1: \documentclass[prl,twocolumn,showpacs,amsmath,amssymb]
2: {revtex4}
3:
4: \usepackage[dvips]{color}
5: \usepackage{graphicx}
6:
7: \begin{document}
8:
9: \title{Emergence of quasi-condensates of hard-core bosons
10: at finite momentum}
11:
12: \author{Marcos Rigol}
13: \affiliation{Institut f\"ur Theoretische Physik III, Universit\"at
14: Stuttgart, Pfaffenwaldring 57, D-70550 Stuttgart, Germany.}
15:
16: \author{Alejandro Muramatsu}
17: \affiliation{Institut f\"ur Theoretische Physik III, Universit\"at
18: Stuttgart, Pfaffenwaldring 57, D-70550 Stuttgart, Germany.}
19:
20: \begin{abstract}
21: An exact treatment of the non-equilibrium dynamics of hard-core bosons
22: on one-dimensional lattices shows that, starting from a pure Fock state,
23: quasi-long-range correlations develop dynamically, and that they lead to
24: the formation of quasi-condensates at finite momenta.
25: Scaling relations characterizing the quasi-condensate and the dynamics
26: of its formation are obtained. The relevance of our findings for
27: atom lasers with full control of the wave-length by means of a lattice
28: is discussed.
29: \end{abstract}
30:
31: \pacs{03.75.Kk, 03.75.Lm, 03.75.Pp, 05.30.Jp}
32: \maketitle
33:
34: The study of the non-equilibrium dynamics of quantum gases has been
35: shown to be a very useful tool for the understanding of their properties
36: \cite{dalfovo99}. Since the achievement of Bose-Einstein condensation (BEC)
37: \cite{1995}, where the anisotropic expansion of the gas revealed the
38: importance of the inter-particle interaction, many experiments studying the
39: dynamics of such systems have been developed. Recently, the possibility
40: of realizing one-dimensional (1-D) quantum gases has
41: attracted a lot of attention. They can be obtained experimentally in very
42: elongated traps \cite{1D}, or using optical lattices \cite{greiner01}.
43: In these quasi-1D systems it is possible to go from the weakly interacting
44: regime to an impenetrable gas of bosons, i.e.\ hard-core bosons (HCB)
45: \cite{HCB}. Theoretical studies have analyzed the density evolution
46: between these two extreme regimes \cite{santos02}, showing that in general
47: it does not follow a self-similar solution. In addition, in the HCB limit
48: the Fermi-Bose mapping \cite{girardeau60} was generalized to the
49: time-dependent case, where the density dynamics revealed dark soliton
50: structures, breakdown of the time-dependent mean-field theory, and interference
51: patterns of the thermal gas on a ring \cite{girardeau02}.
52:
53: In 1D systems in equilibrium it was shown that
54: quasi-condensates of HCB develop in the homogeneous \cite{homog},
55: harmonically trapped \cite{trap}, and in general cases in
56: the presence of a lattice \cite{rigol04_1}. This is because
57: the ground state of these systems exhibits off-diagonal
58: quasi-long-range order determined by a power-law decay
59: of the one-particle density matrix (OPDM)
60: $\rho_{ij}\sim |x_i-x_j|^{-1/2}$ \cite{homog,rigol04_1},
61: i.e. there is no BEC in the thermodynamic limit \cite{yang62}.
62: The occupation of the lowest natural orbital (NO)
63: (the highest occupied one) is then proportional to
64: $\sqrt{N_b}$ ($N_b$ being the number of HCB) \cite{homog,trap,rigol04_1}.
65: The NO ($\phi^\eta_i$) are effective single particle states
66: defined as the eigenstates of the OPDM \cite{penrose56},
67: i.e.\ they are obtained by the equation
68: $\sum_j \rho_{ij}\phi^\eta_j= \lambda_{\eta}\phi^\eta_i$,
69: with $\lambda_{\eta}$ being their occupations.
70:
71: In the present work we study within an {\it exact} approach the
72: non-equilibrium dynamics of HCB in 1D configurations with
73: an underlying lattice. The presence of the lattice enhances
74: correlations between particles, and its experimental realization
75: (the so-called optical lattice) allowed the study of the Mott
76: superfluid-insulator transition for soft-core bosons \cite{greiner02_1}.
77: In contrast to the continuous case, in a lattice it is possible to create
78: pure Fock states of HCB (a HCB per lattice site) where there is no
79: coherence in the system. We show that quasi-long-range correlations
80: develop in the equal-time-one-particle density matrix (ETOPDM) when
81: such states are allowed to evolve freely, and that they lead to the
82: formation of quasi-condensates of HCB at finite momentum. In addition
83: we obtain an universal power law describing the population of
84: the quasi-condensate as a function of time, independent of the initial
85: number of particles in the Fock-state. Finally,
86: we analyze how such systems can be used to create atom lasers with a
87: wave-length that can be controlled through the lattice parameter. Our
88: analysis is based on a generalization of the method used to study
89: the ground-state properties of 1D HCB on a lattice \cite{rigol04_1}.
90:
91: The HCB Hamiltonian on a lattice can be written as
92: \begin{equation}
93: \label{HamHCB} H_{HCB} = -t \sum_{i} \left( b^\dagger_{i} b^{}_{i+1}
94: + h.c. \right) + V_\alpha \sum_{i} x_i^\alpha\ n_{i },
95: \end{equation}
96: with the addition of the on-site constraints for the creation
97: ($b^{\dagger}_{i}$) and annihilation ($b_{i}$) operators:
98: $b^{\dagger 2}_{i}= b^2_{i}=0$, and $\left\lbrace
99: b^{}_{i},b^{\dagger}_{i}\right\rbrace =1$. The hopping parameter is denoted
100: by $t$ and $n_{i }= b^{\dagger}_{i}b^{}_{i}$ is the particle number operator.
101: The last term in Eq.\ (\ref{HamHCB}) describes an arbitrary confining
102: potential.
103:
104: The Jordan-Wigner transformation \cite{jordan28},
105: \begin{equation}
106: \label{JordanWigner} b^{\dag}_i=f^{\dag}_i
107: \prod^{i-1}_{\beta=1}e^{-i\pi f^{\dag}_{\beta}f^{}_{\beta}},\ \
108: b_i=\prod^{i-1}_{\beta=1} e^{i\pi f^{\dag}_{\beta}f^{}_{\beta}}
109: f_i \ ,
110: \end{equation}
111: is used to map the HCB Hamiltonian into the one of non-interacting
112: fermions
113: \begin{equation}
114: H_F =-t \sum_{i} \left( f^\dagger_{i}
115: f^{}_{i+1} + h.c. \right)+ V_\alpha \sum_{i} x_i^\alpha \
116: n^f_{i }, \label{eq5}
117: \end{equation}
118: where $f^\dagger_{i}$ and $f^{}_{i}$ are the creation
119: and annihilation operators for spinless fermions, and
120: $n^f_{i }=f^\dagger_{i}f^{}_{i}$.
121:
122: This mapping allows to express the equal-time Green's function
123: for the HCB in a non-equilibrium system as
124: \begin{eqnarray}
125: \label{green1} G_{ij}(\tau)&=&\langle \Psi_{HCB}(\tau)|
126: b^{}_{i}b^\dagger_{j}|\Psi_{HCB}(\tau)\rangle \\
127: &=&\langle \Psi_{F}(\tau)| \prod^{i-1}_{\beta=1}
128: e^{i\pi f^{\dag}_{\beta}f^{}_{\beta}} f^{}_i f^{\dag}_j
129: \prod^{j-1}_{\gamma=1} e^{-i\pi f^{\dag}_{\gamma}f^{}_{\gamma}}
130: |\Psi_{F}(\tau)\rangle, \nonumber
131: \end{eqnarray}
132: where $\tau$ is the real time variable, $|\Psi^{G}_{HCB}(\tau)\rangle$
133: is the time evolving wave-function for the HCB and
134: $|\Psi^{G}_{F}(\tau)\rangle$ is the corresponding one
135: for the non-interacting fermions.
136:
137: In what follows, we study the time evolution of initial Fock states of HCB
138: once they are allowed to evolve freely. Experimentally such states can be
139: created by a strong confining potential $V_\alpha(N_ba/2)^{\alpha}\gg t$
140: ($a$ is the lattice constant), which avoid vacancies. Since the
141: equivalent fermionic system is a non-interacting one, the time evolution of
142: an initial wave-function ($|\Psi^I_{F}\rangle$) can be easily calculated
143: \begin{equation}
144: \label{time}
145: |\Psi_{F}(\tau)\rangle=e^{-iH_F\tau/\hbar}|\Psi^I_{F}\rangle
146: = \prod^{N_f}_{\delta=1}\ \sum^N_{\sigma=1} \ P_{\sigma
147: \delta}(\tau)f^{\dag}_{\sigma}\ |0 \rangle,
148: \end{equation}
149: which is a product of single particle states, with $N_f$ being
150: the number of fermions ($N_f=N_b$), $N$ the number of lattice sites and
151: ${\bf P(\tau)}$ the matrix of components of $|\Psi_{F}(\tau)\rangle$.
152: From here on the method is identical to the one used in Ref.\
153: \cite{rigol04_1}, the action of
154: $\prod^{j-1}_{\gamma=1} e^{-i\pi f^{\dag}_{\gamma}f_{\gamma}}$
155: on the right in Eq.\ (\ref{green1}) generates only a
156: change of sign on the elements $P_{\sigma \delta}(\tau)$
157: for $\sigma \leq j-1$, and the further creation of a particle at site
158: $j$ implies the addition of one column to ${\bf P(\tau)}$ with the
159: element $P_{jN_f+1}(\tau)=1$ and all the others equal to zero
160: (the same applies to the action of $\prod^{i-1}_{\beta=1} e^{i\pi
161: f^{\dag}_{\beta}f_{\beta}} f_i$ on the left of Eq.\ (\ref{green1})).
162: Then the HCB Green's function can be calculated exactly as \cite{rigol04_1}
163: \begin{eqnarray}
164: \label{determ}
165: G_{ij}(\tau)
166: &=&\langle 0 | \prod^{N_f+1}_{\delta=1}\ \sum^N_{\beta=1} \
167: P'^{A}_{\beta \delta}(\tau)f_{\beta}
168: \prod^{N_f+1}_{\sigma=1}\ \sum^N_{\gamma=1} \ P'^{B}_{\gamma
169: \sigma}(\tau)f^{\dag}_{\gamma}\ |0 \rangle \nonumber \\
170: &=&\det\left[ \left( {\bf P}^{'A}(\tau)
171: \right)^{\dag}{\bf P}^{'B}(\tau)\right],
172: \end{eqnarray}
173: where ${\bf P'}^{A}(\tau)$ and ${\bf P'}^{B}(\tau)$ are the new matrices
174: obtained from ${\bf P}(\tau)$ when the required signs are changed and
175: the new columns added. The evaluation of $G_{ij}(\tau)$ is done
176: numerically, and the ETOPDM ($\rho_{ij}(\tau)$) is determined by the expression
177: $\rho_{ij}(\tau)=\left\langle b^\dagger_{i}b_{j}\right\rangle_\tau
178: =G_{ji}(\tau)+\delta_{ij}\left(1-2 G_{ii}(\tau)\right)$.
179:
180: \begin{figure}[h]
181: \includegraphics[width=0.48\textwidth,height=0.225\textwidth]
182: {Perfil}
183: \caption{(color online). Evolution of density (a) and momentum (b)
184: profiles of 300 HCB in 1000 lattice sites. The times are
185: $\tau=0$ (\textcolor{magenta}{$\triangle$}),
186: $50\hbar/t$ (\textcolor{red}{$\bigcirc$}),
187: $100\hbar/t$ (\textcolor{green}{$\times$}), and
188: $150\hbar/t$ (\textcolor{blue}{$\nabla$}). Positions (a)
189: and momenta (b) are normalized by the lattice constant $a$.}
190: \label{Perfil}
191: \end{figure}
192: Figure \ref{Perfil} shows density profiles (a), and their corresponding
193: momentum distribution functions (MDF) (b) for the time evolution of an
194: initial Fock state. Initially, the MDF
195: is flat as corresponds to a pure Fock state, and during the evolution
196: of the system sharp peaks appear at $k=\pm \pi/2a$.
197: Notice that although in the equivalent fermionic system
198: the density profiles are equal to the ones of the HCB,
199: the MDF remains flat since the fermions do not interact, making evident the
200: non-trivial differences in the off-diagonal correlations between both systems.
201:
202: Since the peaks in the MDF may correspond to quasi-condensates
203: at finite momenta, we diagonalize the ETOPDM to study the NO.
204: In Fig.\ \ref{NO}(a) we show the lowest NO
205: ($|\phi^0|$ since $\phi^0$ is complex) corresponding to the results in
206: Fig.\ \ref{Perfil} for $\tau>0$ (at $\tau=0$ the NO are delta functions
207: at the occupied sites). They are two-fold degenerate, corresponding to
208: right- (for $x>0$) and left- (for $x<0$) moving solutions.
209: It can be seen that some time after the $n=1$
210: plateau disappears from the system, the NO lobes almost do not change
211: their form and size (see inset in Fig.\ \ref{NO}(a) for the
212: right lobe at three different times). Furthermore, they move with constant
213: velocities $v_{NO}=\pm 2at/\hbar$ (see inset in Fig.\ \ref{NO}(b) for the
214: positive one). These are in fact, the group velocities
215: $v=1/\hbar\ \partial \epsilon_k/\partial k$
216: for a dispersion $\epsilon_k=-2t\cos ka$ (the one of HCB on a lattice)
217: at $k=\pm \pi/2a$. This is further confirmed by considering
218: the Fourier transforms of the lowest NO Fig.\ \ref{NO}(b), which show
219: sharp peaks centered at quasi-momenta $k=\pm \pi/2a$.
220: The peak at $k=+\pi/2a$ appears due to the
221: Fourier transform of the right lobe in Fig.\ \ref{NO}
222: (a) (i.e.\ $\phi^0_i \approx |\phi^0_i| e^{i\pi x_i/2a}$ for $x_i>0$),
223: and the one at $k=-\pi/2a$ due to the fourier transform
224: of the left lobe (i.e.\ $\phi^0_i \approx |\phi^0_i| e^{-i\pi x_i/2a}$
225: for $x_i<0$). Studying the Fourier transform of the other NO we have
226: seen that they have very small or zero weight at $k=\pm \pi/2a$ so that
227: we can conclude that the peaks at $n_{k=\pm \pi/2a}$ are reflecting
228: the large occupation of the lowest NO.
229: \begin{figure}[h]
230: \includegraphics[width=0.48\textwidth,height=0.235\textwidth]
231: {NO}
232: \caption{(color online). Lowest NO evolution (a) and its fourier
233: transform (b). The times are $\tau=50\hbar/t$ (\textcolor{red}{$\bigcirc$}),
234: $100\hbar/t$ (\textcolor{green}{$\times$}), and
235: $150\hbar/t$ (\textcolor{blue}{$\nabla$}). The insets show: (a)
236: the superposed right lobe of the lowest NO at
237: $\tau=100\hbar/t$ (\textcolor{green}{$\times$}),
238: $125\hbar/t$ (\textcolor{red}{$\bigcirc$}),
239: and $150\hbar/t$ (\textcolor{blue}{$\nabla$}); (b) the evolution of
240: the lowest NO right lobe position, the line has a slope 2.}
241: \label{NO}
242: \end{figure}
243:
244: Whether the large occupation of the lowest NO corresponds to a
245: quasi-condensate, can be determined studying the ETOPDM. Fig.\ \ref{ETOPDM}(a)
246: shows the results for $|\rho_{ij}|$ (with $i$ taken at the beginning
247: of the lowest NO left lobe's and $j>i$) at the same times of
248: Figs.\ \ref{Perfil}, \ref{NO}. It can be seen that off-diagonal
249: quasi-long-range order develops in the ETOPDM. It is reflected by a
250: power-law decay of the form $|\rho_{ij}|=0.25\ |(x_i-x_j)/a|^{-1/2}$, that
251: remains almost unchanged during the evolution of the system.
252: A careful inspection shows that this power-law behavior is restricted
253: to the regions where each lobe of the lowest NO exists, outside these
254: regions the ETOPDM decays faster. This quasi-coherent behavior in two
255: different segments of the system is the reason for the
256: degeneracy found in the lowest NO. In addition, a detailed analysis
257: of the ETOPDM Fourier's transform shows that the peak in the MDF at
258: $k=+\pi/2a$ is originated by components of the ETOPDM with $x_i,x_j>0$, and
259: the one at $k=-\pi/2a$ by the components with $x_i,x_j<0$, so that in
260: the regions of the lobes
261: $\rho_{ij}\approx 0.25\ |(x_i-x_j)/a|^{-1/2}e^{i\pi(x_i-x_j)/2a}$
262: for $x_i,x_j>0$, and
263: $\rho_{ij}\approx 0.25\ |(x_i-x_j)/a|^{-1/2}e^{-i\pi(x_i-x_j)/2a}$
264: for $x_i,x_j<0$. The prefactor of the power law (0.25) was found to be
265: independent of the number of particles in the initial
266: Fock state.
267: \begin{figure}[h]
268: \includegraphics[width=0.48\textwidth,height=0.23\textwidth]
269: {ETOPDM}
270: \caption{(color online). (a) ETOPDM for:
271: $\tau$=$50\hbar/t$ (\textcolor{red}{$\bigcirc$}),
272: $100\hbar/t$ (\textcolor{green}{$\times$}), and
273: $150\hbar/t$ (\textcolor{blue}{$\nabla$}), the line is
274: $0.25\ |(x_i-x_j)/a|^{-1/2}$ (see text for details).
275: (b) Scaled lowest NO right lobe's {\it vs} $x/N_ba$ for
276: $N_b=101$ (\textcolor{red}{$\bigcirc$}),
277: $201$ (\textcolor{green}{$\times$}), and
278: $301$ (\textcolor{blue}{$\nabla$}).
279: The inset in (a) shows the maximum occupation of the lowest NO
280: {\it vs} $N_b$ of the initial Fock state, the line is
281: $0.72\ N_b^{1/2}$.}
282: \label{ETOPDM}
283: \end{figure}
284:
285: After having observed off-diagonal quasi-long-range correlations in
286: the ETOPDM, it remains to consider how the occupation of the lowest NO
287: behaves in the thermodynamic limit, in order to see whether it corresponds
288: to a quasi-condensate. The only new information we need at this point is the
289: scaling relation (if any) of the modulus of the lowest NO
290: with the number of particles in the initial Fock state.
291: For this, we consider the NO at those times, where as shown in
292: the inset of Fig.\ \ref{NO}, they almost do not change in time,
293: i.e.\ after the plateau with $n=1$ disappears.
294: In Fig.\ \ref{ETOPDM}(b) we show that a scaled NO
295: ($|\varphi_0|=N_b^{1/2}|\phi^0|$) exists when its size is normalized
296: by $N_ba$. (The results for the left lobe are identical due to inversion
297: symmetry.) Figure \ref{ETOPDM}(b) also shows that the lobe size $L$ is
298: approximately $N_ba$. Then evaluating
299: $\lambda_0=\sum_{ij}\phi^{*0}_i \rho_{ij}\phi^0_j$
300: as the double of the integral over a single lobe ($L\gg a$) one obtains
301: \begin{eqnarray}
302: \lambda_0 &\sim& 2/a^2
303: \int^L_{0}dx \int^L_{0}dy \ \phi^{*0}(x)\rho(x,y) \phi^0(y)
304: \\ &=& N_b^{1/2}
305: \int^{1}_{0}dX \int^{1}_{0}dY
306: \frac{|\varphi^0(X)|0.25|\varphi^0(Y)|}{|X-Y|^{-1/2}}
307: =A\sqrt{N_b}, \nonumber \label{lambda0}
308: \end{eqnarray}
309: where we did the change of variables $x$=$XN_ba$, $y$=$YN_ba$,
310: $\phi^0$=$N_b^{-1/2}\varphi^0$, and we notice that the phase
311: factors between the NO and the ETOPDM cancel out. The integral over $X,Y$
312: is a constant that we call $A$. A confirmation of the validity of the
313: previous calculation is shown in the inset of Fig.\ \ref{ETOPDM}(a).
314: There we plot the maximum occupation of the lowest NO (reached when the
315: lobes of the NO have a stable form), as a function of the number of
316: particles in the initial Fock state. The $\sqrt{N_b}$ power-law behavior
317: is evident and a fit allowed to obtain the constant $A=0.72$.
318: Hence, this and the previous results demonstrate that the peaks of the MDF
319: in Fig.\ \ref{Perfil}(b) correspond to quasi-condensates with
320: momenta $k=\pm \pi/2a$.
321:
322: The appearance of such quasi-condensates at $k=\pm \pi/2a$
323: can be understood on the basis of total energy conservation.
324: Given the dispersion relation of HCB on a lattice,
325: since the initial Fock state has a flat MDF, its total energy
326: is $E_T = 0$. Would all the particles condense into one state,
327: it would be the one with an energy corresponding to
328: $\overline{\epsilon}_k = E_T/N$.
329: Taking into account the dispersion relation for HCB on a lattice
330: ($\epsilon_k=-2t\cos ka$), $\overline{\epsilon}_k=0$ corresponds
331: to $k=\pm \pi/2a$. Actually, in the 1D case there is only
332: quasi-condensation, so that the argument above applies only to
333: maximize the occupation of a given state. In addition, the minimum
334: in the density of states at these quasi-momenta reinforces
335: quasi-condensation into a single momentum state.
336:
337: The process of formation of the quasi-condensate is also characterized
338: by a power law. In Fig.\ \ref{LASER}(a) we plot
339: the occupation of the lowest NO as a function of the evolution
340: time. The {\it log-log} scale shows that the occupation of the
341: quasi-condensate increases quickly and decays slowly. A more
342: detailed examination shows that the population of the quasi-condensate
343: increases in an universal way ($1.38\sqrt{\tau t/\hbar}$, continuous line in
344: Fig.\ \ref{LASER}(a)) independently of the initial number of particles
345: in the Fock state. The power law is determined by the universal behavior
346: of the off-diagonal part of the ETOPDM shown before, and the fact that
347: during the formation of the quasi-condensate, the NO increases its size
348: linearly with time at a rate given by $|v_{NO}| = 2at/\hbar$.
349: Such a power law is followed almost up to the point where the maximal
350: occupation is reached. The time at which it is reached depends linearly
351: on the number of particles of the initial Fock state
352: $\tau_m=0.32N_b\hbar /t$ as shown in the inset of Fig.\ \ref{LASER}(a).
353: For a tipical experimental setup of rubidium atoms, in a lattice with a
354: recoil energy of $E_r=20$kHz and a depth of 20$E_r$, the time $\tau_m$
355: can be estimated as $\tau_m\sim 5.7N_b$(ms).
356: \begin{figure}[h]
357: \includegraphics[width=0.48\textwidth,height=0.235\textwidth]
358: {LASER}
359: \caption{(color online). (a) Time evolution of the lowest NO
360: occupation for $N_b=101$ (\textcolor{red}{$\bigcirc$}),
361: $201$ (\textcolor{green}{$\times$}), and
362: $301$ (\textcolor{blue}{$\nabla$}), the straight line
363: is $1.38\sqrt{\tau t/\hbar}$. The inset shows the
364: time at which the maximum occupation of the NO is reached
365: ($\tau_m t/\hbar$) {\it vs} $N_b$, the line is $0.32N_b$.
366: (b) Density profiles (inset) and MDF of 150 HCB evolving only
367: to the right in 1000 lattice sites for
368: $\tau=0$ (\textcolor{magenta}{$\triangle$}),
369: $50\hbar/t$ (\textcolor{red}{$\bigcirc$}),
370: $100\hbar/t$ (\textcolor{green}{$\times$}), and
371: $150\hbar/t$ (\textcolor{blue}{$\nabla$}).}
372: \label{LASER}
373: \end{figure}
374:
375: The results above showing that a quasi-coherent matter
376: front forms spontaneously from Fock states of HCB, suggest that
377: such an arrangement could be used to create atom lasers with a wave-length
378: that can be fully controlled given the lattice parameter $a$.
379: No additional effort is needed to
380: separate the quasi-coherent part from the rest since the quasi-condensate is
381: traveling at the maximum possible velocity on the lattice so that the front
382: part of the expanding cloud is the quasi-coherent part.
383: The actual realization would imply to restrict the evolution of the initial
384: Fock state to one direction, as shown in Fig.\ \ref{LASER}(b), where we
385: display the MDF for 150 HCB restricted to evolve
386: to the right in 1000 lattice sites at the same evolution times of Fig.\
387: \ref{Perfil}. It can be seen that the values of $n_{k=\pi/2}(\tau)$ are
388: almost the same in both situations, although in Fig.\ \ref{LASER}(b)
389: the initial Fock state has almost half of the particles. The same occurs to
390: the occupation of the lowest NO that in the latter case is not degenerated
391: anymore.
392:
393: The previous results suggest how to proceed in order to obtain
394: lasers in higher dimensional systems where real condensation can occur
395: \cite{lieb02}. One can employ Mott insulator states with one particle per
396: lattice site created by a very strong on-site repulsive potential $U$.
397: The latter is required in order to obtain a close realization of a
398: pure Fock state, since quantum fluctuations of the particle number present
399: in a Mott insulator for any finite $U$ \cite{hubbard} will be strongly
400: suppressed. Then the geometry of the lattice should be designed in order
401: to restrict the evolution of the Mott insulator to one direction only,
402: and to have a low density of states around the mean value of energy per
403: particle in the initial state. With these conditions the sharp features
404: observed in 1D should be reproduced by a condensate in higher dimensions.
405:
406: In summary, we have studied the non-equilibrium dynamics of a
407: Fock states of HCB when they are allowed to evolve freely. We have
408: shown that quasi-long-range correlations develop dynamically in these
409: systems and that they lead to the formation of quasi-condensates of HCB
410: at quasi-momenta $k\pm \pi/2a$. We have studied the dynamics of the formation
411: of these quasi-condensates, and their occupations were found to scale
412: proportionally to $\sqrt{N_b}$. These systems can be used to create
413: atom lasers since the quasi-condensate develops quickly and decays slowly.
414: The wave-length of such lasers is simply determined by the lattice parameter.
415: Finally, we have discussed the possibility of creating atom lasers in higher
416: dimensional systems where true condensation can occur.
417:
418: We thank HLR-Stuttgart (Project DynMet) for allocation of computer
419: time, and SFB 382 for financial support. We are gratefull to
420: F. de Le\'on for useful discussions.
421:
422: \begin{thebibliography}{99}
423:
424: \bibitem{dalfovo99} F. Dalfovo, S. Giorgini, L. P. Pitaevskii and S.
425: Stringari, Rev. Mod. Phys. {\bf 71}, 463 (1999).
426:
427: \bibitem{1995} M. H. Anderson {\it et~al.}, Science {\bf 269}, 198 (1995);
428: C. C. Bradley {\it et~al.}, Phys. Rev. Lett. {\bf 75}, 1687 (1995);
429: K. B. Davis {\it et~al.}, Phys. Rev. Lett. {\bf 75}, 3969 (1995).
430:
431: \bibitem{1D} F. Schreck {\it et~al.}, Phys. Rev. Lett. {\bf 87},
432: 080403 (2001); A. G\"orlitz {\it et~al.}, Phys. Rev. Lett. {\bf 87},
433: 130402 (2001).
434:
435: \bibitem{greiner01} M. Greiner {\it et~al.}, Phys. Rev. Lett. {\bf 87},
436: 160405 (2001); H. Moritz {\it et~al.}, Phys. Rev. Lett. {\bf 91},
437: 250402 (2003); T. St\"oferle {\it et~al.}, Phys. Rev. Lett. {\bf 92},
438: 130403 (2004).
439:
440: \bibitem{HCB} M. Olshanii, Phys. Rev. Lett. {\bf 81}, 938 (1998);
441: D. S. Petrov, G. V. Shlyapnikov, and J. T. M. Walraven, Phys. Rev. Lett.
442: {\bf 85}, 3745 (2000); V. Dunjko, V. Lorent, and M. Olshanii,
443: Phys. Rev. Lett. {\bf 86}, 5413 (2001).
444:
445: \bibitem{santos02} P. \"Ohberg and L. Santos, Phys. Rev. Lett. {\bf 89},
446: 240402 (2002); P. Pedri, L. Santos, P. \"Ohberg, and S. Stringari,
447: Phys. Rev. A {\bf 68}, 043601 (2003).
448:
449: \bibitem{girardeau60} M. Girardeau, J. Math. Phys. {\bf 1}, 516 (1960).
450:
451: \bibitem{girardeau02} M. D. Girardeau and E. M. Wright, Phys. Rev. Lett.
452: {\bf 84}, 5691 (2000); M. D. Girardeau and E. M. Wright, Phys. Rev. Lett.
453: {\bf 84}, 5239 (2000); K. K. Das, M. D. Girardeau, and E. M. Wright,
454: Phys. Rev. Lett. {\bf 89}, 170404 (2002).
455:
456: \bibitem{homog} A. Lenard, J. Math. Phys. {\bf 5}, 930 (1964),
457: H. G. Vaidya and C. A. Tracy, Phys. Rev. Lett. {\bf 42}, 3 (1979).
458:
459: \bibitem{trap} M.D. Girardeau, E.M. Wright, and J.M. Triscari,
460: Phys. Rev. A {\bf 63}, 033601 (2001); T. Papenbrock,
461: Phys. Rev. A {\bf 67}, 041601(R) (2003); P.J. Forrester, N.E. Frankel,
462: T.M. Garoni, and N.S. Witte, Phys. Rev. A {\bf 67}, 043607 (2003).
463:
464: \bibitem{rigol04_1} M. Rigol and A. Muramatsu, Phys. Rev. A {\bf 70},
465: 031603(R) (2004).
466:
467: \bibitem{yang62} C. N. Yang, Rev. Mod. Phys. {\bf 34}, 694 (1962).
468:
469: \bibitem{penrose56} O. Penrose and L. Onsager, Phys. Rev. {\bf 104},
470: 576 (1956).
471:
472: \bibitem{greiner02_1} D. Jaksch {\it et~al.}, Phys. Rev. Lett. {\bf 81},
473: 3108 (1998); M. Greiner {\it et~al.}, Nature {\bf 415}, 39 (2002).
474:
475: \bibitem{jordan28} P. Jordan and E. Wigner, Z. Phys. {\bf 47}, 631
476: (1928).
477:
478: \bibitem{lieb02} E. H. Lieb and R. Seiringer, Phys. Rev. Lett.
479: {\bf 88}, 170409 (2002).
480:
481: \bibitem{hubbard} G. G. Batrouni {\it et~al.}, Phys. Rev.
482: Lett. {\bf 89}, 117203 (2002); M. Rigol, A. Muramatsu, G. G. Batrouni,
483: and R. T. Scalettar, Phys. Rev. Lett. {\bf 91}, 130403 (2003).
484:
485: \end{thebibliography}
486:
487: \end{document}
488: