squeezed state approach.
493: \end{references}
494:
495: \vskip 3 truecm
496: \centerline{\bf Figures captions}
497: \begin{description}
498:
499: \item[Fig.~1] Structure of the quantum ground state for $K=5$ and winding
500: number $P/Q=610/987$ at $\tilde\hbar=2$ (black dots), $6$ (white dots) and $7$
501: (black curve). (a) quantum hull
502: function plotted against unperturbed atomic positions; (b) $g$-function
503: plotted
504: against actual atomic positions (the dashed curves represent eq.(\ref{gf2})
505: with $G_i$ satisfying (\ref{G}); (c) and (d) quantum fluctuations $G_i$
506: plotted against the actual and
507: unperturbed positions, respectively. The dashed curves in
508: (c) represent the curves of eq.(\ref{G}) for different $\tilde\hbar$.
509:
510: \item[Fig.~2] Same as Fig.~1 for $K=1.5$ and $\tilde\hbar=0.5$
511: (black dots), $1.0$ (white dots) and $2$ (black curve).
512:
513: \end{description}
514: \end{document}
515:
516: