Manuscript Details

Manuscript number JOCS_2017_171

Title A hybrid approach for numerical solution of nonlinear Schrodinger equation

Article type Full Length Article

Abstract
The present study aims to develop a hybrid differential quadrature method with modified trigonometric cubic B-spline
basis functions for solving nonlinear Schrodinger equation in both one and two dimensions. This method reduces the
nonlinear equation into a set of ordinary differential equations which can be solved by a modified form of Ruge-Kutta
method. This proposed method has been applied to this equation using two different approaches and has also been
tested for proficiency on several numerical examples of one and two dimensions. The insignificant errors are found
when compared with the exact solution. The obtained numerical results are also in good agreement with the results
available in the literature. Comparison of numerical and the exact solution is depicted in the form of figures and tables.

Keywords Differential Equation; Modified Trigonometric B-Spline; Differential Quadrature

Method; Schrodinger Equation.

Corresponding Author Varun Joshi

Order of Authors Geeta Arora, Varun Joshi

Suggested reviewers Y N Reddy, Ujwal Warbhe, S Dinesh Kumar

Submission Files Included in this PDF

File Name [File Type]
Cover letter.doc [Cover Letter]

SE.pdf [Manuscript File]

Biogra.docx [Author Biography]

Photos.docx [Author Photo]

highlights.docx [Highlights]

To view all the submission files, including those not included in the PDF, click on the manuscript title on your EVISE
Homepage, then click 'Download zip file'.
 The Editor, Date: 2nd April ' 17

Journal of Computational Science,

Dear Sir,

Find the attached copy of my manuscript entitled A hybrid approach for numerical solution
of nonlinear Schrodinger equation for favor of publication in your reputed Journal, Journal of
Computational Science.
Hope you find it suitable for publication in your Journal.

Thanks!

Yours truly,

Varun Joshi
 Journal
Logo
00 (2016) 117

A hybrid approach for numerical solution of nonlinear Schrodinger

equation
Geeta Arora, Varun Joshi1,
Department of Mathematics, Lovely Professional University, Punjab, India

Abstract
The present study aims to develop a hybrid dierential quadrature method with modified trigonometric cubic B-spline basis func-
tions for solving nonlinear Schrodinger
equation in both one and two dimensions. This method reduces the nonlinear equation into
a set of ordinary dierential equations which can be solved by a modified form of Ruge-Kutta method. This proposed method has
been applied to this equation using two dierent approaches and has also been tested for proficiency on several numerical examples
of one and two dimensions. The insignificant errors are found when compared with the exact solution. The obtained numerical
results are also in good agreement with the results available in the literature. Comparison of numerical and the exact solution is
depicted in the form of figures and tables.

Keywords: Dierential Equation, Modified Trigonometric B-Spline, Dierential Quadrature Method, Schrodinger
Equation.
2000 MSC: 65M06, 35K57

1. Introduction

Material science has many fundamental equations for portraying the quantum mechanical behaviour [1]. Among
them, the most important equation is nonlinear Schrodinger
(NLS) equation which is used to elucidate the changes of
quantum system with time. This partial dierential equation also known as Schrodinger wave equation, with cubic
nonlinearity, delineates the advancement of wave function of a physical framework after some time. Additionally,
this equation has application in various fields of physical sciences such as in configuration of optoelectronic gadgets,
in study of electromagnetic wave proliferations, quantum flow computations, underwater acoustics, portraying the
mobility of Bose-Einstein condensate at temperature near absolute zero, signal propagation in optical fibers and in
many other physical nonlinear systems having instability phenomena (See Refs [2-7]). This proposal utilizes the NLS
equation with trapping potential for its extensive use. This equation has many remarkable features such as existence
of plane wave stationary solution, solution in form of localized wave with preservation of shape with advancement in
time, etc.
The nonlinear Schrodinger
equation in one dimension is as follows:

i Ut = U xx + |U|2 U + F(x, t)U, x [a, b], t 0 (1)

with initial condition

U(x, 0) = U0 (x) (2)

Email addresses: geetadma@gmail.com (Geeta Arora), varunjoshi20@yahoo.com (Varun Joshi)

1
117 2

and boundary condition

lim U(x, t) = 0 (3)
|x|

In two dimension, nonlinear Schrodinger

equation can be considered in the following form:

i Ut = (U xx + Uyy ) + |U|2 U + F(x, y, t)U, x, y [a, b] [c, d], t 0 (4)

with initial condition

U(x, y, 0) = U0 (x, y) (5)
and boundary condition
lim U(x, y, t) = 0 (6)
|x,y|

where and are arbitrary real values and F(x, t) or F(x, y, t) represents a bounded real valued function or
trapping potential and U = u + iv represents the complex valued wave function. The NLS equation results in two
types of solutions [5]. When the solution, along with spatial derivatives, vanishes at |x| = , it is called a bright
soliton solution, and the solution repeats itself after a specific domain L, it is called an L-periodic solution.
Due to the wide applicability of this equation in various fields of science, various schemes have been developed
for numerical simulation of this equation and researchers are still improving or developing new numerical schemes
to get better results with less computational cost. The solution of the equation was obtained by finite dierence
schemes [8]-[9]. The fully implicit finite dierence schemes along with the Barakat and Clark type explicit formula
was developed by Deghan and Shokri [10]. The authors have also implemented a scheme based on collocation and
radial basis functions to solve this equation numerically. An implicit semi-discrete compact scheme of higher order
was used by Kalita et al. [11]. A split-step fourier pseudospectral scheme was developed by Weideman et al. [12].
For studding the behavior of the singular solutions of the two-dimensional cubic NLS equation, spectral method was
implemented by Sulem [13]. Several schemes was used by Taha and Ablowitz [14] to study the analytical and nu-
merical solutions of the linear and nonlinear one-dimensional NLS equation that includes finite element approaches
and various implicit and explicit finite dierence schemes. Chang et al. [15] used an inverse scattering transform
scheme to find the solution of the generalized NLS equation. Gardner et al. [16] applied various finite element based
schemes for solution of the one dimensional NLS equation. To find numerical solution to NLS equation, a Hermite
function based pseudospectral approach was developed by Muruganandam et al. [17]. Perez-Garcia et al. [18] solved
the NLS equation in one dimension by using several numerical methods based on finite dierence schemes and also
done the comparative study of the schemes regarding accuracy. In last few years, the equation has also been solved
by a time splitting Fourier spectral approximation [19], an explicit finite dierence method [20], an explicit Dufort-
Frankel type scheme [21], Hermite functions based Galerkin spectral scheme[22], spectral collocation method [23],
finite dierence method [24], [25], radial basis functions [26], compact boundary value method [27] and Chebyshev
Spectral collocation method [28].

In recent years, the B-spline based methods gained popularity in solving the linear and nonlinear dierential
equations numerically. The B-spline basis functions have been used to solve many well known linear and nonlinear
partial dierential equations such as Fishers equation [29], Burgers equation [30], and many more well known
equations. B-spline basis functions have been used successfully in Finite Dierence method, Dierential Quadrature
method, and Finite Element method. The use of B-spline basis functions to solve partial dierential equations is
popular because of their properties of continuity, compact support, orthogonality, and in particular due to property of
localization. The B-spline basis functions, used in their classical and modified form to solve various well known partial
dierential equations, are of dierent types such as trigonometric B-Splines, Exponential B-splines and Cardinal B-
splines, each having some distinct properties.
This paper has developed a method based on the dierential quadrature with modified trigonometric cubic B-spline
as basis function (MTCB-DQM) to solve Schrodinger equation (1), with altered primary and boundary conditions.
Additionally, some numerical illustrations with neighbourhood limit conditions and another with distinctive boundary
conditions are examined and the obtained numerical results are observed to be empowering. Thus, the aim of this
work is to explore the practicality of trigonometric B-spline dierential quadrature scheme for the numerical solution
of NLS equation. The developed scheme has also been implemented to solve well known Telegraph equation [31] and
2
117 3

Generalized Burgers-Huxley equation [32]. This work has been accepted and is under publication.
This research article is prearranged as follows: Section 2 describes the trigonometric cubic B-spline basis functions
and the idea to modify the basis function. Section 3 gives the description of the proposed numerical method. Section
4 discusses numerical illustrations to establish the relevance and the accuracy of the given method. In the last, section
5 presents the conclusion of the paper.

2. Numerical scheme

The domain [a, b] [c, d] is divided into the consistently estimated limited components of spans h1 and h2 by the
knots xm1 and ym2 such that a = x0 < x1 < x2 < ... < xm1 1 < xm1 = b and c = y0 < y1 < y2 < ... < ym2 1 < ym2 = d.
Let T Bi (x) and T Bi (y) be trigonometric cubic B-splines with knots at the points xm1 , m1 = 0, 1, 2, ..., N1 and ym2 , m2 =
0, 1, 2, ..., N2 . The arrangement of splines {T B1 , T B0 , T B1 , ..., T BN1 , T BN1 +1 } and {T B1 , T B0 , T B1 , ..., T BN2 , T BN2 +1 }
be the basis functions defined over [a, b] [c, d]. The solution U(x, y, t) at any time on the knot (xi , y j ) is U(xi , y j , t)
for i = 1, 2, ..., N1 and j = 1, 2, ..., N2 . The approximate values of first and second order spatial derivatives are given
by
N
N
Ux = ai j u(x j , y, t), U xx = bi j u(x j , y, t), f or i = 1, 2, ..., N1 (7)
j=1 j=1

N


N

Uy = ai j u(x, y j , t), Uyy = bi j u(x, y j , t), f or i = 1, 2, ..., N2 (8)
j=1 j=1

where ai j , bi j , ai j and bi j are weighting coecients of the first and second order derivatives with respect to x and y
The cubic trigonometric B-spline basis function T Bm1 (x), for m1 = 1, 0, ..., N1 + 1 is characterized as

p3 (xm1 ), if x [xm1 , xm1 +1 )



1

p(xm1 )(p(xm1 )q(xm1 +2 ) + q(xm1 +3 )p(xm1 +1 )) + q(xm1 +4 )p2 (xm1 +1 ), if x [xm1 +1 , xm1 +2 )
T Bm1 (x) = (9)
w


q(xm1 +4 )(p(xm1 +1 )q(xm1 +3 ) + q(xm1 +4 )p(xm1 +2 )) + p(xm1 )q2 (xm1 +3 ), if x [xm1 +2 , xm1 +3 )



q3 (xm +4 ), ifx [xm1 +3 , xm1 +4 )
1

xx xx
where p(xm1 ) = sin ( 2 m1 ), q(xm1 ) = sin ( 2 m1 ), w = sin ( h21 ) sin(h1 ) sin ( 3h21 ) and h1 = ba
n and T Bm1 (x) is piece-
wise cubic trigonometric function with geometric properties like C continuity, non-negativity and partition of unity.
The values of T Bm1 (x), T Bm1 (x) and T Bm1 (x) are given by Table 1. where

sin2 ( h21 ) 2
a1 = , a2 = ,
sin(h1 )sin( 3h21 ) 1 + 2cos(h1 )
3 3
a3 = , a4 = ,
4sin( 3h21 ) 4sin( 3h21 )

3(1 + 3cos(h1 )) 3cos2 ( h21 )

a5 = , a6 = .
16sin2 ( h21 )(2cos( h21 ) + cos( 3h21 )) sin2 ( h21 )(2 + 4cos(h1 ))
The cubic trigonometric B-spline basis functions are modied to result a diagonally dominant matrix system of equa-
tions. The modified cubic trigonometric B-spline basis functions at the knots are defined as follows:

MT B1 (x) = T B1 (x) + 2T B0 (x),

MT B2 (x) = T B2 (x) T B0 (x),
MT B j (x) = T B j (x) f or j = 3, 4, ..., N1 2.
3
117 4

Table 1. The values of trigonometric B-spline basis functions and their derivatives at dierent node points

x xm1 2 xm1 1 xm1 xm1 +1 xm1 +2

T Bm1 0 a1 a2 a1 0

T Bm1 0 a3 0 a4 0

T Bm1 0 a5 a6 a5 0

MT BN1 (x) = T BN1 (x) + 2T BN1 +1 (x),

MT BN1 1 (x) = T BN1 1 (x) T BN1 +1 (x).
where {MT B1 , MT B2 , ..., MT BN1 } form a basis over the region [a, b]. In the similar way we can define the trigono-
metric B-spline basis function in y direction.

3. Computation of the weighting coecients ai j , bi j , ai j and bi j

The first order approximation at the grid point xi , i = 1, 2, ..., N1 is

N
MT Bk (xi ) = ai j T Bk (x j , y) f or k = 1, 2, ..., N1 . (10)
j=1

that results in a matrix system

a [i] =
A

R [i] (11)
where A is the coecient matrix given by


a2 + 2a1 a1 0 0 . 0


0 a2 a1 . . 0



0 a1 a2 a1 . 0
(12)

0 0 a1 a2 a1 0



0 . . . . 0


0 0 . . a1 a2 + 2a1

and a [i] denotes the weighting coecient vector corresponding to grid point x , that is
i
a [i] = [a , a , ..., a ]T
i1 i2 iN1


and the coecient matrix R [i] = [ri1 , ri2 , ..., riN1 ] corresponding to xi , i = 1, 2, ..., N1 , are evaluated as R [1] =
T

[2a4 , a3 a4 , 0, ..., 0]T , R [2] = [a4 , 0, a3 , 0, ..., 0]T , R [3] = [0, a4 , 0, a3 , 0, ..., 0]T ,..., R [N1 1] = [0, ..., 0, a4 , 0, a3 ]T ,

R [N1 ] = [0, ..., 0, a4 a3 , 2a3 ]T . Now, by using designed program for the proposed method on MATLAB, we can
solve the resulting tridiagonal system of equations which provides the vector a [i], that is, the weighting coecients
ai1 , ai2 , ..., aiN1 for i = 1, 2, ..., N1 . Using the coecient ai j , the weighting coecient bi j for i = 1, 2, ..., N1 and
j = 1, 2, ..., N1 as follows
( 1 ) N
bi j = 2ai j aii , f or i , j, and bii = bi j
xi x j i=1,i, j
4
117 5

In similar way one can obtain the ai j and bi j to obtain the approximate first and second order derivatives with respect
to y.
Now the approximate values of first and second spatial derivatives with respect to x as well as y can be calculated
and substituted in equation (1) that results in to a system of ordinary dierential equations using two approaches,
named as Method 1 and Method 2 as described in the next section.

3.1. Method 1
In this approach one can split the equation (1) by expanding in real and imaginary parts. This reduce the equation
(1) in to the system of coupled equations, given as

vt = (u xx + uyy ) + (u2 + v2 )u + F(x, y, t)u

(13)
ut = (v xx + vyy ) + (u2 + v2 )v + F(x, y, t)v
now substituting the values of first and second derivatives from the equation (8), we get

N
N

vt = ( bi j u(x j , y j , t) + bi j u(x j , y j , t)) + (u2 + v2 )u + F(x, y, t)u
j=1 j=1
(14)

N
N

ut = ( bi j v(x j , y j , t) + bi j v(x j , y j , t)) + (u + v )v + F(x, y, t)v
2 2

j=1 j=1

This obtained system is a system of ordinary dierential equations (ODEs) which is solved by a modified form of
Runge kutta method called as strong stability-preserving time-stepping Runge Kutta (SSP-RK43) scheme [33]

3.2. Method 2
In this approach, substituting the approximation of first and second order of the spatial derivatives directly in the
equation (1) results in the following system

N
i Ut = bi j U j + |U j |U j + F(x, t) U j = R(U) (15)
j=1

To solve the resulted system of ODEs we use strong stability-preserving time-stepping RungeKutta (SSP-RK43)
scheme [33] and hence obtained the numerical solutions.

4. Numerical experiment and discussion

In this section we have solved seven test problems of Schrodinger

equation, and also examined the numerical
results. The accuracy and eciency of the developed method is demonstrated by evaluating the discrete average error
and L norm defined as follows

Maximum Error = L = max |ui (Un )i |

i
N
|ui (Un )i |
Average Error = i=0
N
Example 1. For = 1, = 2 and F(x, t) = 0, consider the equation (1) in following form

i Ut (x, t) = U xx (x, t) 2|U(x, t)|2 U(x, t)

with x [1, 1] and t > 0. The initial and the boundary conditions can be obtained from the exact solution given by
[23]
U(x, t) = exp(i(2x 3t))sech(x 4t)
5
117 6

The L error is computed using both of the proposed methods at dierent times t 4.5 with h = 0.02 and t = 0.0001.
Numerical results are compared with Javidi [23] and enlisted in Table 2. Comparison of numerical and the analytical
solution using both of the methods are presented in form of Figures 1 and 2.

Example 2. The NLS equation (1) is solved with = 12 , = 1 and F(x, t) = cos2 (x), that results in

1
i Ut (x, t) = U xx (x, t) + |U(x, t)|2 U(x, t) + cos2 (x)U(x, t)
2
on x [0, 2], t > 0. The initial and the boundary conditions can be obtained from exact solution given by [23]
3it
U(x, t) = sin(x)exp( )
2
L error is computed using both of the methods at dierent times t 20 with N = 100, t = 0.0001, and shown
in Table 3. Comparison of numerical and the analytical solution using both of the methods are presented in form of
Figures 3 and 4.

Example 3. Consider the one soliton solution of the equation (1) with = 1, = 2 and F(x, t) = 0,

i Ut (x, t) = U xx (x, t) + 2|U(x, t)|2 U(x, t)

with x [15, 15] and t > 0. Initial and boundary condition are calculated from the exact solution given by [25]

U(x, t) = exp(i(2x + 4 3t))sech(x + 2 4t)

To show the accuracy of the method, L error is computed at dierent time t 3 using both of the methods with
N = 100 and t = 0.0001, and shown in Table 4. Comparison of numerical and the exact solution using both of the
methods are presented in form of Figures 5 and 6.

Example 4. Consider the collision of two solitons solution of the equation (1) with = 1, = 2 and F(x, t) = 0,
we get
i Ut (x, t) = U xx (x, t) + 2|U(x, t)|2 U(x, t)
with x [15, 15] and t > 0. The initial and boundary conditions for this equation can be obtained from the exact
solution [25]

U(x, t) = exp(i(2x 20 3t))sech(x 10 4t) + exp(i(2x + 20 + 3t))sech(x + 10 + 4t)

In computation L error is computed using both of the methods at time t 3 with N = 100 and t = 0.0001 and
shown in Table 5. Comparison of numerical and the exact solution using both of the methods are presented in form of
Figures 7 and 8.

Example 5. Consider the 2D NLS equation (4) with = 1 and = 0

i Ut (x, y, t) = U xx (x, y, t) + Uyy (x, y, t) + F(x, y)U(x, y, t)

where x, y [0, 1], t > 0 and F(x, y) = 3 2tanh2 (x) 2tanh2 (y). The initial and boundary conditions can be obtained
from the exact solution [27]
iexp(it)
U(x, y, t) =
cosh(x)cosh(y)
In numerical computations maximum or L and average errors are calculated using both of the methods at time t 1
with N = 30,t = 0.0001 and enlisted in Table 6, comparison of the error is given with Mohebbi [27] and Dehghan
[26]. Comparison of numerical and the exact solution are presented in form of Figure 9.

6
117 7

Example 6. By taking = 1 and = 0, 2D form of the NLS equation (4) become

i Ut (x, y, t) = U xx (x, y, t) + Uyy (x, y, t) + F(x, y)U(x, y, t)

where x, y [0, 1], t > 0 and F(x, y) = 1 2

x2
2
y2
. The equation is solved with initial and boundary conditions
obtained from exact solution [27]
U(x, y, t) = x2 y2 exp(it)
To check the accuracy of the proposed methods maximum or L and average errors are calculated using both of
the methods at time t 1 with N = 30, t = 0.0001 and shown in Table 7, comparison of the error is given with
Abdur[28] and Dehghan [26]. Comparison of numerical and the exact solution are presented in form of Figure 10.
Example 7. Consider the 2D NLS equation (4) with = 12 and = 1

1
i Ut (x, y, t) = (U xx (x, y, t) + Uyy (x, y, t)) + F(x, y)U(x, y, t) + |U(x, y, t)|2 U(x, y, t)
2
where x, y [0, 2], t > 0 and F(x, y) = 1 sin2 (x)sin2 (y). Initial and boundary conditions can be obtained from the
exact solution [24]
U(x, y, t) = sin(x)sin(y)exp(i2t)
In numerical computations maximum or L and average errors are calculated using both of the methods at time t 16
with N = 40 and t = 0.0001 and shown in Table 8, comparison of the error is given with Wang [24]. Comparison of
numerical and the exact solution are presented in form of Figure 11.

5. Conclusion

In recent years, due to the existence of soliton solution of NLS equation, numerical simulation has become popular
among researchers. Several distinctive methods were developed and implemented to solve this equation. As per the
authors knowledge, it is for the first time that the modified trigonometric cubic B-spline basis function has been used
with dierential quadrature method to solve NLS equation in both one and two dimensions. To establish the eciency
of the scheme, the approximated results are compared with the exact solutions as well as with numerical solutions
obtained by other researchers. From the obtained discrete average error and L error, it can be inferred that the
obtained numerical results are in great congruity with the analytical solutions. The proposed method is, therefore, an
economical and ecient technique for solving a variety of linear and nonlinear physical models numerically, hence
can be implemented to solve various one and two dimensional partial dierential equations in an ecient way.

7
117 8

1 0.5

Ex. Sol. at t=1

0.8
Num. Sol. at t=1

0.6

0.4 0

0.2
u(x,t)

v(x,t)
0

0.2 0.5

0.4

Ex. Sol. at t=1

0.6
Num. Sol. at t=1
0.8 1
10 5 0 5 10 10 5 0 5 10
x x

Figure 1. The comparison of numerical and the Exact solution of Example 1 for N = 110 at t = 1 with method 1

0.02
Exact Solution at t=1
Numerical Solution at t=1
0
Imaginary Part of U(x,t)

0.02

0.04

0.06

0.08

0.1
0.04 0.02 0 0.02 0.04 0.06
Real Part of U(x,t)

Figure 2. The comparison of numerical and the Exact solution of Example 1 for N = 100 at t = 1 with method 2

8
117 9

0.08 1
Ex. Sol. at t=1
Num. Sol. at t=1 0.8
0.06
0.6
0.04
0.4
0.02
0.2
u(x,t)

v(x,t)
0 0

0.2
0.02
0.4
0.04
0.6
0.06 Ex. Sol. at t=1
0.8
Num. Sol. at t=1
0.08 1
0 2 4 6 8 0 2 4 6 8
x x

Figure 3. The comparison of numerical and the Exact solution of Example 2 for N = 100 at t = 1 with method 1

1
Exact Solution at t=1
0.8 Numerical Solution at t=1

0.6

0.4
Imaginary Part of U(x,t)

0.2

0.2

0.4

0.6

0.8

1
0.08 0.06 0.04 0.02 0 0.02 0.04 0.06 0.08
Real Part of U(x,t)

Figure 4. The comparison of numerical and the Exact solution of Example 2 for N = 100 at t = 1 with method 2

9
117 10

0.6 1
Ex. Sol. at t=2 Ex. Sol. at t=2
0.4 Num. Sol. at =2 0.8 Num. Sol. at =2

0.6
0.2

0.4
0
0.2
u(x,t)

v(x,t)
0.2
0
0.4
0.2

0.6
0.4

0.8 0.6

1 0.8
20 10 0 10 20 20 10 0 10 20
x x

Figure 5. The comparison of numerical and the Exact solution of Example 3 for N = 200 at t = 2 with method 1

0.8

0.6
Imaginary Part U(x,t)

0.4

0.2 Ex. Sol. at t=2

Num. Sol. at t=2
0

0.2

0.4

0.6

0.8
1 0.8 0.6 0.4 0.2 0 0.2 0.4 0.6
Real Part of U(x,t)

Figure 6. The comparison of numerical and the Exact solution of Example 3 for N = 200 at t = 2 with method 2

10
117 11

1 1
Ex. Sol. at t=1 Ex. Sol. at t=1
0.8 Num. Sol. at t=1 Num. Sol. at t=1

0.6

0.4 0.5

0.2
u(x,t)

v(x,t)
0

0.2 0

0.4

0.6

0.8 0.5
20 10 0 10 20 20 10 0 10 20
x x

Figure 7. The comparison of numerical and the Exact solution of Example 4 for N = 300 at t = 1 with method 1

Ex. Sol. at t=1

Num. Sol. at t=1
Imaginary Part of u(x,t)

0.5

0.5
0.8 0.6 0.4 0.2 0 0.2 0.4 0.6 0.8 1
Real Part of U(x,t)

Figure 8. The comparison of numerical and the Exact solution of Example 4 for N = 350 at t = 1 with method 2

11
117 12

Numerical Solution Exact Solution

0.4 0.4
0 0
0.5 0.5
u(x,y,t)

u(x,y,t)
0.5 0.5
0.6 0.6

1 0.7 1 0.7
40 40
40 0.8 40 0.8
20 20 20 20
y 0 0 x y 0 0 x

Numerical Solution Exact Solution

0.8 0.8
0.5 0.5
0.45 0.45
0.6 0.6
v(x,y,t)

v(x,y,t)
0.4 0.4

0.4 0.35 0.4 0.35

0.3 0.3
0.2 0.25 0.2 0.25
40 20 40 40 20 40
0 0 20 0 0 20
y x y x

Figure 9. The Surface plot of real and imaginary parts of the solution for Example 5 at t = 1 and N = 30

Numerical Solution Exact solution

0.5 0.5
1 1
0.4 0.4
u(x,y,t)

u(x,y,t)

0.5 0.3 0.5 0.3

0.2 0.2
0 0
40 0.1 40 0.1
40 40
20 20 20 20
0 0
y 0 0 x y 0 0 x

Numerical Solution Exact Solution

0.8 0.8
1 1
0.6 0.6
v(x,y,t)

v(x,y,t)

0.5 0.5
0.4 0.4

0 0.2 0 0.2
40 40
40 40
20 20 20 20
0 0
y 0 0 x y 0 0 x

Figure 10. The Surface plot of real and imaginary parts of the solution for Example 6 at t = 1 and N = 30

12
117 13

Numerical solution Exact Solution

0.4 0.4
0.5 0.5
0.2 0.2
u(x,y,t)

u(x,y,t)
0 0 0 0

0.5 0.2 0.5 0.2

40 40
40 40
20 20 0.4 20 20 0.4
y 0 0 x y 0 0 x

Numerical solution Exact Solution

1 0.5 1 0.5
v(x,y,t)

v(x,y,t)
0 0 0 0

1 0.5 1 0.5
40 40
40 40
20 20 20 20
y 0 0 x y 0 0 x

Figure 11. The Surface plot of real and imaginary parts of the solution for Example 7 at t = 1 and N = 30

Table 2. Comparison of errors obtained by applying both of the methods in Example 1 for h = 0.02, t = 0.0001 at dierent time levels.

Method 1 Method 2 Javidi[23]

t L for u L for v L2 for U L for U L2 for U

(h=0.125 and
t = 0.0005)

0.5 6.03E-4 3.93E-4 5.28E-4 1.82E-3 7.08E-6

1 3.48E-4 2.90E-4 3.37E-4 3.09E-4

1.5 2.42E-4 4.51E-4 3.54E-4 4.53E-4 7.08E-6

2 3.96E-4 2.96E-4 3.54E-4 4.59E-4

2.5 3.55E-4 4.16E-4 3.54E-4 4.21E-4 7.09E-6

3 3.44E-4 2.21E-4 3.54E-4 4.05E-4

3.5 2.78E-4 3.01E-4 3.53E-4 3.92E-4 7.09E-6

4 3.70E-4 3.09E-4 3.54E-4 4.60E-4

4.5 2.83E-4 3.01E-4 3.54E-4 3.75E-4 7.10E-6

13
117 14

Table 3. Comparison of errors obtained by applying both of the methods in Example 2 for N = 100, t = 0.0001 at dierent time levels.

Method 1 Method 2

t L for u L for v L for U

1 1.43E-4 1.39E-5 1.69E-4

5 1.25E-4 4.40E-5 6.57E-4

10 7.51E-5 8.77E-5 2.17E-3

20 8.02E-5 1.35E-5 8.26E-3

Table 4. Comparison of errors obtained by applying both of the methods in Example 3 for N = 200, t = 0.0001 at dierent time levels.

Method 1 Method 2

t L for u L for v L for U

0.5 2.17E-4 2.54E-4 2.54E-4

1 1.95E-4 1.84E-4 1.97E-4

1.5 1.51E-4 2.31E-4 2.32E-4

2 2.45E-4 3.32E-4 3.40E-4

2.5 3.97E-4 4.63E-4 4.94E-4

3 6.44E-4 5.64E-4 6.68E-4

14
117 15

Table 5. Comparison of errors obtained by applying both of the methods in Example 4 for N = 350, t = 0.0001 at dierent time levels.

Method 1 Method 2

t L for u L for v L for U

0.5 2.72E-4 2.77E-4 2.85E-4

1 2.27E-4 2.47E-4 2.50E-4

1.5 2.06E-4 1.98E-4 2.10E-4

2 4.81E-4 1.83E-4 4.87E-4

2.5 1.92E-3 2.31E-3 2.35E-3

3 2.27E-3 2.40E-3 2.42E-3

Table 6. Comparison of errors obtained by applying both of the methods in Example 5 for N = 30, t = 0.0001 at dierent time levels.

Method 1 Method 2 Mohebbi[27] Dehghan[26]

t L for u L for v Max. Error Average Error L for u L for v L for u L for v

0.1 6.64E-5 1.14E-4 1.32E-4 3.59E-5 4.61E-7 5.08E-7 2.44E-5 2.99E-5

0.3 1.47E-4 1.04E-4 1.55E-4 6.15E-5 4.89E-7 4.25E-7 2.94E-5 2.94E-5

0.5 9.29E-5 1.09E-4 1.19E-4 4.08E-5 4.33E-7 5.30E-7 2.74E-5 3.40E-5

0.7 1.04E-4 7.19E-5 1.10E-4 5.72E-5 5.00E-7 3.03E-7 2.54E-5 1.86E-5

1 1.02E-4 1.52E-4 1.68E-4 5.65E-5 4.08E-7 4.76E-7 2.94E-5 2.42E-5

15
117 16

Table 7. Comparison of errors obtained by applying both of the methods in Example 6 for h = 0.033, t = 0.0001 at dierent time levels.

Method 1 Method 2 Abdur[28] Dehghan[26]

t L for u L for v Max. Error Average Error Max. Error in u Max. Error in v Max. Error in u Max. Error in v

0.1 3.79E-5 5.11E-5 5.55E-5 1.36E-5 5.15E-5 4.68E-5 4.04E-4 3.57E-4

0.3 4.14E-5 3.87E-5 5.17E-5 1.08E-5 6.23E-5 4.16E-5 5.12E-4 8.05E-4

0.5 4.64E-5 5.61E-5 6.78E-5 1.58E-5 5.74E-5 4.06E-5 4.63E-4 3.95E-4

0.7 3.87E-5 4.73E-5 4.81E-5 1.32E-5 4.90E-5 5.27E-5 3.89E-4 4.16E-4

1 4.53E-5 5.01E-5 5.01E-5 1.45E-5 4.83E-5 5.23E-5 3.72E-4 4.12E-4

Table 8. Comparison of errors obtained by applying both of the methods in Example 7 for N = 40, t = 0.0001 at dierent time levels.

Method 1 Method 2 Wang [24]

t L for u L for v Max. Error Average Error Max. Error

4 3.22E-4 1.06E-4 3.34E-4 9.40E-05 8.11E-4

8 2.40E-4 6.74E-4 6.89E-4 2.61E-4 1.62E-3

12 9.65E-4 4.68E-4 1.04E-3 4.34E-4 2.43E-3

16 8.22E-4 1.29E-3 1.53E-3 6.010E-4 3.24E-3

16
117 17

References
[1] A. Arnold, Numerically absorbing boundary conditions for quantum evolution equations, VLSI Design 6 (1998) 313-319.
[2] M. Levy, Parabolic Equation Methods for Electromagnetic Wave Propagations, IEEE, 2000.
[3] F.D. Tappert, The parabolic approximation method, in: J.B. Keller, J.S. Papadaskis (Eds.), Wave Propagation and Underwater Acoustics, in:
Lecture Notes in Physics, Springer, Berlin, 70 (1977) 224-287.
[4] Y.V. Kopylov, A.V. Popov, A.V. Vinogradov, Applications of the parabolic wave equations to X-ray diraction optics, Optics Comm. 118
(1995) 619-636.
[5] W. Huang, C. Xu, S.T. Chu, S.K. Chaudhuri, The finite-dierence vector beam propagation method, J. Lightwave Technol. 10 (3) (1992)
295-304.
[6] F.Y. Hajj, Solution of the Schrdinger equation in two and three dimensions, J. Phys. B 18 (1985) 1-11.
[7] L.Gr. Ixaru, Operations on oscillatory functions, Comput. Phys. Comm. 105 (1997) 1-9.
[8] M. Dehghan, Finite dierence procedures for solving a problem arising in modeling and design of certain optoelectronic devices, Math.
Comput. Simulation 71 (2006) 16-30.
[9] M. Subasi, On the finite-dierence schemes for the numerical solution of two dimensional Schrdinger equation, Numer. Methods Partial
Dierential Equations 18 (2002) 752-758.
[10] M. Dehghan, A. Shokri, A numerical method for two-dimensional Schrdinger equation using collocation and radial basis functions, Comput.
Math. Appl. 54 (2007) 136-146.
[11] J.C. Kalita, P. Chhabra, S. Kumar, A semi-discrete higher order compact scheme for the unsteady two-dimensional Schrdinger equation, J.
Comput. Appl. Math. 197 (2006) 141-149.
[12] J.A.C. Weideman, B.M. Herbst, Split-step methods for the solution of the nonlinear Schrodinger equation, SIAM J.Numer.Anal.23 (3) (1986)
485507.
[13] P.L. Sulem, C. Sulem, A. Patera, Numerical simulation of singular solutions to the twodimensional cubic Schrodinger equation, Commun.Pure
Appl.Math. 37 (1984) 755778.
[14] T.R. Taha, A numerical scheme for the nonlinear Schrodinger equation, Comput.Math.Appl. 22 (9) (1991) 7784.
[15] Q.Chang, E.Jia, W.Sun, Dierence schemes for solving the generalized nonlinear Schrodinger equation, J.Comput.Phys. 148 (1999) 397415.
[16] L.R.T. Gardner, G.A. Gardner, S.I. Zaki, Z. El Sahrawi, B-spline finite element studies of the non-linear Schrodinger equation, Com-
put.Methods Appl.Mech.Eng. 108 (1993) 303318.
[17] P.Muruganandam, S.K Adhikari, BoseEinstein condensation dynamics in three dimensions by the pseudospectral and nite-dierence methods,
J.Phys.B 36 (2003) 25012513.
[18] V.M. Perez-Garcia, X.Y. Liu, Numerical methods for the simulation of trapped nonlinear Schrodinger systems, Appl.Math.Comput. 144 (23)
(2003) 215235.
[19] W.Bao, D.Jaksch, P.A.Markowich, Numerical solution of the GrossPitaevskii equation for BoseEinstein condensation, J.Comput.Phys. 187
(1) (2003) 318342.
[20] M.M. Cerimele, M.L. Chiofalo, F. Pistella, S. Succi, M.P. Tosi, Numerical solution of the GrossPitaevskii equation using an explicit finite-
dierence scheme: an application to trapped BoseEinstein condensates, Phys.Rev.E. 62 (2000) 13821389.
[21] M.C. Lai, C.Y. Huang, a simple DufortFrankel type scheme for the GrossPitaevskii equation of BoseEinstein condensates on dierent ge-
ometries, preprint.
[22] C.M. Dion, E. Cances, Spectral method for the time-dependent GrossPitaevskii equation with a harmonic trap, Phys.Rev.E 67 (2003) 046706-
1046706-9.
[23] M. Javidi, A. Golbabai, Numerical studies on nonlinear Schrdinger equations by spectral collocation method with preconditioning,J. Math.
Anal. Appl. 333 (2007) 1119-1127.
[24] H. Wang, Numerical studies on the split-step finite dierence method for nonlinear Schrodinger equations, Appl. Math. Comput. 170 (2005)
17-35.
[25] Q. Chang, E. Jia and W. Sun, Dierence Schemes for Solving the Generalized Nonlinear Schrodinger Equation, Journal of Computational
Physics 148(1999) 397-415.
[26] M. Dehghan, A. Shokri, A numerical method for two-dimensional Schrdinger equation using collocation and radial basis functions, Comput.
Math. Appl. 54 (2007) 136-146.
[27] A. Mohebbi, M. Dehghan, The use of compact boundary value method for the solution of two-dimensional Schrdinger equation, Journal of
Computational and Applied Mathematics, 225 (2009) 124-134.
[28] R. Abdur,A.I.B.M. Ismail, Numerical Studies on Two Dimensional Schrodinger Equation By Chebyshev Spectral Collocation Method, U.P.B.
Sci. Bull., Series A, Vol. 73 (1) (2011) 101-110.
[29] 3. Mittal R.C. and Arora G., Ecient Numerical solution of Fishers Equation by using B-spline method, International Journal of Computer
Mathematics, 87 (13) (2010) 30393051.
[30] 5. G. Arora and B.K. Singh, Numerical solution of Burgers equation with modified cubic B-spline dierential quadrature method, Applied
Mathematics and Computation 224 (2013) 166177.
[31] G. Arora, V, Joshi, Comparison of Numerical Solution of 1D Hyperbolic Telegraph Equation using B-spline and Trigonometric B-spline by
Dierential Quadrature Method, Accepted for publication in Indian Journal of Science and Technology, (2016).
[32] G. Arora, V, Joshi, Travelling wave solutions of the Generalized BurgersHuxley equation, Accepted for publication in Dierential Equations
and Dynamical Systems, (2016).
[33] J.R. Spiteri, S.J. Ruuth, A new class of optimal high-order strong stability-preserving time-stepping schemes, SIAM J. Numer. Anal. 40 (2)
(2002) 469-491.

17
Dr. Geeta Arora is an Assistant Professor in Department of Mathematics, School of Physical Sciences,
Lovely Professional University, Punjab, India.

Varun Joshi is an Assistant Professor in Department of Mathematics, School of Physical Sciences,

Lovely Professional University, Punjab, India.
Dr. Geeta Arora

Varun Joshi
In this paper, the following are the highlights

1. An accurate numerical solution of the Schrodinger equation is obtained by using trigonometric

differential quadrature method.
2. Implementation of the proposed scheme is done by two different way named as Method 1 and
method 2 successfully.
3. In this article first time ever trigonometric b-spline basis functions are used with differential
quadrature method, and author obtained good results by this experiment.