Higher Dimensional Gaussian-Type Solitons of Nonlinear Schrödinger Equation with Cubic and Power-Law Nonlinearities in PT-Symmetric Potentials

Two families of Gaussian-type soliton solutions of the (n+1)-dimensional Schrödinger equation with cubic and power-law nonlinearities in -symmetric potentials are analytically derived. As an example, we discuss some dynamical behaviors of two dimensional soliton solutions. Their phase switches, powers and transverse power-flow densities are discussed. Results imply that the powers flow and exchange from the gain toward the loss regions in the cell. Moreover, the linear stability analysis and the direct numerical simulation are carried out, which indicates that spatial Gaussian-type soliton solutions are stable below some thresholds for the imaginary part of -symmetric potentials in the defocusing cubic and focusing power-law nonlinear medium, while they are always unstable for all parameters in other media.


Introduction
The construction of the exact solutions of nonlinear partial differential equations is one of the most important and essential tasks in various branches from mathematical physics, engineering sciences, chemistry to biology [1,2]. Many powerful methods have been presented, such as the (G'/G)-expansion method [3][4][5], the variable separation method [6,7], the multiplier approach [8], the similarity transformation method [9] and the extended generalized Riccati equation mapping method [10], and so on.
The nonlinear Schrödinger equation (NLSE) and its relatives play important role in in physics, biology and other fields. Researchers have studied abundant mathematical solutions and physical localized structures of various NLSEs, including solitons and nonautonomous solitons [11,12], similaritons [13], rogue waves [14] and breathers [15] etc.
In recent years, the propagation of solitons in parity-time (PT ) symmetric potentials are presently attracting a great interest both from the theoretical and from the applicative point of view [16][17][18][19][20][21][22][23][24][25][26]. The definitions of PT potentials were given by Bender and coworkers in classical quantum mechanics, namely, the PTsymmetric potential satisfies V(r)~V Ã ({r) with Ã denoting complex conjugation [27]. Considering the mathematical correspondence between the quantum Schrödinger equation and the paraxial equation of diffraction-NLSE, the concept of PT symmetry has been introduced in the field of optics. Pioneering theoretical contributions of Christodoulides and co-workers [16,17] stimulated recent experimental observations [18,19]. After then, optical solitons in PT -symmetric Rosen-Morse potential [20], periodic potential [21] and Scarf II potential [22] were discussed. Two-dimensional (2D) solitons in nonlocal media [23] with PTsymmetric potentials have also been reported. More recently, the propagation of nonautonomous solitons in optical media with PT symmetry has been a subject of intense investigation [24][25][26].
However, all soliton solutions in [20][21][22][23][24][25][26] are sech-type. Other type of soliton solutions in PT -symmetric potentials is less studied. Especially, higher dimensional Gaussian-type soliton solutions in PT -symmetric potentials with cubic and power-law nonlinearities are hardly studied. In this paper, we aim to obtaining some analytical higher dimensional Gaussian-type soliton solutions of NLSE with cubic and power-law nonlinearities in PT -symmetric potentials. Two issues are firstly studied in this present paper: i) higher dimensional Gaussian-type soliton solutions are analytically presented in PT -symmetric media with cubic and power-law nonlinearities, and ii) linear stability analysis and direct simulation are used together to investigate the stability of solutions in PT -symmetric media with cubic and power-law nonlinearities.

Analytical higher dimensional Gaussian-type soliton solutions
The propagation of spatial soliton and LB in a PT -symmetric nonlinear medium of non-Kerr index can be described by the following NLSE where u(z,x) with x:fx n g is the complex envelope of the electrical field, x 1 ,x 2 and x 3 represent the transverse spatial coordinates x,y and the retarded time t, respectively. The subscripts z and x n in the first and second terms in Eq. (1) denote the derivatives to them. Parameters b 1 ,b 2 and b 3 are respectively the coefficients of the diffraction and dispersion, c is the cubic nonlinear coefficient and c 2mz1 for m~2,:::,j describe the nonlinearities of orders up to 2jz1. For m~2 one has the quintic nonlinearity, for m~3 the septic, and so on. Functions V(x) and W(x) describe the index guiding and the gain/loss distribution respectively. The real and imaginary components of the complex PT -symmetric potential satisfy V(x)~V({x) and W(x)~{W({x). If N~1, Eq. (1) is 1DNLSE, and its solutions are 1D spatial soliton solutions. If N~2, Eq. (1) is 2DNLSE, and its solutions are 2D spatial soliton solutions. If N~3, Eq. (1) is 3DNLSE, and its solutions are LB solutions. We seek solutions of NLSE (1) in the form: Substituting it into Eq. (1) leads to two differential equations about real functions r and W In the following, we derive analytical Gaussian-type soliton solutions of Eqs.
with real parameters V 1n~{ 4b n a 4 n ,V 2n~W 2 n =(36b n a 4 n ),V 4~{ c 2mz1 ({V 3 =c) m and arbitrary constants V 3 ,W n and a n , the localization condition r?0 as x?+? brings into solution of Eq. (1) as follows with l~{2 P N n~1 b n a 2 n and the error function erf (o). Family 2 Solution in the second type of extended PT -symmetric potential. When the PT -symmetric potential has the form with real parameters V 1n~{ 4b n a 4 n =m 2 ,V 2n~m 2 W 2 n =½4(mz2) 2 b n a 4 n ,V 3{ c 2mz1 ({V 4 =c) m and arbitrary constants V 4 ,W n and a n , the localization condition r?0 as x?+? leads to solution of Eq. (1) in the form with l~{ 2 m X N n~1 b n a 2 n and the error function erf (o).

Characteristic quantities of analytical solutions
It is obvious that the real and imaginary parts of the PT -symmetric potentials (5) and (7) are both even and odd functions with regards to x n . Thus, V and W exhibit the symmetric and anti-symmetric properties. As an example, we present 2D case of these properties in Fig. 1. From Fig. 1(b), V possesses the plateau-like structure with m~2. with the increase of m, this structure turns into a two-hump structure, and the humps protrude little by little.
In the PT -symmetric potentials above, the phase switches of solutions (6) and (8) can be found. Fig. 2(a) presents the phase switch of solution (6). The comparison of phase switch of solution (8) with the different m at x~0 is shown in Fig. 2(b). The span of switch gradually enlarges with the increasing order m of nonlinearity.
The power P and power-flow density (Poynting vector)S are two important quantities. They can be calculated from P~Ð z?
{? ju(z,x)j 2 dx, and S~(i=2)½Y+Y Ã {Y Ã +Y. From these definitions, the powers P of solution (6) and (8) respectively. The power-flow densitiesS of solution (6) and (8) can be expressed z1)a 2 n x 2 n g, respectively. Note that the notation f : g combines multi-component case, which means that there is one component for 1D case, and two components ½ : , : for 2D case, etc. From these detailed expression above, We can find that the power and the power-flow density of solution (6) are both independent of the parameter m, while those of solution (8) both depend on the parameter m. Considering V 3 cv0 or V 4 cv0, S is everywhere positive, which implies that the power flow and exchange for solutions (6) and (8) in the PT cells are always in one direction, that is, from the gain toward the loss regions. Two examples to this exchange among gain or loss regions are shown in Fig. 2(c) and 2(d) for 2D case.  When N is chosen as other values, such as N~1 and N~3, the phase switch and the directional power-flow density can also be found. Here we omit these discussions for the limit of the length. In the next section, we only focus on the dynamical behaviors of solutions (6) and (8) with N~2.

Linear stability analysis of analytical solutions
In order to discuss the linear stability of analytical solutions (6) and (8) of Eq. (1), we consider a perturbed solution [28] , where E is an infinitesimal amplitude, u 0 (x) is a respective eigenmode [analytical solution of Eq. (1)], R(x) and I(x) are the real and imaginary parts of perturbation eigenfunctions, which may grow upon propagation with the perturbation growth rate s. It is obvious that the perturbed solution becomes linearly unstable if there exist nonzero imaginary parts of s, otherwise solution is stable.
Substituting the perturbed solution into Eq. (1) and linearizing it around the unperturbed one (the first-order term of E), we arrive at the eigenvalue problem where s is an eigenvalue, R and I are eigenfunctions with Hermitian operators In the following, we discuss the whole eigenvalue spectra of the above problem (9). In order to perform the numerical computation, here we have restricted b 1~1 ,b 2~1 :1,V 3~5 ,V 4~4 with a 1~0 :45,a 2~0 :4 in solution (6) and a 1~0 :4,a 2~0 :45 in solution (8) with N~2. We discuss the linear stability analysis of solutions (6) and (8) in the self-focusing and self-defocusing cubic and power-law nonlinearities. Fig. 3(a) presents stable and unstable regions of some order parameters m for the self-defocusing cubic (DC) and self-focusing power-law (FP) nonlinearities in 2D case when other parameters are chosen as b 1~1 ,b 2~1 :1,a 1~0 :45,a 2~0 :4, V 3~5 ,W 2~0 :015,c~{1,c 2mz1~0 :7. For a certain m, if other parameters are fixed, there exists a threshold value of W 1 , above which analytical solution becomes unstable and below which analytical solution evolves stably. From Fig. 3(a), the threshold value of W 1 decreases at first, next adds to a maximum when m~4, then attenuates again, and finally increases to a certain value when m~13. For m~2, the threshold value of W 1 is close to 0.0234. For m~3, it is W 1 *0:021, and for m~4, W 1 *0:0297. For m~8, W 1 *0:0231. After mw8 the threshold value of W 1 increases, and reaches to a fixed value (W 1 *0:0245) after m~13.
For other nonlinearities such as the self-focusing cubic (FC) and self-focusing power-law (FP) nonlinearities, the self-focusing cubic (FC) and self-defocusing power-law (DP) nonlinearities and the self-defocusing cubic (DC) and selfdefocusing power-law (DP) nonlinearities, the eigenvalue s of solution (6) exists many imaginary parts shown in Fig. 3(b)-3(d), and thus solution (6) is always unstable in these nonlinear media.
The linear stability analysis of solution (8) has the similar result with that of solution (6). In the FC and FP, the FC and DP, and the DC and DP nonlinear media, the eigenvalue s of solution (8) appears many imaginary parts, and thus solution (8) is also linearly unstable. In the DC and FP nonlinear medium, if all other parameters are fixed, solution (8) is linearly stable only in the case when the values of W 1 and W 2 are chosen below their threshold values. Table 1 lists the thresholds of W 1 and W 2 of solution (8) for b 1~1 ,b 2~1 :1,a 1~0 :4,a 2~0 :45, V 4~4 ,c~{1,c 2mz1~0 :7 in 2D case with N~2. Table 1 indicates that the values of W 1 and W 2 decrease quickly with the increase of the order parameter m. Different from the case in Fig. 3(a), the values of W 1 and W 2 always decrease in Table 1. The gain (loss) related to the values of W 1 and W 2 are fairly small compared with the bigger value of V 4 , otherwise, all analytical solutions finally result in instability.

Numerical calculation for the stability of analytical solutions
The linear stability analysis gives the stable regions of analytical solutions in different 2D extended PT -symmetric potentials. However, the analytical solutions are not exactly satisfied in real situations, thus it is important to discuss the stability of solutions with respect to finite perturbations. In the following, we further discuss the stability of these solutions against a perturbation of 5% white noise by the direct numerical simulation (a split-step Fourier beam technique).
The stable and unstable 2D spatial soliton solution (6) in the 2D extended PTsymmetric potential (5) are presented in Figs. 4 and 5, respectively. From Fig. 4, 2D spatial soliton solution (6) with m~2,3,5 stably propagate over tens of diffraction lengths in the DC and FP nonlinear medium, and do not yield any visible instability except for some small oscillations. We can find a good agreement with results from the linear stability analysis for analytical solution (6). This indicates that the PT complex potential is strong enough to inhibit the collapse of spatial soliton solutions caused by diffraction and DC and FP nonlinearities. From Fig. 4(b), when m~2, the white noise only influences the background of soliton and produces some small oscillations around the soliton. When m~3 and 5, the white noise brings some small oscillations on the top part of soliton shown in Fig. 4(c) and 4(d).
In other nonlinear media, solution (6) is unstable in the 2D extended PTsymmetric potential (5). As two examples, Fig. 5 displays this kind of instability. For the DC and DP nonlinearities and the FC and DP nonlinearities, these solitons can not maintain their original shapes. Along the propagation distance, these solitons collapse, and finally decay into noise. With the increase of nonlinear order parameter m, the instability of solution (6) adds. Compared Fig. 5(a) and 5(c) with Fig. 5(b) and 5(d), solution (6) with m~3 are more unstable than that with m~2.
Spatial soliton solution (8) also exhibits stable and unstable propagations in different nonlinear media. In the DC and FP nonlinear medium, solution (8) with m~2,3,5 can stably propagate over tens of diffraction lengths. The influence of initial 5% white noise is suppressed, and only some small oscillations happen. Similar to the case shown in Fig. 4, the white noise only exerts an effect on the background of soliton and generates some small oscillations around the soliton for m~2 in Fig. 6(b). When m~3 and 5, the white noise only affect the top part of soliton, and there are some small oscillations on the top part of soliton shown in Fig. 6(c) and 6(d). Compared Fig. 4(a) and 4(b) with Fig. 6(a) and 6(b), solution (6) is more stable than solution (8) for the case of m~2 because white noise produce more small oscillations in Fig. 4(a) and 4(b) than that in Fig. 6(a) and 6(b). Some examples of unstable spatial soliton solution (8) are presented in Fig. 7. In the DC and DP nonlinear medium and the FC and FP nonlinear medium, spatial solitons are broken down propagating after several diffraction lengths, and their original shapes can not be preserved, especially to the case of large nonlinear order m. When m~2 in Fig. 7(a) and 7(c), spatial solitons are distorted, then spread to the background, and next decay into noise. When m~3 in Fig. 7(b) and 7(d), spatial solitons almost become noise. Therefore, solution (8) with m~3 are more unstable than that with m~2, and this instability also adds with the increase of nonlinear order parameter m.

Conclusions
We analytically obtain two families of Gaussian-type soliton solutions of the (n+1)-dimensional Schrödinger equation with cubic and power-law nonlinearities in PT -symmetric potentials. As an example, we discuss some dynamical behaviors of two dimensional soliton solutions. Their phase switches, powers and transverse power-flow densities are discussed. Results imply that the power flow and exchange from the gain toward the loss regions in the PT cell. Moreover, the linear stability analysis and the direct numerical simulation are carried out, which indicates that spatial Gaussian-type soliton solutions are stable below some thresholds for the imaginary part of PT -symmetric potentials in the defocusing cubic and focusing power-law nonlinear medium, while they are always unstable for all parameters in other media. These results will enrich the variety of higher dimensional structures in PT -symmetric potential in the field of mathematical physics, and may also have potential values to the application of synthetic PTsymmetric systems in nonlinear optics.