One-dimension cubicquintic GrossPitaevskii equation in Bose-Einstein condensates in a trap potential

C. Trallero-Giner1 and R. Cipolatti2

2

Centro Brasileiro de Pesquisas Fisicas, Rua Xavier Sigaud 150, 22290-180 Rio de Janeiro-RJ, Brazil Instituto de Matemtica, Universidade Federal do Rio de Janeiro, C.P. 68530, Rio de Janeiro, RJ, Brasil a (Dated: August 14, 2012) By means of new general variational method we report a direct solution for the quintic self-focusing nonlinearity and cubic-quintic 1D Gross Pitaeskii equation (GPE) in a harmonic conned potential. We explore the inuence of the 3D transversal motion generating a quintic nonlinear term on the ideal 1D pure cigar-like shape model for the attractive and repulsive atom-atom interaction in Bose Einstein condensates (BEC). Also, we oer a closed analytical expression for the evaluation of the error produced when solely the cubic nonlinear GPE is considered for the description of 1D BEC.
PACS numbers: 03.75.Lm, 03.75.Hh, 03.75.Kk, 05.45.Yv

arXiv:1208.2371v1 [physics.atom-ph] 11 Aug 2012

I.

INTRODUCTION

Nowadays one and quasi-one dimensional BoseEinstein condensates (BEC) are common experimental procedures1. The transition from 3D to 1D system was invoked long time ago2 . In general, the 3D Gross-Pitaesvkii equation (GPE) cannot be factorized into transverse and longitudinal motions, nevertheless, under certain parameter regions we can assert that the BEC follows a 1D behavior (for a detailed discussion see Ref. 3). In the case of the harmonic trapping potential and considering that the atoms are tightly conned in two transverse directions, a transition to the quasi-1D description is possible. Starting with the standard 3D GPE, employing the adiabatic approximation and using the anzat wavefunction (x, r;t) = exp(i0 t/)(x)(r;t), we can derive an eective 1D GPE, which describes the physical characteristics of the cigar-like shape condensate 46 d2 1 + m 2 x2 +g1D ||2 g ||4 = 0 , (1) 2m dx2 2 where 0 R is the chemical potential, m is the atomic mass, is the longitudinal harmonic oscillator frequencies, and g1D R, g R are the eective 1D nonlinear self-interaction coecients. These two coecients depend on the total number N of particles in the condensate, the transverse harmonic oscillator frequency r and the scattering length as (as > 0 or as < 0 for attractive or repulsive interatomic interaction, respectively) by 2 the relations g1D = 2as N r and g = 6 ln(4/3)g1D / r 5 , where we have chosen for the stationary state the nor2 malization condition dx || = 1. Equation (1) is a cubic-quintic nonlinear Schdinger o equation (NLSE) with real coecients. The presence 4 of the g || term in (1) is due to the deviation from one dimension on the longitudinal condensate dynamics, i.e. a residual three dimensionality on an eective onedimensional GPE. In the case of a homogeneous medium, i.e. assuming that is zero, the cubic-quintic NLSE was
R 2

widely used to describe the physical process of an optical medium with a nonlinear polarization including susceptibilities up to fth order7 . Also, the soliton solutions have been extensively studies in Refs. 5,6,8. As we have mentioned above, Eq. (1) is the cigar-like shape approach from the 3D GPE. An important issue is the range of validity of (1), which is directly linked to the existence and stability of set of ground states solutions of the 3D NLSE. It is well known that for any value of as > 0 the 3D GPE does not collapse4 . However, for attractive interatomic interactions, the solution is dynamically stable if and only if as is within the range6,9 N |as | < 0.627, a (2)

with a = /mr . Hence, the validity of the cigar-like shape approach represented by Eq. (1) is also restricted to the constrain (2). Rescaling to dimensionless variables lo = /m, = x/lo , = 2g1D /(lo ), = 20 /( ), () = (lo )/ lo , = 3 ln(4/3)/r, Eq. (1) can be cast as d2 + 2 + ||2 2 ||4 = . d 2

(3)

The main task of this paper is the implementation of a more general variational mathematical approach to solve Eq. (3). Based on this result, we provide approximate solutions for the order parameter, the chemical potential and minimal energy for the quintic and cubic-quintic GPEs. The paper is organized as follows. First, in Sec. II we present the bases of our formalism, i.e., we present exact formulae for the energy and the chemical potential as functions of relevant parameters of (3). By considering a trial function for the ground state, we derive in Sec. III a representation for the energy, Eapp (), and chemical potential, app (). Section IV is devoted to the application of our results to get explicit approximate solutions for the quintic and also cubic-quintic NLSEs. An estimation of the error due to the inuence of the interaction

2 between the axial and radial degrees of freedom on the 1D cigar-shape model is presented both graphically and analytically as function of the self-interaction parameter and the coecient leading the quintic nonlinear term.
II. CUBIC-QUINTIC NONLINEAR GROSS-PITAEVSKII EQUATION

where
4 4 4 6 6 6

()

=
R

|()| d and

()

=
R

|()| d

In the following we will consider a more general nonlinear Gross-Pitaevskii equation d2 2 4 2 + 2 + a || b2 || = , d where a 0 and b are real constants. Let V = H 1 (R)|
R

(8) are the usual norms of the standard Banach spaces L4 (R) and L6 (R). Taking into account the relation between the energy and the chemical potential, i.e. E [ ]/ = , it follows that E [ ] d ; d d d Q [ ] = = 0. 2 = ;

R, (4)

(9)

2 |()|2 d < +

be the

Hence, from Eqs. (7) and (9) we obtain the useful formula dE [ ] a = d 2
4 4

Hilbert space endowed10 with the norm

R

and the corresponding inner product

|()| d +
R

2 |()| d

1/2

2b 3

6 6

Thus, the minimum energy is given by

Emin () = 1 +

a 2

s
0

4 4

ds

2b 3

s s
0

6 6

ds. (10)

(|)V =
R

d d + 2 ()() d. d d

Moreover, from Eq. (3) is straightforward that the chemical potential can be written as a 2
4 4

In V we dene the energy functional E [] =

R

d () d + d
R

min () = Emin () + |()| d b2 3

R 2 2

2b2 3

6 6

or equivalently10
6

a 2
R

|()| d

|()| d. (5)

We denote by G the set of ground states of Eq. (4), i.e., the set of functions of V that minimize the energy functional E [] under the condition
2

Q[] =
R

|()| d = 1.

(6)

Notice that in the case of attractive interaction where b > 0, it is possible to show (by applying the GagliardoNirenberg inequalities) that the set of ground states G is nonempty if the condition || < 2/ |b| is satised. Hence, for G we obtain dE [ ] = d E [ ] d ; d + a 2
4 4

It is important to remark that Eqs. (10) and (11) are exact under the condition of knowing the ground state G and therefore, independent of the method or approach we employe to get the solution of the Eq. (3).
III. APPROXIMATE FORMULAE

a min () = 1 + 2 2b 2 3

4 4

+
0

4 s 4

ds
6 6

6 6

+
0

s s

ds . (11)

2b 3

6 6

(7)

It is possible to show10 that any solution of Eq. (4) belonging to the Hilbert space V has the asymptotic behavior exp( 2 ), with > 0, as || . So, to evaluate the minimal energy Emin ( ), we can consider for the ground state the trial function

3 More precisely, the value of min (min () := min ) that minimizes the function E [ ] full the equation a 1 (1 2 ) 4 + 3 = . 4 4 (16)

() =

1/4

exp( 2 ).

(12)

Using the function (12) and evaluating the energy functional (5), we obtain the algebraic expression a 1 (13) E [ ] = (1 2 ) 2 + 2 + 4 2 with = 2b/(3 3) and = . In the case 0 the Eq. (13) presents a global minimum in (0, +) for any a and R, but for > 0 a global minimum is guaranteed if || < 1/ = 3 3/ 2b. Let () = () be the minimizer of Eq. (13), in this way the ground state solution in G can be searched considering the function () := () (). By using Eqs. (10) and (11) we obtain the approximate energy, Eapp (), and chemical potential, app (), namely a Eapp () = 1 + 2 and

Notice that, for 0 or > 0 under the condition || < 1/ , the function (13) is strictly convex and coercive on the interval (0, +), and in consequence Eq. (16) has a unique solution, while for || > 1/ the function (13) is not bounded from below.
A. Quintic NLSE

Firstly and for sake of comparison, we will consider the quintic NLSE in a harmonic potential11 . Choosing a = 0, from (16) we have
2 5 =

1 2 (1 2 )
1/2

(17)

(s)ds 2
0 0

s (s)ds

(14)

Inserting (17) into Eqs. (14) and (15) with a = 0, we

Repulsive

= -0.06

Mixture

interaction
=-0.06

a app () = 1 + () + (s)ds 2 0 2 2 2 () +
0

2
= -0.012

= -0.012

1
Attractive interaction Mixture

It becomes clear that || < 2/ b < S = 3 3/ 2b, ensuring that the set G is nonempty and that the functions Eapp () and app () can be good approximations of Emin () and min (), respectively.
IV. APPLICATIONS AND DISCUSSION OF THE RESULTS

s 2 (s)ds .

(15)

0
= 0.06 = 0.012

-1
= 0.06

= 0.012

-2

-8

-6

-4

-2

Explicit formulae (10), (11) and the approximate relationships (14), (15) are among the main results of our work. Nevertheless, more explicit expressions for the order parameter , the energy and the chemical potential as a function of the atom-atom interaction term would be desirable. To do so, let min = min (a, ; ) be the global minimizer of Eq. (13). Thus, Eqs. (14) and (15) with = min allow to obtain the dependence Eapp and app on the relevant physical parameters a, and . Depending on the values of and the sign of the parameter , we can distinguish several phases linked to both type of interaction strengths, i.e. i) pure attractive ( < 0, > 0), ii) pure repulsive ( > 0, < 0), and iii) a mixture ( < 0, < 0 or > 0, < 0).

FIG. 1: (Color online) 5 map for the parameter values = 0.06 and 0.012. According to the signs of and the character of the 5 , as approximated solution of Eq. (4) with a = 0, can be mapped into fourth zones: > 0, < 0 -pure repulsive interaction; < 0, > 0 -pure attractive interaction; > 0, > 0 and < 0, < 0 -mixture region.

obtain for the approximate energy E5 and chemical potential 5 the expressions E5 = 1 2 and 5 = 1 2
1/2 1/2

(18)

2 (1 2 )1/2

(19)

4 Figure 1 displays the 5 map diagram for several values of the parameter . It can be seen that the reduced chemical potential 5 =20 /( ) shows a strong dispersion as a function of , moreover and following the symmetry properties of Eq. (19), we observe that 5 increases (decreases) for the pure repulsive phase, > 0, < 0 (pure attractive phase, < 0, > 0), while in the mixture region an opposite behavior is reached with respect to the pair of values (, ). Following Eqs. (12) and (17) we obtain for the wavefunction 5 , the expression 5 , () = 1 (1 2 )1/2
1/4

3 2 1 0 -1 -2 -3 -4 -5 -6
0.1 0.2 0.3
5

=0

=0.05
24

<

=0.1
16

= 2.857/( )

1/2

<

3D

= 2.33/( )

1/2

exp

2 2 (1 2 )1/2

(20) valid for 1 > 2 . The above obtained wave function exhibits dierent behavior depending on the sign of and independent of the type interaction (attractive with < 0 or repulsive for > 0). The function becomes eectively less conned for < 0, i.e. 5 , is delocalized and its maximum decreases, while for > 0 the function 5 , () gets more localized and the maximum increases as the nonlinear potential 2 increases.
B. Cubic-quintic NLSE

-7 -8 -6 -4 -2 0 2 4 6 8

FIG. 2: (Color online) Reduced chemical potential 3 5 as a function of the dimensionless self-interaction parameter , calculated from Eqs. (22)-(27) and several values of coecient . Inset: - map showing the validity ranges of Eqs. (21)(27) and the stability region of 3D GPE given by < S = 2.857/()1/2 (orange) and < 3D = 2.33/()1/2 (gray lines), respectively.

Although one can solve Eq. (16) numerically and to obtain from Eqs. (14) and (15) the energy E3 5 (, b) and the chemical potential 2 5 (, b), it will be very useful to report explicit compact approximate solution of the cubic-quintic nonlinear 1D GPE (4). Searching the solution of Eq. (16) with a = 1 and b = > 0 as a Taylor

series on , 3 5 =

n n , we get
n=0

3 3 1 1 2 2 2 + E3 () = 1 + 4 64 12 256 2 1 1 4 3 (24) + 13 2 6 5 , 32 2 256 2 2

1 3 5 = + 2 16 2 3 2 + 2 256 6 3 1 1 3 + + 2 512 12 3 2 45 42 4 + ... + 3 2 211 128 3

1 3 2 2 2 1 3 + 3 () = 1 + 2 64 256 5 1 4 3 2 3 (25) + 12 2 6 5 . 2 256 32 2 2 (21) The terms E3 5 (, ) and 3 5 (, ) tell us how much error is made by neglecting the potential 2 ||4 in the 1D cigar-like shape model. These quantities are expressed by 3 2 + 2 9 8 7 2 4 1 3 + 11 , (26) 32 2

Under the condition || << 1 and by substituting Eq. (21) in (14) and (15) we have, for small enough E3 5 = E3 () + E3 5 (, ), (22)

E3 5 (, ) =

3 5 = 3 () + 3 5 (, ),

(23)

where E3 () and 3 () correspond to the energy and the chemical potential, respectively, for the cubic NLSE and are given, up to the 5th order, by10

and

5 and repulsive interatomic ( = 5) interactions. Also, in the gure is represented the values of 3 5 for = 0, 0.05, and 0.1. It becomes clear that for a given > 0, the variation of the function Uef f (; , ) with respect to the parameter is negligible. Hence, the corresponding chemical potentials 3 5 (, ) 3 5 (, 0). Thus, 2 the nonlinear potential || for > 0 becomes a very good description to study the physical properties of a 1D cigar-shape BEC under the condition that the transverse harmonic oscillator frequency r is much larger than the perpendicular frequency trap . This result is in agreement with the conclusions of Ref. 12. As one can see from Fig. (2) the error 3 5 ( > 0, ) ranges between 2-7 % with respect to the value of 3 (). Now, if we consider the opposite case, i.e. an attractive interaction, the behavior of Uef f (; , ) presents a strong localized potential and, in correspondence, the chemical potential will change drastically as the parameter increases. This characteristic of the eective potential (28) for < 0 de2 termines clearly that the term || is not enough for a correct description of 1D cigar-like shape BECs. In this 4 case, the residual three dimensionality term 2 || , for the eective 1D GPE under a harmonic trap, strongly modify the corresponding chemical potential. Small variation of the strength 2 leads to strong change of 3 5 value and hence the ground state energy E3 5 as well. This peculiar behavior is related to the orbital stability of the nonlinear Schrdinger equation (4). As mentioned o above, the 3D GPE presents a set of ground states which is orbitally stable for any value of the self-repulsive interaction, while for the attractive interparticle interaction regime (as < 0) the solution does not collapse if and only if the condition N as > a 0.627 is fullled. The for mer criterion can be rewritten as < 3D = 2.33/ and it should be compared with condition of existence and validity, for < S = 2.857/ , of the obtained solutions Eqs. (22)-(27). In the inset Fig. (2) for the attractive interaction, the intersection of these two sets, 3D () S (), is represented by a shaded region in the diagram. In conclusion, a new variational approach is presented, which allows to construct for the cubic-quintic GPE closed analytical expressions for the order parameter, the minimal energy, and the corresponding chemical. Focusing on the compact analytical expressions, we report the contribution of the quintic term and the systematic error of the residual 3D contribution to the 1D cigar-shape model for both attractive and repulsive interaction. By the calculations the obtained chemical potential solution highlight the strong dependence on the sign of strength and the values of quintic self-interaction parameter .

3 2 + 2 3 5 (, ) = 3 6 7 2 4 5 3 + 10 . (27) 96 2 Figure (2) is devoted to the calculated chemical potential 3 5 using Eqs. (23), (25), and ( 27) as a function of for = 0, 0.05, and 0.1. First, by comparing Figs. (1) and (2) we see the strong qualitative dierence between the two types equations here considered, quintic and cubicquintic NLSEs. Notice that this dierence is remarkable even at 0. From the Fig. (2) we can assert that the cigar-like shape approximation retaining term up to 2 || is a good approach solely for the repulsive case. The chemical potential (and also, the energy) for > 0 and << r is almost independent of = 3 ln(4/3)/r, while a not negligible contribution is reached to 3 5 if the atom-atom interaction is attractive, even for very small value of . In the former case the term 2 ||4 is responsible for the strong obtained dispersion compared to the = 0 limit. In order to understand from the physical point of view the behavior of the chemical potential on , we dene the eective potential

Repulsive interaction

=0.05

Effective potential

=0
2

= 5

=0.1 Atractive interaction

=0
-2

= - 5

-4

=0.05

=0.1

-6

-2

-1

FIG. 3: (Color online) The potential Uef f given by Eq. (28) for = 5 (solid lines) and = 5 (dot-solid lines). In the calculation we employed the values of = 0 (green), 0.05(blue) and 0.1 (black). The chemical potential 3 5 for each is indicated by at lines.

Uef f = 2 + | | 2 | | ,

(28)

Acknowledgments

where the order parameter () := () () has been substituted by the trial function (12) with 3 5 (; ) := 3 5 . Figure (3) shows the potential Uef f for both considered cases, attractive ( = 5)

C.T-G acknowledges the hospitality at the MaxPlanck-Institut fr Physik Komplexer Systeme and u thanks Alexander von Humboldt Foundation for nan-

6 cial support.

A. Grlitz, et al., Phys. Rev. Lett. 87, 130402, (2001); o M. Greiner , I. Bloch, O. Mandel, T. W. Hnsch, and T. a Esslinger, Phys. Rev. Lett. 87, 160405 (2001); F. S. Cataliotti, et al., Science 293, 843 (2001). E. H. Lieb, Phys. Rev. 130, 1616 (1963); M. Olshanii, Phys. Rev. Lett. 81, 938 (1998); D. S. Petrov, G. V. Shlyapnikov and J. T. M. Walraven, Phys. Rev. Lett. 85, 3745 (2000). E. H. Lieb, R. Seiringer, and J. Yngvason, Commun. Math. Phys. 244, 347 (2004). R. Carretero-Gonzlez, D. J. Frantzeskakis, and P. G. a Kevrekidis, Nonlinearity 21, R139, (2008). A. E. Muryshev , G. V. Shlyapnikov , W. Ertmer, K. Sengstock, and M. Lewenstein, Phys. Rev. Lett. 89, 110401 (2002). L. Khaykovich and B. A. Malomed, Phys. Rev. A 74 , 023607 (2006). Kh. I. Pushkarov, D. I. Pushkarov, and I. V. Tomov, Opt.

10

11

12

Quantum Electron., 11, 471 (1979). S. Sinha, A. Y. Cherny, D. Kovrizhin, and J. Brand, Phys. Rev. Lett. 96, 030406 (2006); L. Salasnich, A. Parola , and L. Reatto, Phys. Rev. A 65, 043614 (2002). V. M. Prez-Garc H. Michinel, and H. Herrero, Phys. e a, Rev. A 57, 3837 1(998); A. Gammal, L. Tomio, and T. Frederico, Phys. Rev. A 66, 043619 (2002); L. D. Carr and Y. Castin, Phys. Rev. A 66, 063602 2002 . R. Cipolatti, J. Lpez Gondar, and C. Trallero-Giner, o Physica D-Nonlinear phenomena 241, 755 (2012). This type of equation, excluding the connemet potential, have considered for studing the dynamic of the dark soliton and the properties of a 1D BEC [see V.A. Brazhnyi, V. V. Konotop and L. P. Pitaevskii, Phys. Rev. A 73 053601(2006) and reference therein]. E. H. Lieb, R. Seiringer, and J. Yngvason, Phys. Rev. Lett. 91, 150401 (2003).