cond-mat0404381/KT.tex
1: \documentstyle[aps,preprint,pra,epsfig]{revtex} 
2: %\documentstyle[aps,multicol,pra,epsfig]{revtex} 
3: %\documentclass{article}                   
4:  
5: \begin{document} 
6:  
7: \draft 
8:  
9: \title{Observable Signature of the Berezinskii-Kosterlitz-Thouless Transition  
10: in a Planar Lattice of Bose-Einstein Condensates} 
11:  
12: \author{A. Trombettoni$^{1}$, A. Smerzi$^{2,3}$, and P. Sodano$^{4}$} 
13: \address{$^1$ Istituto Nazionale per la Fisica della Materia and 
14: Dipartimento di Fisica, Universita' di Parma, 
15: parco Area delle Scienze 7A, I-43100 Parma, Italy\\ 
16: $^2$ Istituto Nazionale per la Fisica della Materia BEC-CRS and 
17: Dipartimento di Fisica, Universita' di Trento, I-38050 Povo, Italy\\ 
18: $^3$ Theoretical Division, Los Alamos 
19: National Laboratory, Los Alamos, NM 87545, USA\\ 
20: $^4$ Dipartimento di Fisica and Sezione I.N.F.N., Universit\`a di  
21: Perugia, Via A. Pascoli, I-06123 Perugia, Italy\\ 
22: } 
23: \date{\today} 
24: \maketitle 
25:  
26: \begin{abstract} 
27: We investigate the possibility that Bose-Einstein condensates 
28: (BECs), loaded on a $2D$ optical lattice, undergo - at finite temperature -  
29: a Berezinskii-Kosterlitz-Thouless (BKT) transition.  
30: We show that - in an experimentally attainable range  
31: of parameters - a planar lattice of BECs is described by the $XY$ model  
32: at finite temperature. We demonstrate that  
33: the interference pattern of the expanding condensates provides  
34: the experimental signature of the BKT transition  
35: by showing that, near the critical temperature,  
36: the $\vec{k}=0$ component of the momentum distribution and  
37: the central peak of the atomic density profile sharply decrease. 
38: The finite-temperature transition for a $3D$ optical lattice 
39: is also discussed, and the analogies with superconducting 
40: Josephson junction networks are stressed through the text. 
41: \end{abstract} 
42: %\pacs{PACS: 63.20.Pw, 05.45.-a} 
43:  
44: %\begin{multicols}{2} 
45:  
46: \section{Introduction} 
47:  
48: Recent advances in the manipulation of cold atomic gases  
49: allow nowadays for storing Bose-Einstein condensates (BECs) in $1D$  
50: \cite{anderson98,cataliotti01,morsch01,hensinger01,eiermann03},  
51: $2D$ \cite{greiner01} and $3D$ \cite{greiner02} optical lattices,  
52: opening a pathway for the experimental and theoretical analysis of finite  
53: temperature transitions in atomic systems. 
54: In a way similar to the spectacular  
55: realizations obtained with superconducting Josephson networks  
56: \cite{beasley79,resnick81}, arrays of BECs can provide 
57: new experimentally realizable systems on which to test well known   
58: paradigms of the statistical mechanics: our interest here 
59: is to evidence how to detect a finite-temperature defect-mediated  
60: transition in $2D$ optical lattice of BECs.  
61:  
62: According to the well-known Mermin-Wagner theorem \cite{nelson83},  
63: two-dimensional systems with a continuous symmetry 
64: %of the Hamiltonian 
65: cannot sustain long-range order in the thermodynamic limit  
66: at finite temperature; however,  
67: a phase transition is still possible and it occurs via the 
68: unbinding of point defects like dislocations or vortices.  
69: Defect-mediated transitions are widely studied \cite{nelson83} since  
70: they provide crucial insights for a wide variety of 
71: experiments on thin films. The BKT transition is the 
72: paradigmatic example of a defect-mediated transition \cite{nelson83}  
73: which is exhibited by the two-dimensional  
74: $XY$ model \cite{minnaghen87,kadanoff00},  
75: describing $2$-components spins on a two-dimensional lattice.  
76: In the low-temperature phase, characterized by the presence of  
77: bound vortex-antivortex pairs, the spatial correlations exhibit a power-law  
78: decay; above a critical temperature $T_{BKT}$, the decay is exponential  
79: and there is a proliferation of free vortices.  
80: The BKT transition has been observed in superconducting Josephson arrays  
81: \cite{resnick81} and its predictions are well verified in  
82: the measurements of the superfluid density  
83: in $^4He$ films \cite{bishop78}. 
84: 
85: In finite magnetic systems the BKT transition point  
86: is signaled by the drop to $0$ of a suitably defined 
87: magnetization \cite{berezinskii73}; the existence  
88: of such a magnetization in finite systems does not contradict  
89: the Mermin-Wagner theorem, since the latter 
90: is valid only in the thermodynamic limit.  
91: In a large class of experimental situations, the finite size  
92: $XY$ model predicts magnetization exponents 
93: in agreement with the measurements  
94: carried out for layered magnets with planar spin symmetry \cite{bramwell93}. 
95: We shall show in the sequel that the Fourier transform 
96: at $\vec{k}=0$ of the condensates wavefunction provide 
97: the analogous of this magnetization for atomic systems, 
98: which are naturally finite (since they are trapped by an harmonic potential).
99:  
100: In this paper we show that - at a critical temperature $T_{BKT}$ 
101: lower than the temperature $T_{BEC}$ at which condensation  
102: in each well occurs - $2D$ lattices of BECs may undergo a 
103: phase transition to a superfluid regime  
104: where the phases of the single-well condensates are coherently aligned 
105: allowing for the observation  
106: of a Berezinskii-Kosterlitz-Thouless (BKT) transition.  
107: We show in the following that the recently realized  
108: $2D$ optical lattice of Bose-Einstein  
109: condensates \cite{greiner01} may allow for the observation  
110: of a BKT transition in a finite bosonic system, 
111: since the thermodynamic properties  
112: of the bosonic lattice at finite temperature may be well described,  
113: in a suitable and experimentally attainable range of parameters,  
114: by the Hamiltonian of the $XY$ model. It is easy to convince oneself 
115: that the transition we propose to observe  
116: is different from the quantum phase transition reported in \cite{greiner02}, 
117: where the system is at $T=0$ and the insulator phase 
118: (signaled by a reversible 
119: destruction of the phase coherence across the lattice) 
120: is reached varying the optical lattice  
121: parameters: at variance, here, 
122: we propose to fix the system parameters in the superfluid phase and  
123: increase the temperature $T$ until the thermally induced vortex  
124: proliferation determines the BKT transition.   
125:  
126: We shall also show that the experimental signature  
127: of the BKT transition in bosonic planar lattices  
128: is obtained by measuring, as a function of the temperature,  
129: the central peak of the interference pattern  
130: of the expanding condensates released  
131: from the trap when the confining potentials are switched off.  
132: In fact, the peak of the momentum distribution at  
133: $\vec{k}=0$ may be regarded as the magnetization of a  
134: finite size $2D$ $XY$ magnet, and,  
135: as shown in the following, it should exhibit a sharp decrease  
136: around $T_{BKT}$. As a consequence, also the central peak of the atomic  
137: density profile decreases around $T_{BKT}$, signaling the occurrence  
138: of the BKT transition. 
139: 
140: The plan of the paper is the following: 
141: in Section II we discuss the Hamiltonian  
142: describing a $2D$ lattice of Bose-Einstein condensates at finite temperature  
143: and we show that - in a suitable range of parameters - 
144: the system is described by 
145: the $XY$ Hamiltonian. 
146: %The validity limits of such a description will be discussed. 
147: %The coefficients of the Hamiltonian are variationally  
148: %estimated and compared with the findings for $1D$ and 
149: %$3D$ optical lattices. 
150: In Section III we show how the signature of the BKT transition 
151: may be evidenced in a bosonic planar lattice; there we shall also 
152: discuss interesting analogies 
153: with superconducting Josephson junction networks.
154: %We also present results for the finite-temperature transition in 
155: %$3D$ optical lattices. 
156: Section IV is devoted to our concluding remarks.
157: %The Appendixes contain technical material  
158: %omitted for brevity in the main text: 
159: In Appendix A we report the details of a variational estimate 
160: of the coefficients of the Hamiltonian describing 
161: $1D$, $2D$ and $3D$ optical lattices, while Appendix B contains the  
162: analytical computation of the Fourier transform of a vortex on a lattice.  
163:  
164: \section{The Finite-Temperature Hamiltonian for the System} 
165:  
166: For $2D$ optical lattices, when the polarization vectors of the two  
167: standing wave laser fields are orthogonal, the resulting periodic  
168: potential for the atoms is 
169: \begin{equation} 
170: V_{opt}(\vec{r})=V_0 [ \sin^2{(kx)} + \sin^2{(ky)} ] 
171: \label{opt_potential_2D} 
172: \end{equation} 
173: where $k=2 \pi/\lambda$ is the wavevector of the laser beams.  
174: The potential maximum of a single standing wave,   
175: $V_0=sE_R$, may be controlled by changing the intensity of the optical  
176: lattice beams, and it is conveniently measured in units of the recoil energy 
177: $E_R=\hbar^2 k^2/2m$ ($m$ is the atomic mass),  
178: while, typically, $s$ can vary from $0$ up to $30$ (gravity  
179: is assumed to act along the $z$-axis).  
180: 
181: Usually, superimposed to the optical potential, 
182: there is an harmonic magnetic potential   
183: \begin{equation} 
184: V_{m}(\vec{r})=\frac{m}{2} [ \omega_r^2 (x^2+y^2) + \omega_z^2 z^2 ], 
185: \label{magn_potential_2D} 
186: \end{equation} 
187: where $\omega_z$ ($\omega_r$) is the axial (radial) trapping frequency.  
188: The minima of the $2D$ periodic potential (\ref{opt_potential_2D}) 
189: are located at the points $\vec{j}=(j_1,j_2) \cdot \frac{\lambda}{2}$  
190: with $j_1$ and $j_2$ integers, and the potential around the minima is 
191: $V_{opt} \approx \frac{m}{2} \tilde{\omega}_r^2 [ (x-\lambda j_1/2)^2 +   
192: (y-\lambda j_2/2)^2 ]$ with  
193: \begin{equation} 
194: \tilde{\omega}_r=\sqrt{2 V_0 k^2 / m} 
195: \label{omega_tilde} 
196: \end{equation} 
197: providing the frequency of the wells. 
198: When $\tilde{\omega}_r \gg \omega_r,\omega_z$,  
199: the system realizes a square array of tubes, i.e. an array of harmonic traps  
200: elongated along the $z$-axis \cite{greiner01}.  
201: Although the axial dynamics is very interesting in its  
202: own right \cite{pedri03}, in the following we analyze only  
203: the situation in which the axial degrees of freedom are frozen out,  
204: which is realizable if $\omega_z$ is sufficiently large. We also 
205: assume that the harmonic oscillator length  
206: of the magnetic potential $\sqrt{\hbar/m \omega_r}$ is  
207: larger than the size $L$ (in lattice units) of the optical  
208: lattice in order to reduce density inhomogeneity effects and to allow for  
209: a safe control of finite size effects. 
210: %We mention here that a very intersting  
211: %issue to address (but, as said, not considered here)  
212: %would be the effect of the inhomogeneity due to the harmonic 
213: %confining potential on the BKT transition. 
214: In addition, when all the relevant energy scales  
215: are small compared to the excitation  
216: energies one can safely expand the field operator \cite{jaksch98} as  
217: \begin{equation} 
218: \hat{\Psi}(\vec{r},t)= \sum_{\vec{j}} \hat{\psi}_{\vec{j}}(t)  
219: \Phi_{\vec{j}}(\vec{r}) 
220: \label{TB}  
221: \end{equation} 
222: with $\Phi_{\vec{j}}(\vec{r})$ 
223: the Wannier wavefunction localized at the $\vec{j}$-th well  
224: (normalized to $1$) and $\hat{N}_{\vec{j}}=\hat{\psi}^{\dag}_{\vec{j}}  
225: \hat{\psi}_{\vec{j}}$ the bosonic number operator. 
226: Substituting the expansion in the full quantum Hamiltonian  
227: describing the bosonic system, leads to  
228: the Bose-Hubbard model \cite{fisher89,jaksch98}  
229: \begin{equation} 
230: \hat{H}=-K \sum_{<\vec{i}, \vec{j}>}  
231: (\hat{\psi}^{\dag}_{\vec{i}} \hat{\psi}_{\vec{j}}+ h.c.) 
232: + {U \over 2} \sum_{\vec{j}} \hat{N}_{\vec{j}}  
233: (\hat{N}_{\vec{j}}-1)  
234: \label{BHM} 
235: \end{equation} 
236: where $\sum_{<\vec{i}, \vec{j}>}$ denotes a sum over  
237: nearest neighbours, $ U = (4 \pi \hbar^2 a / m) \int d\vec{r} \,  
238: \Phi_{\vec{j}}^4$ ($a$ is the $s$-wave scattering length) and  
239: $K \simeq - \int d\vec{r} \, \big[ \frac{\hbar^2}{2m}  
240: \vec{\nabla} \Phi_{\vec{i}} \cdot \vec{\nabla} \Phi_{{\vec{j}}} +  
241: \Phi_{\vec{i}}  V_{ext} \Phi_{\vec{j}} \big]$.  
242:   
243: Upon defining  
244: \begin{equation} 
245: J = 2 K N_0 
246: \label{Jos_energy} 
247: \end{equation}  
248: (where $N_0$ is the average number of atoms per site),    
249: when $N_0 \gg 1 $ and $J/N_0^2 \ll U$, the Bose-Hubbard model reduces to  
250: \begin{equation} 
251: \hat{H}=H_{XY}- {U \over 2}\sum_{\vec{j}}  
252: \frac{\partial^2}{\partial \theta_{\vec{j}}^2}, 
253: \label{QPM} 
254: \end{equation}  
255: which describes the so-called quantum phase model  
256: (see e.g. \cite{simanek94,fazio01}): in Eq.(\ref{QPM}) 
257: $\theta_{\vec{j}}$ is the phase of the $j$-th condensates and  
258: \begin{equation} 
259: H_{XY}=-J \sum_{<\vec{i},\vec{j}>} \cos{(\theta_{\vec{i}}- 
260: \theta_{\vec{j}})} 
261: \label{X-Y} 
262: \end{equation} 
263: stands for the Hamiltonian of the classical $XY$ model.  
264: When $J \gg U$ and at temperatures $T \gg U/k_B$, 
265: %the quantum phase Hamiltonian (\ref{QPM}) may be  
266: %well approximated by Eq.(\ref{X-Y}); therefore,  
267: the pertinent partition function  
268: describing the thermodynamic behaviour  
269: of the BECs stored in an optical lattice may be computed  
270: using the classical $XY$ model (\ref{X-Y}). 
271:  
272: The close relationship of the Bose-Hubbard model (\ref{BHM}) and 
273: the quantum phase model (\ref{QPM}) is by now well known 
274: \cite{fisher88} and it has been  
275: widely used in the analysis of superconducting Josephson networks; 
276: since, in superconducting networks, 
277: $N_0$ is the average number of excess Cooper pairs at a given site, 
278: the requirement $N_0 \gg 1 $ is   
279: of course satisfied and the condition $J/N_0^2 \ll U$ may 
280: also be easily realized 
281: for reasonable values of $J$. 
282: %the two models have been thought as equivalent for most practical 
283: %purposes. 
284: At variance, in bosonic lattices  
285: $N_0$ varies usually between $\sim 1$ and $\sim 10^3$ and the validity  
286: of the mapping between the Bose-Hubbard model and 
287: the quantum phase Hamiltonian is not always guaranteed; 
288: furthermore, in order to get the $XY$ model (\ref{X-Y}) from the  
289: Bose-Hubbard model (\ref{BHM}) $U$ should be much smaller than $J$, 
290: but not vanishing,  
291: in order to satisfy to the condition $J/N_0^2 \ll U$.  
292: Of course the Bose-Hubbard model - for $U=0$ -   
293: describes harmonic oscillators and  
294: thus cannot sustain any BKT transition; nevertheless, 
295: when $N_0 \gg 1 $ and $J/N_0^2 \ll U \ll J$, 
296: the Bose-Hubbard Hamiltonian reduces in the $XY$ model  
297: (\ref{X-Y}), which {\em do} display the BKT transition.  
298:  
299: A simple estimate of the coefficients $J$ and  
300: $U$ may be obtained by  
301: approximating the Wannier functions in Eq.(\ref{TB}) 
302: with gaussians, whose widths are 
303: determined variationally. 
304: The details of the computation are reported in Appendix A.  
305: In Fig.1 we plot $J/U$ as a function of $V_0$  
306: for a $2D$ optical lattice; for comparison, we plot the same ratio  
307: for $1D$ and $3D$ optical lattices also.  
308: Since, in a mean-field approach, the Mott insulator-superfluid transition  
309: occurs at $T=0$ approximately at $2 z J/U \sim 1$ \cite{fisher89,fazio01}  
310: ($z$ is the number of nearest-neighbours), from Fig.1  
311: one sees that it is much easier to detect a quantum phase transition   
312: for the three-dimensional array (and indeed it has been recently detected   
313: \cite{greiner02}); no quantum phase transition has yet been  
314: reported in literature for $2D$ and $1D$ optical lattices, since   
315: a larger laser power $V_0$ is required for its observation.  
316:  
317: For a $2D$ lattice with $V_0$ between $20$ and $25E_R$ (and  
318: $N_0 \approx 170$ as in \cite{greiner01}),  
319: the conditions $J \gg U \gg J/N_0^2$ are rather well  
320: satisfied and the BKT critical temperature, $T_{BKT} \sim J/k_B$,  
321: is between $10$ and $30nK$. Since the BKT transition for the $XY$ model  
322: occurs at a temperature $T_{BKT} \sim J/k_B$, 
323: it should be possible to evidence  
324: the transition with measurements performed  
325: at different temperatures. 
326: Using a coarse-graining approach to determine the finite-temperature  
327: phase-boundary line of the Bose-Hubbard model \cite{kampf93}, 
328: yields, in $2D$ and for $N_0 \gg 1$ and $J / U \gg 1 $, 
329: a critical temperature $T_{BKT} \approx J / k_B$, 
330: in reasonable agreement with the $XY$ estimates of $T_{BKT}$; 
331: in addition, in the chosen range of parameters, the condition  
332: $U N_0^2/J=U N_0/K \gg 1$, satisfies the finite-temperature stability 
333: criterion recently derived by Tsuchiya and Griffin  \cite{tsuchiya03}.  
334:  
335: We notice that the $XY$ Hamiltonian,  
336: in the limits $N_0 \gg 1$ and $J \gg U \gg J/N_0^2$, can be 
337: also retrieved   
338: extending the Gross-Pitaevskii Hamiltonian to  
339: finite temperature $T \gg U/k_B$.  
340: Replacing the tight-binding ansatz  
341: $\Psi(\vec{r},t)= \sum_{\vec{j}} \psi_{\vec{j}}(t)  
342: \Phi_{\vec{j}}(\vec{r})$ (where $\psi_{\vec{j}}$ are classical fields  
343: and not operators) in the Gross-Pitaevskii Hamiltonian  
344: one gets the lattice Hamiltonian  
345: \begin{equation}
346: H=-K \sum_{<\vec{i}, \vec{j}>}  
347: (\psi^{\ast}_{\vec{i}} \psi_{\vec{j}}+ c.c.) 
348: + \frac{U}{2} \sum_{\vec{j}} N_{\vec{j}} (N_{\vec{j}}-1),
349: \label{CBHM}
350: \end{equation}  
351: which is the classical version of the Bose-Hubbard model   
352: ($N_{\vec{j}} \equiv \mid \psi_{\vec{j}}\mid^2$).  
353: Writing $N_{\vec{j}}=N_0 + \delta N_{\vec{j}}$, one may  
354: neglect  
355: the quadratic terms [i.e. $(\delta N_{\vec{j}})^2$]  
356: in the hopping part of the Hamiltonian (\ref{CBHM}), which then
357: reduces in the considered limit to $H_{XY}$ (\ref{X-Y}).  
358:  
359: Accurate Monte Carlo simulations yield - for the  
360: $XY$ model - the BKT critical temperature  
361: $T_{BKT}=0.898J/k_B$ \cite{gupta88}.
362: When $U \ll J$, a BKT transition still occurs at a slightly  
363: lower critical temperature $T_{BKT}(U)$.  
364: Intuitively speaking, when $U$ increases, the superfluid region  
365: in the phase diagram decreases and, thus, one has $T_{BKT}(U) < T_{BKT}$  
366: \cite{fazio01}.  
367: An estimate of the dependence of $T_{BKT}(U)$ on $U$ obtained 
368: using a renormalization group approach has been reported in \cite{smerzi04}:  
369: when $U \ne 0$, but still $U/J \ll 1$, the only effect of the interacting  
370: term amounts to renormalize $J$ and to 
371: lower the BKT critical temperature \cite{rojas96}.  
372: For this reason in the following we shall present results from Monte Carlo  
373: simulations of the classical $XY$ model only.  
374: 
375: In numerical simulations of the finite $XY$ lattice one defines the critical  
376: temperature as the temperature $T_C(L)$  
377: at which the correlation length equals the size $L$ of the square lattice  
378: \cite{bramwell93}; of course, as $L \to \infty$, $T_C(L) \to T_{BKT}$.  
379: For $T=T_C(L)$, the BKT transition is signaled in a finite system  
380: by the fact that a suitably defined magnetization,  
381: defined by Eq.(\ref{M}), drops to zero \cite{bramwell93}.  
382:  
383: The emerging physical picture is the following:  
384: There are two relevant temperatures for the system,  
385: the temperature $T_{BEC}$, at which in each well there is a condensate,  
386: and the temperature $T_{BKT}$ at which the condensates phases  
387: start to coherently point in the same direction. Of course,  
388: in order to have well defined condensates phases one should have  
389: $T_{BKT} < T_{BEC}$. 
390: The critical temperature $T_{BEC}$ is given by $T_{BEC} \approx 0.94 \hbar N_0^{1/3}  
391: \bar{\omega}/k_B$ where $\bar{\omega}= (\tilde{\omega}_r^2 \omega_z)^{1/3}$  
392: \cite{pitaevskii03}. With the numerical values given in \cite{greiner01},  
393: $T_{BEC}$ turns out to be   
394: $ \gtrsim \ 500 nK$ for $s \gtrsim 20$. When $T < T_{BEC}$,  
395: the atoms in the well $\vec{j}$  
396: of the $2D$ optical lattice may be described by a macroscopic  
397: wavefunction $\psi_{\vec{j}}$.  
398: Furthermore, when the fluctuations around  
399: the average number of atoms per site $N_0 \gg 1$ are strongly suppressed,  
400: one may put, apart from the factor $\sqrt{N_0}$ constant across the array,   
401: $\psi_{\vec{j}}=e^{i \theta_{\vec{j}}}$.  
402: The temperature $T_{BKT}$ is of order  
403: of $J/k_B$: with the experimental parameters of  
404: \cite{greiner01} and $V_0$ between $20$ and $25E_R$,  
405: one has that $T_{BKT}$ is between $10$ and $30nK$, which is  
406: sensibly smaller than the condensation temperature  
407: $T_{BEC}$ of the single well.  
408: 
409: A similar scenario describes   
410: also the phases of planar arrays of superconducting Josephson junctions  
411: \cite{simanek94,fazio01}: they  
412: exhibit a temperature $T_{BCS}$ at which the metallic grains  
413: placed on the sites of the array become (separately) superconducting  
414: and the Cooper pairs may be described by macroscopic wavefunctions.  
415: At a temperature $T_{BKT}<T_{BCS}$,  
416: the array undergo a BKT transition and the system - as a whole -  
417: becomes superconducting. 
418:  
419: \section{Observable Signature of the BKT Transition} 
420:  
421: In this Section we shall evidence that the 
422: experimental signature of the BKT transition in bosonic planar lattices  
423: is obtained by measuring, as a function of the temperature,  
424: the central peak of the interference pattern  
425: obtained after turning off the confining potentials.  
426: 
427: Firstly, we show that the peak of the momentum distribution at  
428: $\vec{k}=0$ is the direct analog of the magnetization of a  
429: finite size $2D$ $XY$ magnet. In fact, 
430: for the $XY$ magnets, the spins can be written  
431: as $\vec{S}_{\vec{j}}=(\cos{\theta_{\vec{j}}},\sin{\theta_{\vec{j}}})$ and  
432: the magnetization is defined as $M=(1/N) \cdot   
433: \langle \, \mid \sum_{\vec{j}} \vec{S}_{\vec{j}} \mid \, \rangle$  
434: where $\langle \cdots \rangle$ stands for the thermal average; 
435: a spin-wave analysis at low temperatures yields  
436: $M=(2N)^{-k_B T/8 \pi J}$ \cite{bramwell93,berezinskii73}.  
437: With discrete BECs at $T=0$, all the phases  
438: $\theta_{\vec{j}}$ are equal at the equilibrium and  
439: the lattice Fourier transform of $\psi_j$, 
440: \begin{equation}
441: \tilde{\psi}_{\vec{k}}=\frac{1}{N}  
442: \sum_{\vec{j}} \psi_{\vec{j}} \, \, \, 
443: e^{-i \vec{k} \cdot {\vec{j}}},
444: \label{FT}
445: \end{equation}  
446: exhibits a peak at $\vec{k}=0$ ($\vec{k}$  
447: is in the first Brillouin zone of the $2D$ square lattice); 
448: the magnetization is then  
449: \begin{equation} 
450: M=\langle \, \mid \tilde{\psi}_0 \mid \, \rangle. 
451: \label{M} 
452: \end{equation}    
453: The intuitive picture of the BKT transition is the following:  
454: at $T=0$, all the spins point in the same direction. 
455: %Increasing the temperature, bound vortex pairs are thermally induced.   
456: As one can see from Fig.2(A), a single free vortex modifies the phase  
457: distribution far away from the vortex core and, thus, the modulus 
458: square of  
459: its lattice Fourier transform $\mid \tilde{\psi}_{\vec{k}} \mid^2$ 
460: has a minimum at $\vec{k}=0$. At variance, a vortex-antivortex   
461: pair [see Fig.2(B)] modifies the phase distribution only near the center of  
462: the pair (in this sense is a local {\em defect}) and 
463: its lattice Fourier transform  
464: has a maximum at $\vec{k}=0$. Analytical expressions for the  
465: lattice Fourier transform $\tilde{\psi}_{\vec{k}}$ of a single vortex and a  
466: vortex-antivortex pair may be worked out in detail 
467: and are reported in the Appendix B.  
468: 
469: Upon increasing the temperature, vortices are thermally induced.  
470: For $T < T_{BKT}$ only bound vortex pairs are present, and  
471: on average the spins continue to point in the same direction.  
472: When the condensates expand, a large peak (i.e., a magnetization)  
473: is observed in the central $\vec{k}=0$ momentum component,  
474: as shown in Fig.2(C). Rising further the temperature,  
475: due to the increasing number of vortex pairs, the central peak density  
476: decreases. For $T \approx T_{BKT}$,  
477: the pairs start to unbind and free vortices begin to appear  
478: [see Fig.2(D)], determining  
479: a sharp decrease around $T_{BKT}$ of the magnetization.  
480: At high temperatures, only free vortices are present, 
481: leading to a randomization  
482: of the phases and to a vanishing magnetization.  
483: In Fig.3 we plot the intensity of the central peak of the  
484: momentum distribution (normalized to the value at $T=0$) in a 2D  
485: lattice as a function of the reduced temperature $k_B T/J$,  
486: evidencing the sharp decrease  
487: of the magnetization around the BKT critical temperature.  
488:  
489: %Before to discuss the interference patterns of the expanding condensates,  
490: %we briefly comment on what happens for a $3D$ optical lattice resulting form  
491: %the potential $V_{opt}(\vec{r})=
492: %V_0 [ \sin^2{(kx)} + \sin^2{(ky)} + \sin^2{(kz)}]$.  
493: %In a $3D$ lattice, and with $N_0 \sim 100$, the system maps on the $3D$ $XY$  
494: %model, which is the prototype of the universality class including the 
495: %$\lambda$ normal-superfluid transition in $^4He$ \cite{tilley90}. 
496: %Also in this case, the central peak of the interference pattern 
497: %tends to zero at the critical temperature, Fig.4.  
498: %We recall, however, that the $3D$ transition is a simple order-disorder  
499: %transition (whose critical temperature $T_c$  
500: %is given in the thermodynamic limit by $2.202J/k_B$  
501: %\cite{gottlob93}), and is not mediated by  
502: %the creation of vortices-antivortices pairs as in the 2D case. 
503: %A natural question  
504: %arising from the comparison of Figs.3 and 4 is how to 
505: %clearly characterize the  
506: %two transitions: to answer, one may observe that the critical exponent  
507: %$\beta$ (defined by $M \propto (T_c-T)^\beta$) is 
508: %$\beta \approx 0.23$ for the  
509: %$2D$ $XY$ universality class \cite{bramwell93} and 
510: %$\beta \approx 0.35$ for the  
511: %$3D$ $XY$ universality class \cite{campostrini01}. 
512:  
513: Let us now turn our attention to the interference patterns  
514: of the expanding condensates: after the bosonic system reaches the  
515: equilibrium, one may switch off both the harmonic trap and the optical 
516: lattice. At this time ($t=0$) the momentum distribution  
517: $\tilde{\psi}(\vec{p},t=0)$ is given by the Fourier transform of  
518: $\psi(\vec{r},t=0)=\sum_{\vec{j}} \psi_{\vec{j}} \Phi_{\vec{j}}(\vec{r})$: 
519: one finds 
520: \begin{equation}
521: \tilde{\psi}(\vec{p},t=0)= \tilde{\Phi}(\vec{p}) \tilde{\psi}_{\vec{k}},
522: \label{relation}
523: \end{equation}
524: where $\tilde{\Phi}(\vec{p})$ is the Fourier transform  
525: of the $3D$ Wannier functions and $\hbar \vec{k}=(p_x,p_y)$ is the  
526: momentum projection on the  
527: first Brillouin zone of the $2D$ optical lattice.  
528: Using a gaussian for the Wannier functions,  
529: $\Phi(\vec{r}) \propto e^{-x^2/2\sigma_x^2}  
530: e^{-y^2/2\sigma_y^2}e^{-z^2/2 \sigma_z^2}$  
531: (with $\sigma_z \gg \sigma \equiv \sigma_x=\sigma_y$,  
532: since $\tilde{\omega}_r \gg \omega_z$),  
533: one has $\tilde{\Phi}(\vec{p})=\chi(p_x) \chi(p_y) \chi(p_z) \propto  
534: e^{-\sigma^2 (p_x^2+p_y^2)/2\hbar^2}$, where $\chi(p_x) \propto  
535: e^{-\sigma_x^2 p_x^2/ 2 \hbar^2}$ and similarly for $\chi(p_y)$ and  
536: $\chi(p_z)$.  
537: Since for $s \gg1$ $\sigma / d =1/ \pi s^{1/4}$,  
538: the momentum distribution exhibits well  
539: pronounced peaks in the centers of each Brillouin zone: 
540: these peaks have different heights and the central $\vec{k}=0$ peak  
541: is the largest (see Figs.2 and 3 of \cite{greiner01}). 
542:   
543: At $t=0$ the amplitude of the $\vec{k}=0$ peak of the  
544: momentum distribution is simply given by the thermal average  
545: of $\tilde{\psi}_{0}$. By measuring the $\vec{k}=0$  
546: peak (i.e., $\langle \, \mid \tilde{\psi}_0  \mid^2 \,    \rangle$)  
547: at different temperatures, one obtains the results plotted in Fig.2.  
548: The figure has been obtained using a Monte Carlo simulation of the  
549: $XY$ magnet for a $40 \times 40$ array: we find $k_B T_C \approx 1.07 J$.  
550: In Fig.2 we also plot the low-temperature  
551: spin wave prediction \cite{berezinskii73} (solid line),  
552: as well as a fit, first used in  
553: \cite{bramwell93}, valid near $T_C$  
554: and derived form the renormalization group equations (dotted line).  
555: At times different from   
556: $t=0$, the density profiles  
557: are well reproduced by the free expansion of the ideal gas:  
558: one obtains $\tilde{\psi}(\vec{p},t)=\chi(p_z)  
559: e^{-i[(p_z+mgt)^3-p_z^3]/6m^2g\hbar} \tilde{\varphi}(p_x,p_y,t)$, i.e.  
560: a uniformly accelerating motion along $z$ and a free motion in the  
561: $x-y$ plane, with $\tilde{\varphi}(p_x,p_y,t)=  
562: \chi(p_x) \chi(p_y) \tilde{\psi}_{\vec{k}}e^{-i (p_x^2+p_y^2) t/2\hbar m}$ 
563: giving the central and lateral peaks of  
564: the momentum distribution as a function of time for different  
565: temperatures.  
566:  
567: An intense experimental work is now focusing on the  
568: Bose-Einstein condensation in two dimensions:  
569: at present a crossover to two-dimensional behaviour  
570: has been observed for $Na$ \cite{gorlitz01} and $Cs$ atoms \cite{hammes03}.  
571: Our analysis relies on two basic assumptions; namely,  
572: the validity of the tight-binding approximation  
573: for the Bose-Hubbard Hamiltonian and the requirement  
574: that the condensate in the optical lattice may be regarded as planar  
575: \cite{petrov00}. It is easy to see that the first assumption  
576: is satisfied if, at $T=0$, $V_0 \gg \mu$  
577: (where $\mu$ is the chemical potential), and, at finite temperature,  
578: $\hbar \tilde{\omega}_r \gtrsim k_B T$. The second assumption is much more  
579: restrictive since it requires freezing the transverse excitations;  
580: for this to happen one should require a condition on the transverse  
581: trapping frequency $\omega_z$. Namely, one should have that, at $T=0$,  
582: $\hbar \omega_z \gtrsim 8 K$ and that, at finite temperature,  
583: $\hbar \omega_z \gtrsim k_B T$ (since $\omega_z \ll \tilde{\omega}_r$,  
584: the latter condition also implies that $\hbar \tilde{\omega}_r  
585: \gtrsim k_B T$). In \cite{greiner01} it is   
586: $V_0=s \cdot k_B \cdot 0.15 \mu K$ and, for $s \gtrsim 20$,   
587: the tight-binding conditions are satisfied since  
588: $10 Hz \gtrsim 8K/2\pi \hbar$; furthermore, if $\omega_z=2 \pi \cdot 1kHz$,  
589: one may safely regard our finite temperature analysis to be valid at least  
590: up to $T \sim 50 nK$. We notice that the experimental signature for the   
591: BKT transition for a continuous  
592: (i.e., without optical lattice) weakly 
593: interacting 2D Bose gas \cite{prokofev01} is also given  
594: by the central peak of the atomic density of the  
595: expanding condensates. 
596:  
597: \section{Concluding Remarks} 
598:  
599: Our paper analyzes the finite-temperature phase transitions in $2D$  
600: lattices of BECs. Our study evidences the possibility that Bose-Einstein 
601: condensates loaded on a $2D$ optical lattice may exhibit  
602: - at finite temperature - a new coherent  
603: behaviour in which all the phases of the condensates 
604: located in each well  
605: of the lattice point in 
606: the same direction. The finite-temperature transition, which is due  
607: to the thermal atoms in each well, is mediated by vortex defects and may  
608: be experimentally detectable by looking at the 
609: interference patterns of the expanding condensates. Our analysis 
610: relies on the approximation that the effect of the shallow confining 
611: harmonic trap in the $x-y$ plane may be neglected: for a 
612: tight confinement, we expect that interesting mesoscopic effects 
613: will come to play, affecting the BKT transition.
614: 
615: We observe that 
616: for a $3D$ optical lattice (resulting form  
617: the potential $V_{opt}(\vec{r})=
618: V_0 [ \sin^2{(kx)} + \sin^2{(ky)} + \sin^2{(kz)}]$) 
619: with $N_0 \sim 100$, the system maps on the $3D$ $XY$  
620: model, which is the prototype of the universality class including the 
621: $\lambda$ normal-superfluid transition in $^4He$ \cite{tilley90}. 
622: Also in this case, the central peak of the interference pattern 
623: tends to zero at the critical temperature, Fig.4.  
624: We recall, however, that the $3D$ transition is a simple order-disorder  
625: transition (whose critical temperature $T_c$  
626: is given in the thermodynamic limit by $2.202J/k_B$  
627: \cite{gottlob93}), and is not mediated by  
628: the creation of vortices-antivortices pairs as in the 2D case. 
629: A natural question  
630: arising from the comparison of Figs.3 and 4 is how to 
631: clearly characterize the  
632: two transitions: to answer, one may observe that the critical exponent  
633: $\beta$ (defined by $M \propto (T_c-T)^\beta$) is 
634: $\beta \approx 0.23$ for the  
635: $2D$ $XY$ universality class \cite{bramwell93} and 
636: $\beta \approx 0.35$ for the  
637: $3D$ $XY$ universality class \cite{campostrini01}. 
638: 
639: In conclusion, our analysis strengthens 
640: - and extends at finite temperature - the  
641: striking and deep analogy of bosonic systems  
642: with superconducting Josephson junction arrays \cite{anderson98}.  
643:  
644: {\em Acknowledgements:} We thank Prof. E. Rastelli  
645: for stimulating discussions and for pointing our attention to Refs.  
646: \cite{bramwell93}. P.S. and A.T. enjoyed very much the  
647: residence at the CRS-BEC of Trento, where this work was initiated.   
648: This work has been supported by MIUR through grant No. 2001028294  
649: and by the DOE. 
650:  
651: \appendix 
652:  
653: \section{Variational Estimates of the Coefficients $K$ and $U$} 
654:  
655: In this Appendix we derive a variational 
656: estimate of the coefficients $K$ and $U$ entering the 
657: Bose-Hubbard Hamiltonian (\ref{BHM}).  
658: The tunneling rate $K$ and 
659: the coefficient of the nonlinear term $U$ are given by  
660: \begin{equation} 
661: K = - \int d\vec{r} \, \Bigg[ \frac{\hbar^2}{2m}  
662: \vec{\nabla} \Phi_{\vec{i}} \cdot \vec{\nabla} \Phi_{{\vec{j}}} +  
663: \Phi_{\vec{i}}  V_{ext} \Phi_{\vec{j}} \Bigg] 
664: \label{K} 
665: \end{equation} 
666: and  
667: \begin{equation} 
668: U = g_0 \int d\vec{r} \, \Phi_{\vec{j}}^4, 
669: \label{U} 
670: \end{equation} 
671: where $g_0=4 \pi \hbar^2 a / m$ and $\Phi_{\vec{j}}(\vec{r})$ is  
672: the Wannier wavefunction localized at the $\vec{j}$-th well  
673: and normalized to $1$. 
674:  
675: For a $2D$ optical lattice in the $x-y$ plane one has 
676: $V_{ext}=V_{opt}+V_m$, where the optical potential is 
677: $V_{opt}=V_0 [ \sin^2{(kx)} + \sin^2{(ky)} ]$ and  
678: the magnetic potential is  
679: $V_m=\frac{m}{2} [ \omega_r^2 (x^2+y^2) + \omega_z^2 z^2 ]$. 
680: Typical values of the parameters relevant for experiments 
681: \cite{greiner01} are 
682: $\omega_r , \omega_z \sim 2 \pi \cdot 10-100 \, Hz$, $\lambda=852 \, nm $, 
683: for a total number of sites $\sim 3000$ and an average number of 
684: particles per site $\sim 170$. 
685: The condition  $\omega_z \gg \omega_r$ should be imposed in order 
686: to have a $2D$ system. To get a variational estimate  
687: for $K$ and $U$ we assume that the Wannier wavefunction localized 
688: at the $j$-th well is given by  
689: \begin{equation} 
690: \Phi_{\vec{j}}(\vec{r})=C e^{-(x-x_j)^2/2\sigma_x^2}  
691: e^{-(y-y_j)^2/2\sigma_y^2} e^{-(z-z_j)^2/2\sigma_z^2}.  
692: \label{variazionale_2D} 
693: \end{equation}  
694: In Eq.(\ref{variazionale_2D}) 
695: the parameters $\sigma_x$, $\sigma_y$, and $\sigma_z$  
696: have to be determined variationally determined, 
697: $C$ is a normalization constant given by  
698: $C=(\pi^{3/2} \sigma^2 \sigma_z)^{-1/2}$ and $(x_j,y_j,z_j)$ 
699: are the coordinates of the center of the $j$-th well. 
700: Due to the symmetry of the external potential  
701: one has to set $\sigma_x=\sigma_y \equiv \sigma$: furthermore, since  
702: $\tilde{\omega}_r \gg \omega_r,\omega_z$, one has $\sigma_z \gg \sigma$.  
703: To fix $\sigma$ and $\sigma_z$ one recalls that 
704: the Gross-Pitaevskii energy is \cite{pitaevskii03} 
705: \begin{equation} 
706: {\cal E}[\Psi]=\int d\vec{r} \, \Bigg[ \frac{\hbar^2}{2m}  
707: (\vec{\nabla} \Psi)^2 + V_{ext} \mid \Psi \mid^2 + 
708: \frac{g_0}{2} \mid \Psi \mid^4 \Bigg]. 
709: \label{gp_energy}
710: \end{equation} 
711: In the tight-binding approximation,  
712: $\Psi(\vec{r},t)=\sum_j \psi_j(t) \Phi_j(\vec{r})$,  
713: one finds that the energy contribution $E_i$ of the $i$-th well to the total 
714: energy (\ref{gp_energy}) is  
715: $E_i \approx  
716: \int d\vec{r} \, \big[ \mid \psi_i \mid^2 \frac{\hbar^2}{2m}  
717: (\vec{\nabla}\Phi_i)^2 + V_{ext} \mid \psi_i \mid^2 \Phi_i^2 + 
718: \frac{g_0}{2} \mid \psi_i \mid^4 \Phi_i^4 \big]$, and, 
719: since $\mid \psi_i \mid^2 \approx N_0$, $N_0$ being 
720: the average value of particle per site, one gets
721: \begin{equation} 
722: E_i \approx  \int d\vec{r} \, \Bigg[ N_0 \frac{\hbar^2}{2m}  
723: (\vec{\nabla}\Phi_i)^2 + N_0 V_{ext} \Phi_i^2 + 
724: \frac{g_0}{2} N_0^2 \Phi_i^4 \Bigg].
725: \label{energia_singola_buca} 
726: \end{equation} 
727: Substituting (\ref{variazionale_2D}) in (\ref{energia_singola_buca}),  
728: one gets the following approximate expression  
729: for the energy in the $i$-th well   
730: \begin{equation}
731: \frac{E_i}{N_0} \approx \frac{\hbar^2}{2 m}  
732: \Bigg(\frac{1}{2\sigma_z^2} + \frac{1}{\sigma^2} \Bigg)+ 
733: \frac{m}{2} \Bigg(\omega_r^2 \sigma^2+\omega_z^2 \frac{\sigma_z^2}{2} \Bigg)+
734: V_0 k^2 \sigma^2 + \frac{g_0 N_0}{2(2\pi)^{3/2}\sigma^2\sigma_z}.
735: \label{e_i}
736: \end{equation} 
737: The optimal values of $\sigma$ and $\sigma_z$ are determined requiring that 
738: $\partial E_i / \partial \sigma=0$ and 
739: $\partial E_i / \partial \sigma_z=0$. 
740: It should be stressed that being $\sigma$ 
741: the variational width in the directions $x$ and $y$, 
742: it should be less than $\lambda/2$, the distance 
743: between minima of the periodic potential. 
744: We notice also that for realistic values of the parameters, 
745: the variational width $\sigma$ in the $x$ and $y$ directions 
746: has a very weak dependence on the mean field term (and therefore 
747: to the number of particles in each site): in fact, 
748: putting $\sigma \sim 0.2 \mu m$  
749: and $\sigma_z \sim 5 \mu m$, one finds  
750: $\hbar^2/2m\sigma^2 \sim 10^{-23} erg 
751: \gg g_0 N_0 / 2(2\pi)^{3/2}\sigma^2\sigma_z  
752: \sim 10^{-29} erg$, one sees that the last term in Eq.(\ref{e_i}) 
753: contributes very little to the determination of $\sigma$. 
754: Since $V_0 k^2 \gg m \omega_r^2/2$, 
755: from Eq.(\ref{e_i}) one then finds that the dependence of $E_i$ on 
756: $\sigma$ is given by $E_i(\sigma) \approx N_0 
757: (\hbar^2/2m\sigma^2 +V_0 k^2 \sigma^2)$: from 
758: $\partial E_i / \partial \sigma=0$ one obtains 
759: $\sigma^4=\hbar^2/2m k^2 V_0$, which can be cast in the form 
760: $\sigma=\lambda \, \chi$, with
761: \begin{equation}
762: \chi=\frac{1}{2\pi s^{1/4}}
763: \label{chi}
764: \end{equation}
765: and $s=V_0/E_R$. 
766: 
767: We are now in position to have an estimate for $K$ and $U$: 
768: by using the ansatz (\ref{variazionale_2D}) in 
769: Eqs.(\ref{K})-(\ref{U}) we get
770: \begin{equation}
771: \frac{K}{E_R}=\frac{1}{4\pi^2 \chi^4} \Bigg( \frac{1}{16} - 
772: \chi^2 \Bigg) \, e^{-1/16 \chi^2}- s e^{-1/16 \chi^2}
773: \label{K_2D}
774: \end{equation}   
775: and
776: \begin{equation}
777: \frac{U}{E_R}=\frac{2 m g_0}{(2 \pi)^{7/2} \hbar^2 \sigma_z \chi^2}.
778: \label{U_2D}
779: \end{equation}  
780: 
781: The variational procedure used works well also for $1D$ and $3D$ 
782: optical lattices: in a $1D$ optical lattice in the $x$ direction one has 
783: $V_{ext}=V_{opt}+V_m$, where $V_{opt}=V_0 \sin^2{(kx)}$ and 
784: $V_m=\frac{m}{2} [ \omega_x^2 x^2 + \omega_\perp^2 (y^2+z^2) ]$. 
785: Typical values of the parameters relevant for experiments 
786: \cite{cataliotti01} are 
787: $\omega_x \approx 2 \pi \cdot 10 \, Hz$, 
788: $\omega_\perp \approx 2 \pi \cdot 100 \, Hz$, $\lambda=795 \, nm $, 
789: for a total number of sites $\sim 100$ and an average number of 
790: particles per site $\sim 1000$. 
791: The variational ansatz is still given by Eq.(\ref{variazionale_2D}), 
792: but now $\sigma_y=\sigma_z \equiv \sigma_\perp$ and 
793: $\sigma_x \equiv \sigma$ is the variational width 
794: in the laser direction. Proceeding as before,  
795: one obtains that $\sigma=\lambda \, \chi$, 
796: with $\chi$ still given by Eq.(\ref{chi}). 
797: The tunneling rate $K$ is given by  
798: \begin{equation}  
799: \frac{K}{E_R} = \frac{1}{4\pi^2 \chi^4} 
800: \Bigg(\frac{1}{16}-\frac{\chi^2}{2} \Bigg) e^{-1/16\chi^2} -
801: \frac{s}{2} \, (1+e^{-4\pi^2 \chi^2}) \, e^{-1/16\chi^2},
802: \label{K_1D}
803: \end{equation}
804: and the nonlinear coefficient $U$ reads $U/E_R=2 m \lambda g_0 / 
805: (2 \pi)^{7/2} \hbar^2 \sigma_\perp^2 \chi$ where $\chi$ is given 
806: by Eq.(\ref{chi}).
807: 
808: For a $3D$ optical lattice one has 
809: $V_{ext}=V_{opt}+V_m$, where $V_{opt}=V_0 
810: [ \sin^2{(kx)} + \sin^2{(ky)} + \sin^2{(kz)}]$ and 
811: $V_m=\frac{m}{2} \omega (x^2+y^2+z^2)$. 
812: Typical experimental values from \cite{greiner02} are 
813: $\omega \approx 2 \pi \cdot 50\, Hz$, 
814: $\lambda=852 \, nm $, 
815: for a total number of sites $\sim 10^5$ and an average number of 
816: particles per site $\sim 1$. 
817: The variational ansatz is still given by Eq.(\ref{variazionale_2D}), 
818: but now $\sigma_x=\sigma_y=\sigma_z \equiv \sigma$. 
819: One obtains that $\sigma=\lambda \, \chi$, 
820: with $\chi$ given as before by Eq.(\ref{chi}). 
821: The tunneling rate $K$ is given by  
822: \begin{equation}  
823: \frac{K}{E_R} = \frac{1}{4\pi^2 \chi^4} 
824: \Bigg(\frac{1}{16}-\frac{3 \chi^2}{2} \Bigg) e^{-1/16\chi^2} -
825: s \, \Bigg(\frac{3}{2}-\frac{1}{2}e^{-4\pi^2 \chi^2}\Bigg) \, e^{-1/16\chi^2},
826: \label{K_3D}
827: \end{equation}
828: and the nonlinear coefficient $U$ reads $U/E_R=2 m g_0 / 
829: (2 \pi)^{7/2} \hbar^2 \lambda \chi^3$.
830: 
831: We observe that the tunneling rates $K$ for $1D$, $2D$ and $3D$ 
832: optical lattices, given respectively by 
833: Eqs.(\ref{K_1D}), (\ref{K_2D}) and (\ref{K_3D}),  
834: can be compactly written as  
835: \begin{equation}  
836: \frac{K}{E_R} = \frac{1}{4\pi^2 \chi^4} 
837: \Bigg(\frac{1}{16}-\frac{(j+1) \chi^2}{2} \Bigg) e^{-1/16\chi^2} -
838: s \, \Bigg( \frac{j+1}{2}-\frac{j-1}{2}e^{-4\pi^2 \chi^2} 
839: \Bigg) \, e^{-1/16\chi^2},
840: \label{K_general}
841: \end{equation}
842: where $j=D-1$.
843: 
844: \section{Fourier Transform of a Vortex on a Lattice}
845: 
846: In this Appendix we report the analytical 
847: computation of the lattice Fourier transform $\tilde{\psi}_{\vec{k}}$ 
848: of single vortex on a lattice described by 
849: $\psi_{\vec{j}}=\exp{(i \theta_{\vec{j}})}$; 
850: namely, one should compute
851: \begin{equation}
852: \tilde{\psi}_{\vec{k}}=\frac{1}{N}  
853: \sum_{\vec{j}} \psi_{\vec{j}} \, \, \, 
854: e^{-i \vec{k} \cdot {\vec{j}}}=\frac{1}{N}  
855: \sum_{\vec{j}} e^{-i \vec{k} \cdot {\vec{j}}+i\theta_{\vec{j}}}.
856: \label{FT_app}
857: \end{equation}
858: In Eq.(\ref{FT_app}), $\vec{j}=(j_x,j_y)$ denotes the sites 
859: of a square lattice (having $N$ sites) 
860: and $\vec{k}=(k_x,k_y)$ the (quasi)momentum, 
861: with $k_x$ and $k_y$ valued between $-\pi$ and $\pi$. 
862: The relationship between the lattice Fourier transform 
863: (\ref{FT_app}) and the momentum distribution $\tilde{\psi}(\vec{p})$ 
864: in real space is provided by Eq.(\ref{relation}). 
865: 
866: To simplify matters, in this Appendix we set the origin $O$ 
867: of the coordinates in the center 
868: of the central plaquette: setting to $1$ the lattice length, the lattice 
869: sites are labeled by $\vec{j}=(j_1/2,j_2/2)$ 
870: with $j_1$ and $j_2$ positive and negative {\em odd} integer numbers. 
871: 
872: The phase distribution of a vortex is 
873: depicted in Fig.2(A) and is such that $\theta_{\vec{j}}$ equals 
874: the polar angle $\phi$ with respect to the origin $O$ (e.g., 
875: at the site $(1/2,1/2)$ $\theta_{\vec{j}}$ equals $\pi / 4$, 
876: at the site $(3/2,1/2)$ 
877: is $\arctan{(1/3)}$, and so on; an antivortex would be characterized 
878: by $-\phi$). For future convenience, we shall label the site 
879: $(j_1+1/2,j_2+1/2)$ with the index $\ell=(\mid j_1 \mid + \mid j_2 \mid)/2$: 
880: as a consequence, the four sites $(1/2,1/2)$, 
881: $(1/2,-1/2)$, $(-1/2,1/2)$, and $(1/2,1/2)$ of the central plaquette 
882: will be associated to $\ell=1$; the eight sites 
883: $(3/2,1/2)$, $(1/2,3/2)$, $\cdots$, $(-3/2,1/2)$ 
884: will be associated to $\ell=2$. 
885: In other words, the sites with index $\ell$ have chemical 
886: distance $\ell-1$ from the four sites of the central plaquette. 
887: Of course, the number 
888: of sites with index $\ell$ is $4 \ell$. If, for simplicity, 
889: one considers only sites with $\ell$ going from $1$ to a maximum value 
890: ${\cal L}$, then the total number of sites is 
891: \begin{equation}
892: N({\cal L})=\sum_{\ell=1}^{\cal L} 4 \ell=2 {\cal L} ({\cal L}+1).
893: \label{totale}
894: \end{equation} 
895: 
896: To evaluate the lattice Fourier transform (\ref{FT_app}), 
897: one may conveniently separate the sum 
898: $\sum_{\vec{j}} e^{-i \vec{k} \cdot {\vec{j}}+i\theta_{\vec{j}}}$ in 
899: $\cal L$ sums over sites having the same index $\ell$; for instance, 
900: for the four sites $(1/2,1/2)$, 
901: $(1/2,-1/2)$, $(-1/2,1/2)$, and $(1/2,1/2)$ of the central plaquette, 
902: one gets
903: \begin{equation}
904: \sum_{\vec{j}(1)} 
905: e^{-i \vec{k} \cdot {\vec{j}}+i\theta_{\vec{j}}} = 
906: 4 \Bigg[ \sin{\phi^{(1)}_1} \, \cos{\frac{k_x}{2}} \, \sin{\frac{k_y}{2}} 
907: - i \cos{\phi^{(1)}_1} \, \sin{\frac{k_x}{2}} \, \cos{\frac{k_y}{2}}   \Bigg]
908: \label{sum_elle_1}
909: \end{equation}
910: where $\phi^{(1)}_1=\pi/4$ and 
911: the sum is restricted only to the sites having $\ell=1$. 
912: A straightforward generalization of Eq.(\ref{sum_elle_1}) shows that for 
913: the sum on the 
914: $4 \ell$ sites having a fixed $\ell$ one has
915: $$
916: \tilde{\psi}_{\vec{k}}^{(\ell)} \equiv \sum_{\vec{j}(\ell)} 
917: e^{-i \vec{k} \cdot {\vec{j}}+i\theta_{\vec{j}}} = 
918: 4 \sum_{m=1}^{\ell} 
919: \Bigg[ \sin{\phi_m^{(\ell)}} 
920: \cos{ \frac{(2n-2m+1)k_x}{2}} \, \sin{ \frac{(2m-1)k_y}{2}} + 
921: $$
922: \begin{equation}
923: - i \cos{\phi_m^{(\ell)}} \, 
924: \sin{ \frac{(2n-2m+1)k_x}{2}} \, \cos{ \frac{(2m-1)k_y}{2}} \Bigg]
925: \label{sum_elle}
926: \end{equation}
927: where 
928: \begin{equation}
929: \phi_m^{(\ell)}=\arctan{\Bigg[ \frac{2m-1}{2n-2m+1} \Bigg]}
930: \label{theta_m_n}
931: \end{equation}
932: is an angle taking values between $0$ and $\pi / 2$. 
933: 
934: By using Eq.(\ref{sum_elle}), the lattice Fourier transform (\ref{FT_app}) 
935: may be compactly written as
936: \begin{equation}
937: \tilde{\psi}_{\vec{k}}=\frac{1}{N({\cal L})}  
938: \sum_{\ell=1}^{\cal L} \tilde{\psi}_{\vec{k}}^{(\ell)}.
939: \label{FT_an}
940: \end{equation}
941:      
942: If in Eq.(\ref{FT_an}) one puts $k_y=0$, one gets 
943: \begin{equation}
944: \tilde{\psi}_{\vec{k}}=\frac{-4i}{N({\cal L})}  
945: \sum_{\ell=1}^{\cal L} \sum_{m=1}^{\ell} \cos{\phi^{(\ell)}_m} 
946: \sin{\frac{(2n-2m+1)k_x}{2}};
947: \label{FT_an_simpl1}
948: \end{equation}
949: by introducing the index $\tilde{m}=\ell-m$ and rearranging conveniently 
950: the partial sums in Eq.(\ref{FT_an_simpl1}) one obtains 
951: \begin{equation}
952: \tilde{\psi}_{\vec{k}}=
953: \sum_{\tilde{m}=0}^{{\cal L}-1} a_{\tilde{m}} 
954: \sin{\frac{(2\tilde{m}+1)k_x}{2}}
955: \label{FT_an_simpl2}
956: \end{equation}
957: with
958: $$
959: a_{\tilde{m}}=\frac{-4i}{N({\cal L})} \sum_{n=\tilde{m}+1}^{\cal L} 
960: \frac{2 \tilde{m}+1}{\sqrt{(2 \tilde{m}+1)^2+(2n-2 \tilde{m}-1)^2}}.
961: $$
962: Eq.(\ref{FT_an_simpl2}) clearly shows that for $k_x=0$ one has 
963: $\tilde{\psi}_0=0$: of course, with a single vortex, the magnetization 
964: (\ref{M}) is zero; furthermore, in the thermodynamic limit ${\cal L} \to 
965: \infty$, $a_{\tilde{m}} \to 0$, in agreement the Mermin-Wagner theorem. We
966: observe Eq.(\ref{FT_an}) may be easily applied to the analysis 
967: of the lattice Fourier transform of a vortex-antivortex pair.   
968: 
969: \begin{thebibliography}{10} 
970: 
971: \bibitem{anderson98} B. P. Anderson and M. A. Kasevich, 
972: Science {\bf 282}, 1686 (1998). 
973:  
974: \bibitem{cataliotti01} F. S. Cataliotti, S. Burger, 
975: C. Fort, P. Maddaloni, F. Minardi, A. Trombettoni, 
976: A. Smerzi, and M. Inguscio, Science {\bf 293}, 843 (2001).  
977:  
978: \bibitem{morsch01} O. Morsch, J. H. M\"uller, 
979: M. Cristiani, D. Ciampini, and E. Arimondo   
980: Phys. Rev. Lett. {\bf 87}, 140402 (2001).  
981:  
982: \bibitem{hensinger01} W. K. Hensinger {\em et al.},  
983: Nature {\bf 412}, 52 (2001).  
984:  
985: \bibitem{eiermann03} 
986: B. Eiermann, P. Treutlein, T. Anker, 
987: M. Albiez, M. Taglieber, K.-P. Marzlin, and M. K. Oberthaler 
988: Phys. Rev. Lett. {\bf 91}, 060402 (2003). 
989:  
990: \bibitem{greiner01}  
991: M. Greiner, I. Bloch, O. Mandel, T. W. H\"ansch, and 
992: T. Esslinger,  Phys. Rev. Lett. {\bf 87}, 160405 (2001). 
993:  
994: \bibitem{greiner02} M. Greiner, O. Mandel, T. Esslinger, 
995: T. W. H\"ansch, and I. Bloch, 
996: Nature {\bf 415}, 39 (2002). 
997:  
998: \bibitem{beasley79} M. R. Beasley, J. E. Mooij, and T. P. Orlando,  
999: Phys. Rev. Lett. {\bf 42}, 1165 (1979). 
1000:  
1001: \bibitem{resnick81} D. J. Resnick {\em et al.},  
1002: Phys. Rev. Lett. {\bf 47}, 1542 (1981). 
1003:  
1004: \bibitem{nelson83} D. R. Nelson, in {\em Phase Transitions and  
1005: Critical Phenomena}, vol. 7, eds. C. Domb and J. L. Lebowitz  
1006: (New York, Academic Press, 1983) and references therein. 
1007:  
1008: \bibitem{minnaghen87} P. Minnaghen, Rev. Mod. Phys. {\bf 59},  
1009: 1001 (1987). 
1010:  
1011: \bibitem{kadanoff00} L. P. Kadanoff, {\em Statistical Physics: Statics,  
1012: Dynamics and Renormalization} (Singapore, World Scientific, 2000),  
1013: Chapt.s 16-17 and reprints therein. 
1014:  
1015: %\bibitem{martinoli87} P. Martinoli {\em et al.},  
1016: %J. Appl. Phys. {\bf 26}, 1999 (1987). 
1017:  
1018: \bibitem{bishop78}  D. J. Bishop and J. D. Reppy,  
1019: Phys. Rev. Lett. {\bf 40}, 1727 (1978). 
1020: 
1021: \bibitem{berezinskii73} V. L. Berezinskii and A. Ya. Blank, Sov. Phys. JETP  
1022: {\bf 37}, 369 (1973). 
1023: 
1024: \bibitem{bramwell93} S. T. Bramwell and P. C. W. Holdsworth,  
1025: J. Phys.: Condensed Matter {\bf 5}, L53 (1993);  
1026: Phys. Rev. B {\bf 49}, 8811 (1994). 
1027:  
1028: \bibitem{pedri03} P. Pedri and L. Santos,  
1029: Phys. Rev. Lett. {\bf 91}, 110401 (2003). 
1030:  
1031: \bibitem{jaksch98} D. Jaksch, C. Bruder, J. I. Cirac, 
1032: C. W. Gardiner, and P. Zoller,
1033: Phys. Rev. Lett. {\bf 81}, 3108 (1998). 
1034:  
1035: \bibitem{fisher89} M. P. A. Fisher, P. B. Weichman, G. Grinstein, 
1036: and D. S. Fisher, Phys. Rev. B {\bf 40}, 546 (1989).  
1037:  
1038: %\bibitem{anglin01} J.R. Anglin{\em et al.},  
1039: %Phys. Rev. A {\bf 64}, 063605 (2001). 
1040:  
1041: %\bibitem{batrouni02} G.G. Batrouni {\em et al.},  
1042: %Phys. Rev. Lett. {\bf 89}, 117203 (2002). 
1043:  
1044: %\bibitem{dalfovo99} F. Dalfovo {\em et al.},  
1045: %Rev. Mod. Phys. {\bf 71}, 463 (1999). 
1046:  
1047: \bibitem{simanek94} E. Sim\`anek, {\em Inhomogeneous Superconductors}, 
1048: Oxford University Press, New York, 1994. 
1049:  
1050: \bibitem{fazio01} R. Fazio and H. van der Zant,  
1051: Phys. Rep. {\bf 355}, 235 (2001). 
1052: 
1053: \bibitem{fisher88} M. P. A. Fisher and G. Grinstein, 
1054: Phys. Rev. Lett. {\bf 60}, 208 (1988). 
1055:  
1056: \bibitem{kampf93} A. P. Kampf and G. T. Zimanyi,  
1057: Phys. Rev. B {\bf 47}, 279 (1993). 
1058:  
1059: \bibitem{tsuchiya03} S. Tsuchiya and A. Griffin, cond-mat/0311321.  
1060: 
1061: \bibitem{gupta88} R. Gupta, J. DeLapp, G. G. Batrouni, 
1062: G. C. Fox, C. F. Baillie, and J. Apostolakis, 
1063: Phys. Rev. Lett. {\bf 61}, 1996 (1988).
1064:  
1065: %\bibitem{jose84} J.V. Jos\'e,  
1066: %Phys. Rev. B {\bf 29}, 2836 (1984). 
1067: 
1068: \bibitem{smerzi04} A. Smerzi, P. Sodano, and A. Trombettoni,
1069: J. Phys. B {\bf 37}, S265 (2004).
1070:  
1071: \bibitem{rojas96} C. Rojas and J. V. Jos\'e,  
1072: Phys. Rev. B {\bf 54}, 12361 (1996). 
1073: 
1074: \bibitem{pitaevskii03} L. P. Pitaevskii and S. Stringari, 
1075: {\it Bose-Einstein Condensation} 
1076: (Oxford University Press, Oxford, 2003).
1077:  
1078: %\bibitem{sheshadri93} K. Sheshadri {\em et al.}, Europhys. Lett.  
1079: %{\bf 22}, 257 (1993). 
1080:  
1081: %\bibitem{freericks95} J.K. Freericks and H. Monien, Europhys. Lett.  
1082: %{\bf 26}, 545 (1993). 
1083:  
1084: %\bibitem{oosten02} D. van Oosten {\em et al.},  
1085: %Phys. Rev. A {\bf 63}, 053601 (2002).  
1086:  
1087: %\bibitem{pedri01} P. Pedri {\em et al.},  
1088: %Phys. Rev. Lett. {\bf 87}, 220401 (2001). 
1089:  
1090: \bibitem{gorlitz01} A. G\"orlitz {\em et al.},  
1091: Phys. Rev. Lett. {\bf 87}, 130402 (2001). 
1092:  
1093: \bibitem{hammes03} M. Hammes, D. Rychtarik, 
1094: B. Engeser, H.-C. N\'agerl, and R. Grimm,  
1095: Phys. Rev. Lett. {\bf 90}, 173001 (2001). 
1096:  
1097: \bibitem{petrov00} D. S. Petrov, M. Holzmann, and 
1098: G. V. Shlyapnikov, Phys. Rev. Lett. {\bf 84}, 2551 (2000). 
1099:  
1100: %\bibitem{popov83} V. N. Popov, {\em Functional Integrals in Quantum  
1101: %Field Theory and Statistical Physics} (Reidel, Dordrecht, 1983).  
1102:  
1103: \bibitem{prokofev01} N. Prokof\'ev, O. Ruebenacker, and B. Svistunov, 
1104: Phys. Rev. Lett. {\bf 87}, 270402 (2001). 
1105: 
1106: \bibitem{tilley90} D. R Tilley and J. Tilley,  
1107: {\em Superfluidity and Superconductivity} 
1108: (Bristol, Adam Hilger, 1990). 
1109:  
1110: \bibitem{gottlob93} A. P. Gottlob and M. Hasenbusch,  
1111: Physica A {\bf 201}, 593 (1993). 
1112:  
1113: \bibitem{campostrini01} M. Campostrini, M. Hasenbusch,  
1114: A. Pelissetto, P. Rossi, and E. Vicari, Phys. Rev. {\bf B 63},  
1115: 214503 (2003). 
1116:  
1117:  
1118: \end{thebibliography} 
1119:  
1120: \begin{figure} 
1121: \centerline{\psfig{figure=fig1.ps,width=60mm,angle=270}} 
1122: \caption{$J/U$ as a function of $V_0/E_R$ for $1D$, $2D$ and  
1123: $3D$ optical lattices ($z$ is the number of nearest neighbours,  
1124: which is respectively $2$, $4$, and $6$).  
1125: The experimental values are taken, respectively, from 
1126: \protect\cite{cataliotti01}, \protect\cite{greiner01},  
1127: and \protect\cite{greiner02}, with an average number of $^{87} Rb$ atoms  
1128: $N_0=1000$ ($1D$), $170$ ($2D$) and $1$ ($3D$). 
1129: } 
1130: \label{fig1} 
1131: \end{figure} 
1132:  
1133: \begin{figure} 
1134: \centerline{\psfig{figure=fig2.ps,width=60mm,angle=0}} 
1135: \caption{Lattice Fourier transform $\mid \tilde{\psi}_{\vec{k}} \mid^2$ 
1136: (for $k_y=0$) with:  
1137: (A) a single lattice vortex; (B) a lattice vortex-antivortex pair;  
1138: (C) $T=0.5 J/k_B$; (D) $T = 1.1 J/k_B$.  
1139: Figures (C) and (D) are Monte Carlo snapshots after reaching equilibrium.  
1140: The lattice is $20 \times 20$.}  
1141: \label{fig2} 
1142: \end{figure} 
1143:  
1144: \begin{figure} 
1145: \centerline{\psfig{figure=fig3.ps,width=60mm,angle=270}} 
1146: \caption{Intensity of the central peak of the momentum distribution  
1147: (normalized to the value at $T=0$) as a function of the reduced temperature  
1148: $t=k_B T/J$ in a 2D lattice. Empty circles: Monte Carlo simulations;  
1149: solid line: low-temperature spin wave prediction;  
1150: dotted line: fit near $T_C \approx 1.07  
1151: J/k_B$ as in \protect\cite{bramwell93} (in this figure $L=40$).}  
1152: \label{fig3} 
1153: \end{figure} 
1154: 
1155: \begin{figure} 
1156: \centerline{\psfig{figure=fig4.ps,width=60mm,angle=270}} 
1157: \caption{Monte Carlo results for the intensity 
1158: of the central peak of the momentum distribution  
1159: (normalized to the value at $T=0$) as a function of the reduced 
1160: temperature  $t=k_B T/J$ in a 3D lattice (with $20^3$ sites).}  
1161: \label{fig4} 
1162: \end{figure}  
1163: %\end{multicols} 
1164:  
1165: \end{document} 
1166: