cond-mat0507190/gll.tex
1: \documentstyle[pra,aps,psfig]{revtex}
2: %%%%%%%%%%%%%%%%%%
3: \newcommand{\beq}{\begin{equation}}
4: \newcommand{\eeq}{\end{equation}}
5: \newcommand{\beqa}{\begin{eqnarray}}
6: \newcommand{\eeqa}{\end{eqnarray}}
7: \newcommand{\ba}{\begin{array}}
8: \newcommand{\ea}{\end{array}}
9: 
10: \begin{document}
11: \draft
12: 
13: %%%%%%%%%%%%%%%%%%
14: \twocolumn[\hsize\textwidth\columnwidth\hsize\csname
15: @twocolumnfalse\endcsname
16: %%%%%%%%%%%%%%%%%%
17: 
18: \widetext 
19: \title{Quasi One-Dimensional Bosons in Three-dimensional Traps: 
20: \\ From Strong Coupling to Weak Coupling Regime} 
21: \author{L. Salasnich$^{1}$, A. Parola$^{2}$ and L. Reatto$^{1}$} 
22: \address{$^{1}$Dipartimento di Fisica and INFM, Universit\`a di Milano, \\ 
23: Via Celoria 16, 20133 Milano, Italy\\ 
24: $^{2}$Dipartimento di Scienze Fisiche and INFM, 
25: Universit\`a dell'Insubria, \\ 
26: Via Valeggio 11, 23100 Como, Italy} 
27: 
28: \maketitle
29: 
30: \begin{abstract} 
31: We analyze a recent experiment on a Tonks-Girardeau gas  
32: of $^{87}$Rb atoms (T. Kinoshita, T. Wenger, 
33: and D.S. Weiss, Science {\bf 305}, 1125 (2004)). 
34: We find that the experimental data are compatible 
35: with the one-dimensional theory of 
36: Lieb, Seiringer and Yngvason (Phys. Rev. Lett. {\bf 91}, 150401 (2003)) 
37: but are better described by a theory that takes into account 
38: variations in the transverse width of the atomic cloud. 
39: By using this theory we investigate also the free axial expansion 
40: of the $^{87}$Rb gas in different regimes: Tonks-Girardeau gas, 
41: one-dimensional Bose-Einstein condensate and three-dimensional 
42: Bose-Einstein condensate. 
43: \end{abstract}
44: 
45: \pacs{PACS Numbers: 03.75.Kk}
46: 
47: %%%%%%%%%%%%%%%%%%
48: ]
49: %%%%%%%%%%%%%%%%%%
50: 
51: \narrowtext 
52: 
53: \par
54: Two experimental groups \cite{tg-psu,tg-mpi} have reported 
55: the observation of the one-dimensional (1D) Tonks-Girardeau (TG) 
56: gas \cite{g,ll} with ultracold $^{87}$Rb atoms 
57: in highly elongated traps. A rigorous theoretical 
58: analysis of the ground-state 
59: properties of a uniform 1D Bose gas, including the 
60: beyond-mean-field TG regime of 
61: impenetrable bosons, was performed by Lieb and Liniger (LL) 
62: forty years ago \cite{ll}. Recently, motivated by the 
63: experimental achievements, an extension of 
64: the LL theory for finite and inhomogeneous 
65: 1D Bose gases under longitudinal confinement has been 
66: proposed on the basis the local density approximation (LDA) \cite{lda}. 
67: \"Ohberg and Santos \cite{os} have suggested that 
68: the LDA is improved by including a gradient term that represent 
69: additional kinetic energy associated with the inhomogeneity 
70: of the gas. This conjecture has been rigorously proved by 
71: Lieb, Seiringer and Yngvason \cite{lsy}. 
72: More recently we have introduced \cite{gll} 
73: a variational approach, called generalized Lieb-Liniger theory (GLLT), 
74: which reduces to the Lieb-Seiringer-Yngvason theory (LSYT) 
75: in the 1D regime and, in addition, gives 
76: an accurate description of the crossover 
77: from the 1D regime to 3D regime. 
78: \par 
79: In this Brief Report we apply the GLLT to analyze the experimental results 
80: of Kinoshita, Wenger and Weiss \cite{tg-psu}. 
81: Contrary to the data of Paredes {\it et al.} \cite{tg-mpi} 
82: which are deeply in the TG regime, 
83: the data of Ref. \cite{tg-psu} cover different 
84: quantum-dimensional regimes and we show that the experimental atomic cloud 
85: is better described by the GLLT than the LSYT. 
86: The GLLT is then used to determine the axial free expansion 
87: of the $^{87}$Rb gas. We predict a self-similar expansion 
88: whose growth is strictly related to the quantum-dimensional 
89: regime of the initial configuration. 
90: \par 
91: In the LSYT \cite{lsy} the longitudinal density $\rho(z)$ 
92: of a zero-temperature Bose gas is obtained by minimizing 
93: the following energy functional 
94: \beq 
95: E_{LSY}[\rho] = \int_{-\infty}^{+\infty} \left\{ 
96: {\hbar^2 \over 2m} 
97: \left[ 
98: (\partial_z \sqrt{\rho})^2  
99: + \rho^3 e({g\over \rho}) 
100: \right] + V \; \rho \right\} dz \;\; ,  
101: \eeq
102: where $g=2a_s/a_{\bot}^2$ is the interaction parameter, 
103: with $a_s$ the s-wave scattering length and  
104: $a_{\bot}=(\hbar/(m \omega_{\bot})^{1/2}$ the 
105: characteristic length of the transverse harmonic potential 
106: with frequency $\omega_{\bot}$, 
107: $V(z)$ is the longitudinal external potential and 
108: $e(x)$ is the Lieb-Liniger function, 
109: which is defined as the solution of a Fredholm equation 
110: and it is such that $e(x) \approx x-4/(3\pi)x^{3/2}$ for $x\ll 1$ and 
111: $e(x) \approx (\pi^2/3)(x/(x+2))^2$ for $x\gg 1$ \cite{ll}. 
112: \par 
113: As previously stressed, the LSYT is valid in the pure 
114: 1D regime, where the transverse width $R_{\bot}$ of the 
115: Bose gas is frozen and equal to the harmonic length $a_{\bot}$. 
116: In our GLLT \cite{gll} the transverse properties 
117: of the Bose gas are taken into account by considering 
118: the adimensional width $\sigma = R_{\bot}/a_{\bot}$ 
119: as a variational parameter, i.e. $\sigma=\sigma(z)$. 
120: The theory gives the following energy functional 
121: $$
122: E_{GLL}[\rho,\sigma] = 
123: \int_{-\infty}^{+\infty}  
124: \left\{ 
125: {\hbar^2 \over 2m} \left[ 
126: (\partial_z \sqrt{\rho})^2  
127: + \rho^3 e({g\over \rho\sigma^2}) 
128: \right] 
129: \right. 
130: $$
131: \beq 
132: \left. 
133: + V \; \rho + {\hbar \omega_{\bot}\over 2} 
134: ({1\over \sigma^2}+\sigma^2 -2) \rho 
135: \right\} dz \;\; ,    
136: \eeq 
137: where $\hbar \omega_{\bot}(\sigma^{-2}+\sigma^2)\rho/2$ 
138: is the transverse energy density of the Bose gas. 
139: Note that in the low density region where $\sigma \approx 1$ 
140: (1D regime) the functional $E_{GLL}$ reduces to $E_{LSY}$. 
141: Instead, at higher densities (3D regime) where 
142: $e(g/(\rho \sigma^2)) \approx g/(\rho \sigma^2)$, 
143: the functional $E_{GLL}$ 
144: gives the energy functional of the nonpolynomial 
145: Schr\"odinger equation (NPSE) \cite{npse}, 
146: an effective 1D differential equation we have 
147: derived from the 3D Gross-Pitaevskii equation \cite{gpe} 
148: to describe Bose-Einstein condensates (BECs) 
149: under transverse harmonic confinement. 
150: \par 
151: Taking into account the normalization condition of the 
152: longitudinal density $\rho(z)$ to the total number $N$ 
153: of atoms, the minimization of the energy functional 
154: $E_{GLL}$ with respect to $\rho(z)$ gives the equation 
155: $$ 
156: {\hbar^2 \over 2m} \left[ - \sqrt{\rho} \; \partial_z^2 \sqrt{\rho} 
157: + 3 \rho^3 e({g \over \rho\sigma^2}) - {g \rho^2 \over \sigma^2} 
158: e'({g\over \rho\sigma^2}) \right] 
159: $$
160: \beq 
161: + V \; \rho + {\hbar \omega_{\bot}\over 2} 
162: ({1\over \sigma^2}+\sigma^2 -2 ) \rho = \bar{\mu} \; \rho \; ,   
163: \eeq 
164: where $\bar{\mu}$ is the chemical potential fixed by the 
165: normalization condition. 
166: The minimization of the energy functional 
167: $E_{GLL}$ with respect to $\sigma(z)$ gives instead 
168: the equation 
169: \beq 
170: \sigma^4 = 1 + a_{\bot}^2 g \rho \; 
171: e'({g \over \rho \sigma^2}) \;\; . 
172: \eeq 
173: This implicit equation must be solved numerically but 
174: analytical results can be obtained in limiting cases. 
175: In particular, under the condition $a_s \ll a_{\bot}$ \cite{mo}, 
176: one gets $\sigma \approx \sqrt{a_{\bot}} (\rho g)^{1/4}$ 
177: for $\rho \gg 1/a_s$ (3D BEC regime) and $\sigma \approx 1$ for 
178: $\rho \ll 1/a_s$ (1D regime). Note that the 1D regime contains 
179: two subregimes: the 1D BEC regime for 
180: $a_s/a_{\bot}^2 \ll \rho \ll 1/a_s$, and 
181: the TG regime, where $e(g/(\rho\sigma^2))\approx \pi^2/3$, 
182: for $\rho \ll a_s/a_{\bot}^2$. 
183: 
184: \begin{figure}
185: \centerline{\psfig{file=gll-f1.eps,height=3.in}}
186: \small 
187: {FIG. 1 (color online). Axial density $\rho(z)$ of a Bose gas 
188: of $50$ $^{87}$Rb atoms in an anisotropic harmonic 
189: trap with frequencies $\omega_{\bot}$ and $\omega_z$. 
190: Four values of trap anisotropy: $r=\lambda/\lambda_{max}$, 
191: where $\lambda= \omega_{\bot}/\omega_z$ and 
192: $\lambda_{max} = 2570$. Comparison 
193: among various theories: GLLT (full line); 
194: GLLT with TF approximation (dotted line); 
195: LSYT (dashed line); LSYT with TF approximation 
196: (long dashed line). Length in units $a_z$, 
197: and axial density in units $1/a_z$. } 
198: \end{figure}
199: 
200: \par 
201: In the experiment of Kinoshita, Wenger and Weiss \cite{tg-psu} 
202: an ensemble of about 6400 parallel 1D traps has been created 
203: by means of a 2D optical lattice, that strongly confines 
204: atoms in 1D tubes. The traps differ only in the 
205: number of $^{87}$Rb atoms each contains. 
206: As discussed in \cite{tg-psu}, each trap can be modelled 
207: by an anisotropic harmonic potential, where the axial frequency 
208: is $\omega_z=2\pi \times 27.5$ Hz and the transverse frequency 
209: $\omega_{\bot}$ is tuned by changing the energy 
210: depth $U_0$ of the confining optical lattice. The maximum transverse 
211: frequency is $2\pi \times 70.7$ kHz; 
212: it follows that the maximum trap anisotropy is 
213: $\lambda_{max}=2570$, where $\lambda =\omega_{\bot}/\omega_z$. 
214: Kinoshita {\it et al.} \cite{tg-psu} 
215: have worked with $r=\lambda/\lambda_{max}$ ranging 
216: between $r=1$ and $r=10^{-1}$, a regime where the Bose gas 
217: is quasi-1D. 
218: \par
219: By adding an imaginary time term at right side of Eq. (3) and 
220: solving Eq. (3) and Eq. (4) with a 
221: finite-difference Crank-Nicolson predictor-corrector 
222: scheme \cite{cn}, we calculate 
223: the density profile $\rho(z)$ of the Bose gas 
224: using the trap parameters of Ref. \cite{tg-psu}. 
225: In Eq. (3) we set $V(z)=m\omega_z^2 z^2/2$. 
226: The results are shown in Fig. 1, where we compare 
227: LSYT with GLLT, plotting also their Thomas-Fermi 
228: (TF) approximations which neglect the gradient term 
229: in Eq. (1) and Eq. (2). 
230: Figure 1 shows that, for the trap anisotropy $r=1$,  
231: the TF approximation is remarkably 
232: accurate but there are instead 
233: quantitative differences between LSYT and GLLT 
234: due to the fact that $\sigma (\rho(z))$ is not 
235: strictly equal to one. 
236: The differences between LSYT and 
237: GLLT increase by reducing the anisotropy 
238: $r$ of the trap, and also the TF approximation worsens. 
239: In fact, by reducing $r$ one induces a crossover 
240: from the 1D regime to the 3D regime, 
241: whose description cannot be accounted for by 
242: LSYT that is purely 1D \cite{lsy}. 
243: We have verified that the differences between LSYT and GLLT are instead 
244: reduced by taking larger values of the anisotropy ratio $r$. 
245: 
246: \begin{figure}
247: \centerline{\psfig{file=gll-f2.eps,height=2.7in}}
248: \small 
249: {FIG. 2 (color online). Root mean square (RMS) full length of 
250: the Bose cloud of $^{87}$Rb atoms. 
251: $U_0/E_{rec}$ is the trap depth $U_0$ in units 
252: of the atom recoil energy $E_{rec}$, where 
253: $U_0/E_{rec}=87 (\lambda/\lambda_{max})^2$ 
254: with $\lambda= \omega_{\bot}/\omega_z$ and 
255: $\lambda_{max} = 2570$. 
256: Experimental data are obtained in \cite{tg-psu} 
257: by averaging over several clouds with 
258: different numbers of atoms (maximum number $N_{max}=54$). } 
259: \end{figure} 
260: 
261: In Fig. 2 we plot the experimental data of the root mean 
262: square (RMS) length of the atomic cloud 
263: obtained by Kinoshita, Wenger and Weiss 
264: \cite{tg-psu}. This RMS axial length has been derived 
265: as the average over several clouds which have different 
266: numbers of atoms. In Fig. 2 the RMS full length is plotted 
267: versus $U_0/E_{rec}=87 r^2$, where $U_0$ 
268: is the energy depth of the optical potential and 
269: $E_{rec}$ is the recoil energy of the gas. 
270: Kinoshita {\it et al.} \cite{tg-psu} 
271: have compared their data with LSYT \cite{lsy} 
272: in the TF approximation \cite{lda}. 
273: By performing the averaging procedure of Ref. \cite{tg-psu} 
274: on our LSYT results, we obtain the same theoretical curve shown in Fig. 4 
275: of Ref. \cite{tg-psu} and plot it (dashed line) in our Fig. 2. 
276: With the same averaging procedure we obtain also the 
277: GLLT curve (dotted line). 
278: Figure 2 shows that our GLLT fits better than LSYT 
279: the experimental data. As discussed in \cite{tg-psu}, 
280: the experimental data for $U_0/E_{rec}<20$ are strongly 
281: affected by the tunneling between adjacent tubes 
282: and the theoretical analysis based on a single tube 
283: overestimates the axial width of the Bose cloud. 
284: \par 
285: Our GLLT is a simple and useful tool to investigate also dynamical 
286: properties of a Bose gas under transverse harmonic confinement. 
287: The dynamics of the Bose gas can be described by means of 
288: a complex classical field $\Phi(z,t)$ that satisfies 
289: the equation $
290: i \hbar \; \partial_t \Phi = {\delta \over \delta 
291: \Phi^*} E_{GLL}[|\Phi|^2 , \sigma ]  
292: $, 
293: where $|\Phi(z,t)|^2 = \rho(z,t)$. 
294: This time-dependent nonlinear Schr\"odinger equation 
295: can be explicitly written as 
296: \beq 
297: i \hbar \; \partial_t \Phi =  
298: \left( - {\hbar^2 \over 2m} \partial_z^2 
299: + V + \mu[|\Phi |^2] \right) \Phi \; , 
300: \eeq 
301: where $\mu[\rho ]$ is the bulk chemical potential, 
302: given by 
303: \beq 
304: \mu[\rho ] = {\hbar^2 \over 2m} \left[ 
305: 3 \rho^2 e({g \over \rho\sigma^2}) - {g\rho \over \sigma^2} 
306: e'({g\over \rho\sigma^2}) \right] 
307: + {\hbar \omega_{\bot}\over 2} 
308: ({1\over \sigma^2}+\sigma^2 - 2) \; . 
309: \eeq 
310: The Eq. (5) with Eq. (6) must be solved self-consistently with Eq. (4). 
311: It is important to observe that for $\rho \ll a_s/a_{\bot}^2$ 
312: (TG regime) one has $\sigma \approx 1$ and 
313: $\mu \approx \hbar^2 \pi^2 \rho^2/(2m)$  
314: and Eq. (5) reduces 
315: the the time-dependent nonlinear Schr\"odinger equation (NLSE) 
316: introduced by Kolomieski {\it et al.} \cite{ko}. 
317: For $\rho \ll 1/a_s$ (1D regime or TG regime) still 
318: we have $\sigma \approx 1$ but $\mu \approx 3\hbar^2 
319: \rho^2 e({g/\rho})/(2 m) -\hbar^2 g\rho e'(g/\rho )/(2m)$ 
320: and Eq. (5) becomes equal to the time-dependent 
321: NLSE proposed by \"Ohberg and Santos \cite{os}, 
322: which reduces to that of Kolomieski {\it et al.} in the TG regime.  
323: For $\rho \gg a_s/a_{\bot}^2$ (1D regime or 3D regime) the Eq. (5) 
324: gives $\sigma \approx (1 + g a_{\bot}^2 \rho )^{1/4}$, the Eq. (6) reads  
325: $\mu \approx \hbar^2 g\rho/(m \sigma^2) + \hbar \omega_{\bot} 
326: (\sigma^{-2}+\sigma^2-2)/2$ and Eq. (5) 
327: becomes the time-dependent NPSE \cite{npse}. The NPSE coincides with 
328: the equation of \"Ohberg and Santos in the 1D BEC regime, 
329: where $a_s/a_{\bot}^2 \ll \rho \ll 1/a_s$, but it accurately 
330: describes also the 3D BEC regime, where $\rho \gg 1/a_s$ and 
331: $\sigma \approx \sqrt{a_{\bot}}(g\rho )^{1/4}$. 
332: \par 
333: The time-dependent (TD) 
334: GLLT given by Eq. (5) must be carefully employed. In fact,  
335: Girardeau and Wright \cite{gw} have shown, studying the interference 
336: of two Bosonic clouds, that the time-dependent NPSE 
337: of Kolomieski {\it et al.} \cite{ko} overestimates the coherence of the 
338: Tonks-Girardeau gas. However, the TD GLLT can be safely 
339: used to calculate collective properties of the Bose gas. 
340: In Ref. \cite{gll} we have shown that 
341: the time-dependent GLLT gives the expected values 
342: for the axial breathing mode $\Omega_z$ 
343: of the Bose cloud in a harmonic confinement with 
344: $V(z)=m\omega_z^2 z^2/2$. In particular, we have found 
345: $\Omega_z =2 \Omega_z$ in the 
346: TG regime, $\Omega_z = \sqrt{3} \omega_z$ in the 1D BEC regime, 
347: and $\Omega_z = \sqrt{5/2} \omega_z$ in the 3D BEC regime. 
348: \par 
349: Here we analyze the free axial expansion of the Bose gas by means 
350: of the time-dependent GLLT. We consider the expansion of the 
351: Bose gas when the axial harmonic confinement is removed ($\omega_z=0$) 
352: while the radial one is kept fixed. 
353: By setting $\Phi(z,t) = \sqrt{\rho(z,t)} \exp{(i S(z,t)/\hbar)}$, 
354: the Eq. (5) is equivalent to the two hydrodynamics equations   
355: $
356: \partial_t \rho + 
357: \partial_z \left( \rho v \right) = 0 
358: $
359: and 
360: $
361: m \; \partial_t v  + \partial_z  
362: [ 
363: - \hbar^2 / (2 m \sqrt{\rho}) \partial_z^2 \sqrt{\rho}  + 
364: {m v^2/2} + V + \mu 
365: ] = 0 
366: $ 
367: of a 1D viscousless fluid with density field $\rho(z,t)$ and 
368: velocity field $v(z,t) = (\hbar /m)\partial_z S(z,t)$. 
369: If the initial axial width of the cloud is larger than the 
370: healing length $\hbar/\sqrt{2 m \mu}$ then 
371: one can safely neglect the quantum pressure (QP) term 
372: $-\hbar^2 /( 2 m \sqrt{\rho}) \partial_z^2 \sqrt{\rho}$ 
373: in the second hydrodynamics equation. Note that 
374: the bulk chemical potential $\mu$ of Eq. (6) scales as 
375: $\rho^{\gamma}$ in the 
376: three relevant regimes: $\gamma=2$ in the TG regime, $\gamma = 1$ 
377: in the 1D BEC regime, and $\gamma =1/2$ in the 3D BEC regime. 
378: It is straightforward then to prove that in these regimes 
379: the cloud density decreases 
380: during the time evolution following the self-similar solution 
381: $\rho(z,t)=\rho(z/b(t),t=0)/b(t)$, where the adimensional 
382: axial width $b(t)$ satisfies the equation 
383: \beq 
384: {\ddot b} = { \omega_z^2 \over b^{\gamma +1} } \; . 
385: \eeq 
386: The solution of this equation with initial conditions $b(0)=1$ 
387: and ${\dot b}(0) = 0$ can be obtained by quadratures. 
388: For large $t$ the solution is 
389: $b(t) = \sqrt{2/\gamma} \; \omega_z t $. 
390: In general, $\mu(\rho)$ is not 
391: a power law and during the expansion the functional dependence 
392: of $\mu(\rho )$ changes throughout the whole cloud: 
393: the expansion is no more truly self-similar. However, 
394: by evaluating the dynamics at the center of the cloud ($z=0$) 
395: where the initial density is $\rho_0$, from the two 
396: hydrodynamics equations without the QP term one finds 
397: \beq 
398: {1\over 2} {\dot b}^2 = 
399: {\omega_z^2 \left( \mu(\rho_0) - \mu(\rho­_0/b) \right) 
400: \over 
401: \rho_0 {\partial \mu \over \partial \rho}(\rho_0) } \; . 
402: \eeq 
403: For large $t$ the solution of this equation is 
404: $b(t) = \sqrt{2/{\bar \gamma}} \; \omega_z t $, 
405: where naturally appears the effective polytropic index 
406: \beq 
407: {\bar \gamma} = {\rho_0 \over \mu(\rho_0)} 
408: {\partial \mu \over \partial \rho} (\rho_0) \; , 
409: \eeq 
410: that is the logarithmic derivative of the chemical potential $\mu$. 
411: In order to verify these analytical results 
412: we simulate the free axial expansion of the Bose gas 
413: by numerically solving Eq. (5) self-consistently 
414: with Eqs. (4,6). In our simulation we employ a 
415: real-time Crank-Nicholson method \cite{cn}. 
416: \par
417: In Fig. 3 we plot the time evolution of the normalized axial 
418: width $W(t)$ of the Bose gas with $N=50$ atoms and 
419: initial anisotropy of the harmonic trap given by 
420: $r=1$ (${\omega_\bot /\omega_z} = 2570$). After a transient 
421: the width $W(t)$ grows linearly in time, as clearly shown by the lower 
422: panel of Fig. 3, where we plot the width velocity $dW/dt$. 
423: In Fig. 3 we plot also $W(t)$ and $dW/dt$ calculated with two 
424: alternative methods: by using Eq. (8) and by using 
425: Eq. (7) with $\gamma$ given by Eq. (9), in both cases 
426: taking into account Eqs. (4,6). 
427: 
428: \begin{figure}
429: \centerline{\psfig{file=gll-f3.eps,height=2.7in}}
430: \small 
431: {FIG. 3 (color online). Axial free expansion of the Bose gas 
432: of $50$ $^{87}$Rb atoms with initial anisotropy 
433: $\omega_{\bot} /\omega_z =2570$. 
434: Upper panel: time evolution of the normalized axial 
435: width $W(t)=\sqrt{\langle z^2 \rangle(t)/\langle z^2 
436: \rangle(0)}$ of the atomic cloud. 
437: Lower panel: time evolution of the velocity $dW/dt$ 
438: of the axial width. TD GLLT given by Eq. (5) and 
439: numerical solution of Eq. (7) with $\gamma$ from Eq. (9).} 
440: \end{figure} 
441: 
442: Figure 3 shows that 
443: the approximation based on the polytropic index is reliable; 
444: note that in all our tests the relative error in the slope of the 
445: asymptotic linear growth is always within $5\%$. 
446: \par 
447: In Fig. 4 we plot the slope $\alpha=\sqrt{2/{\bar \gamma}}$ 
448: as a function of the initial density $\rho_0$ at the center 
449: of the cloud. $\alpha$ smoothly changes from $1$ to $\sqrt{2}$ and 
450: than to $2$ in the transition from the TG regime to the 1D BEC regime 
451: and then to the 3D BEC regime. 
452: Figure 4 shows that for $r\ll 1$ in addition to the 
453: two plateaus corresponding to $\alpha =2$ at high densities 
454: (3D BEC regime) and to $\alpha =1$ at low densities (TG regime), 
455: an additional plateau appears with $\alpha =\sqrt{2}$ 
456: at intermediate densities (1D BEC regime). 
457: \par 
458: In conclusion, we have analyzed a recent experiment on 
459: dilute Bosons in the Tonks-Girardeau regime 
460: by using a beyond-mean-field theory 
461: that takes into account transverse variations of the atomic cloud. 
462: Remarkably, our theory shows a better agreement with the experimental 
463: results than the simple 1D approach proposed by various authors in the 
464: past years. On the basis of this theory, 
465: we have predicted a quasi self-similar expansion of the gas 
466: when the axial confinement is removed, showing that asymptotic 
467: linear growth of the axial width strongly depends 
468: on the quantum-dimensional regime of the Bosonic cloud. 
469: 
470: \begin{figure}
471: \centerline{\psfig{file=gll-f4.eps,height=2.6in}}
472: \small 
473: {FIG. 4 (color online). 
474: Slope $\alpha =\sqrt{2/{\bar \gamma}}$ of the asymptotic 
475: linear growth of the axial width of the Bose gas as a function of the 
476: initial density $\rho_0$ of the $^8$$^7$Rb cloud at the center 
477: of the trap. Three values of the initial anisotropy 
478: of the harmonic trap: anisotropy ratio $r$ defined as in Fig. 1.} 
479: \end{figure} 
480: 
481: \par 
482: L.S. thanks D.S. Weiss for enlightening e-suggestions. 
483: 
484: \begin{thebibliography}{99}
485: 
486: \bibitem{tg-psu} T. Kinoshita, T. Wenger, and D.S. Weiss, 
487: Science {\bf 305}, 1125 (2004). 
488: 
489: \bibitem{tg-mpi} B. Paredes {\it et al.}, 
490: Nature {\bf 429}, 277 (2004). 
491: 
492: \bibitem{g} M. Girardeau, J. Math. Phys. {\bf 1}, 516 (1960). 
493: 
494: \bibitem{ll} E.H. Lieb and W. Liniger, 
495: Phys. Rev. {\bf 130}, 1605 (1963). 
496: 
497: \bibitem{lda} D.S. Petrov, G.V. Shlyapnikov, 
498: and J.T.M. Walraven, Phys. Rev. Lett. {\bf 85}, 3745 (2000); 
499: V. Dunjko, V. Lorent, and M. Olshanii, 
500: Phys. Rev. Lett. {\bf 86}, 5413 (2001). 
501: 
502: \bibitem{os} P. \"Ohberg and L. Santos,  
503: Phys. Rev. Lett. {\bf 89}, 240402 (2002). 
504: 
505: \bibitem{lsy} E.H. Lieb, R. Seiringer, and J. Yngvason, 
506: Phys. Rev. Lett. {\bf 91}, 150401 (2003); 
507: Commun. Math. Phys. {\bf 244}, 347 (2004).  
508: 
509: \bibitem{gll} L. Salasnich, A. Parola, and L. Reatto, 
510: Phys. Rev. A {\bf 70}, 013606 (2004).
511: 
512: \bibitem{npse} L. Salasnich, Laser Phys. {\bf 12}, 198 (2002); 
513: L. Salasnich, A. Parola, and L. Reatto, 
514: Phys. Rev. A {\bf 65}, 043614 (2002). 
515: 
516: \bibitem{gpe} E.P. Gross, Nuovo Cimento {\bf 20}, 454 (1961); 
517: L.P. Pitaevskii, Sov. Phys. JETP {\bf 13}, 451 (1961). 
518: 
519: \bibitem{mo} Our discussion is valid only if the 
520: condition $a_s\ll a_{\bot}$ is fulfilled. In fact, 
521: when $a_s \approx a_{\bot}$ a 
522: confinement-induced resonance appears 
523: in the s-wave scattering length, as shown by   
524: M. Olshanii, Phys. Rev. Lett. {\bf 81},
525: 938 (1998). 
526: 
527: \bibitem{cn} E. Cerboneschi, 
528: R. Mannella, E. Arimondo, and L. Salasnich, 
529: Phys. Lett. A {\bf 249}, 495 (1998); 
530: L. Salasnich, A. Parola, and L. Reatto, 
531: Phys. Rev. A {\bf 64}, 023601 (2001). 
532: 
533: \bibitem{ko} E.B. Kolomieski, 
534: T.J. Newman, J.P. Straley, and X. Qi, 
535: Phys. Rev. Lett. {\bf 85}, 1146 (2000). 
536: 
537: \bibitem{gw} M.D. Girardeau, and E.M. Wright, 
538: Phys. Rev. Lett. {\bf 84}, 5239 (2000). 
539: 
540: 
541: \end{thebibliography}
542:               
543: \end{document}                                          
544: 
545: 
546: 
547: 
548: 
549: 
550: 
551: 
552: 
553: