EUROPEAN JOURNAL OF PURE AND APPLIED MATHEMATICS

Vol. 2, No. 2, 2009, (268-277)

ISSN 1307-5543 – www.ejpam.com

A Novel Power Series Method for Solving Second Or-

der Partial Differential Equations

Muhammet Kurulay1 and Mustafa Bayram2∗

1
Yildiz Technical University, Faculty of Art and Sciences, Department of Mathematics,
34210-Davutpasa-İstanbul
2
Fatih University, Faculty of Art and Sciences, Department of Mathematics,
34500-Buyukcekmece-İstanbul/Turkey

Abstract. In this article, a new approach is proposed to solve partial differential equation.
This method is based on generalized Taylor’s formula. The solution of partial differential
equations can be expanded using MAPLE. Doing some simple mathematical operations on
these equations, we can get a closed form series solution or approximate solution quickly.
PDE problems with constant and variable coefficients are solved by the present method. With
this method, we can reach same results simpler way than other analytical or approximate
methods.

Key words: Second order partial differential equations; Power Series.

∗
Corresponding author.

Email addresses: muhammetkurulayyahoo. om (M. Kurulay), mbayramfatih.edu.tr (M.

Bayram)

http://www.ejpam.com 268
c 2009 EJPAM All rights reserved.
M. Kurulay and M. Bayram / Eur. J. Pure Appl. Math, 2 (2009), (268-277) 269

1. Introduction

Study about partial differential algebraic equation were done by Marszalek. Marsza-
lek studied analysis of the partial differential algebraic equations [1]. Lucht ve Strehmel
[2, 3] studied numerical solution and indexes of the linear partial differential equa-
tions with constant coefficients. A study about characteristics analysis and differential
index of the partial differential algebraic equations were given by Martinson and Bar-
ton [5, 6]. Debrabant and Strehmel investigated convergence of the Runge-Kutta
Method for linear partial differential algebnraic equations [4].
Chen and Ho [7] proposed methods to Solving partial differential equations by
two-dimensional differential transform method. The method was applied to the
partial differential equation [8–10].Second-order linear Partial differential equations
problems were solved by Yang, Liu and Bai [11]. Chebyshev polynomial solutions of
second-order linear partial differential equations [12].
In this paper, we propose, by making full use of the properties of power series, a
generral scheme for solving the second-order partial differential equation

∂ 2u ∂ 2u ∂ 2u ∂u
a(x, y) 2
+ b(x, y) + c(x, y) 2
+ d(x, y)
∂x ∂ x∂ y ∂y ∂x
∂u
+e(x, y) + f (x, y)u + g(x, y) = 0 (1.1)
∂y

with the initial conditions

∂
u(x, 0) = p(x), u(x, 0) = q(xπ) (1.2)
∂y
Finally, three PDE problems are solved by the present method, and the calculated
results are compared very well with those obtained by other analytical or approximate
methods.
M. Kurulay and M. Bayram / Eur. J. Pure Appl. Math, 2 (2009), (268-277) 270

2. Preliminary Knowledge

The basic theory of power series is stated below. In what follows, we assume that
for each function involved in our study, all its derivatives are existent and continuous
in the region of interest.
The Taylor series may also be generalized to functions of more than one variable
with
∞ ∞
X X ∂n ∂ nd f (a1 , ..., ad )
T (x 1 , ..., x d ) = ... n ... (x 1 − a1 )n1 ...(x d − ad )nd (2.1)
n1 =0 nd =0
∂ x d1 ∂ x dnd n1,..., nd

For example, for a function that depends on two variables, x and y, the Taylor series
to second order about the point (a, b) is:

f (x, y) = f (a, b) + f x (a, b)(x − a) + f y (a, b)( y − b)

1 ” —
= f x x (a, b)(x − a)2 + 2 f x y (a, b)(x − a)( y − b) + f y y (a, b)( y − b)2 + . . . (2.2)
2!

where the subscripts denote the respective partial derivatives.

Definition 1. The coefficient of variables of a function w(x, y) is defined as
– k+h ™
1 ∂ w(x, y)
W (k, h) = , (2.3)
k!h! ∂ x k∂ y h x=0
y=0

where it is be noted that upper case symbol W (k, h) is used to denote the coeffi-
cients of variables in (2.2) which represented by a corresponding lower case symbol
w(x, y).
Definition 2. The inverse of W (k, h) is defined as

∞ X
X ∞
w(x, y) = W (k, h)x k y h . (2.4)
k=0 h=0

From Equations (2.3) and (2.4), we obtain

M. Kurulay and M. Bayram / Eur. J. Pure Appl. Math, 2 (2009), (268-277) 271

∞ X
∞
– ™
X x k yh ∂ k+h w(x, y)
w(x, y) = (2.5)
k=0 h=0
k!h! ∂ x k∂ y h x=0
y=0

Theorem 1. If w(x, y) = u(x, y) ± v(x, y), then W (k, h) = U(k, h) ± V (k, h).
Proof. By definition 1 we have

– ™
1 ∂ k+h
U(k, h) = u(x, y) (2.6)
k!h! ∂ x k∂ y h x=0
y=0
– ™
1 ∂ k+h
V (k, h) = v(x, y) (2.7)
k!h! ∂ x k∂ y h x=0
y=0
– ™
1 ∂ k+h
W (k, h) = u(x, y) ± v(x, y) (2.8)
k!h! ∂ x k∂ y h x=0
y=0

Using equations (2.6)-(2.8), we have

W (k, h) = U(k, h) ± V (k, h) (2.9)

Theorem 2. If w(x, y) = λu(x, y), then W (k, h) = λU(k, h).

Proof. By definition 1 we have
– ™
1 ∂ k+h
U(k, h) = u(x, y) (2.10)
k!h! ∂ x k∂ y h x=0
y=0

– ™
1 ∂ k+h
W (k, h) = λw(x, y) (2.11)
k!h! ∂ x k∂ y h x=0
y=0

Using equations (2.10)-(2.11), we have

W (k, h) = λU(k, h). (2.12)

M. Kurulay and M. Bayram / Eur. J. Pure Appl. Math, 2 (2009), (268-277) 272

3. Numerical Solution of Second-Order Partial Differential

Equations

This section aim at describing a numerical solution of partial differential equations

by power series. We write power series in the form

w(x, y) = W (0, 0) + W (1, 0)x + W (0, 1) y + W (1, 1)x y + ... + ax m y n (3.1)

where W (0, 0), W (1, 0), W (0, 1), W (1, 1)... are known constants but is unknown con-
stant. Substituting (3.1) into (1.1), we can get the following:

W (m, n) = (µa + λ)x m y n−i = 0 (3.2)

where µ and λ are constant and is order of partial differential equation. From
(3.2), we have a constant. Substituting (3.2) into (3.1), we get solution arbitrary or-
der of partial differential equation. Repeating this procedure we can get the arbitrary
order power series of the solution for PDEs in (1.1).

4. Applications

Example 1.
The test problem consider the fallowing partial differential equation [7],
∂ 2 u(x.t) ∂ 2 u(x.t)
− c2 =0 (4.1)
∂ t2 ∂ x2
with initial condition

u(x, 0) = x 3 (4.2)

∂ u(x, t)
= x. (4.3)
∂t
M. Kurulay and M. Bayram / Eur. J. Pure Appl. Math, 2 (2009), (268-277) 273

Using equation (2.4), and initial condition equation (4.2),we obtain

U(i, 0) = 0, i = 0, 1, 2, 4, ..., m,

U(3, 0) = 1. (4.4)

Using equation (2.4), and initial condition equation (4.3),we obtain

U(i, 1) = 0, i = 0, 2, ..., n,

U(1, 1) = 1. (4.5)

Substituting equations (4.4)-(4.5) into equations (2.4), we have

u1 (x, t) = x t + x 3 + ax t 2 (4.6)

Substituting equations (4.6) into equations (4.1), and by recursive method,the

results corresponding to m → ∞, n → ∞ are listed as follows

U(1, 2) = 3c 2 (4.7)

and the others are zero.

We obtain the closed form series solution as follows.

u(x, t) = x t + x 3 + 3c 2 x t 2 (4.8)

By analitik method [7] ,a closed form solution is obtained as follows.

uˆ (x, t) = x t + x 3 + 3c 2 x t 2 (4.9)

From last two equations, we have u(x, t) = uˆ (x, t).

Example 2. We now consider the problem [12]
M. Kurulay and M. Bayram / Eur. J. Pure Appl. Math, 2 (2009), (268-277) 274

ut t = ux x + 6 (4.10)

With initial condition

u(x, 0) = x 2 ,

u t (x, 0) = 4x. (4.11)

Using equation (2.4), and initial condition equation (4.11),we obtain

U(i, 0) = 0, i = 0, 1, 3, 4, ..., m,

U(2, 0) = 1 (4.12)

Using equation (2.4), and initial condition equation (4.12),we obtain

U(i, 1) = 0, i = 0, 2, ..., n,

U(1, 1) = 4 (4.13)

Substituting equations (4.12)-(4.13) into equations (2.4), we have

u1 (x, t) = 4x t + x 2 + at 2 (4.14)

Substituting equations(4.14) into equations (4.10), we get

the results corresponding to m → ∞, n → ∞ are listed as follows

U(0, 2) = 4 (4.15)

and the others are zero.

M. Kurulay and M. Bayram / Eur. J. Pure Appl. Math, 2 (2009), (268-277) 275

Substituting all U(k, h) into equation (2.4),we obtain the closed form series solu-
tion as follows.
u(x, t) = 4x t + x 2 + 4t 2 (4.16)

Again, we have obtained the exact solution of the initial value problem stated
previously.

Example 3. [7]

∂ 2 u(x.t) ∂ 2 u(x.t)
− − x 2 u(x, t) = x (4.17)
∂ t2 ∂ x2
with initial condition

u(x, 0) = 0,
∂ u(x, 0)
= 0. (4.18)
∂t

Using equation (2.4) and initial conditions(4.18) , we have

U(i, 0) = 0, i = 0, 1, 2, ..., m. (4.19)

U(i, 1) = 0, i = 0, 1, 2, ..., n. (4.20)

the results corresponding to m → ∞, n → ∞ are listed as follows

1
U(1, 2) = ,
2
1
U(3, 4) = ,
24
1
U(1, 6) =
120

and the others are zero.

REFERENCES 276

Substituting all U(k, h) into equation(2.4) ,we obtain

x t2 x3t4 x t6
u(x, t) = + + . (4.21)
2 24 120
In this case, we have obtained the exact solution of the targeted equation with the
specified initial conditions

5. Conclusion

Analytic solutions of the second-order linear partial differential equations with

variable coefficients are usually difficult. In many cases, it is required toapproxi-
mate solutions. For this purpose, power series method can be proposed.In this study,
the usefulness power series method presented for the approximate solution of the
second-order linear partial differential equations is discussed. Also, the method can
be applied to the non-homogeneous (Example 2) and homogeneous (Examples 1 )
cases. This method is very simple an effective for most of linear partial differential
equations.

References

[1] Marszalek, W. (1997), Analysis of partial differential algebraic equations, Ph.D. Thesis,
North Carolina State University, Raleigh.
[2] Lucht, W. , Strehmel, K. , Eichler-Liebenow, C. (1997a), “Linear partial differential al-
gebraic equations, Part I: Indexes, consistent boundary/initial conditions”, Report 17,
Fachbereich Mathematik und Informatik, Martin-Luther-Universitat Halle.
[3] Lucht, W. , Strehmel, K. , Eichler-Liebenow, C. (1997b), “Linear partial differential
algebraic equations, Part II: Numerical solution”, Report 18, Fachbereich Mathematik
und Informatik, Martin-Luther-Universitat Halle.
REFERENCES 277

[4] Debrabant, K. ve Strehmel K. (2005), “Convergence of Runge-Kutta methods ap-

plied to linear partial differential-algebraic equations”, Applied Numerical Mathematics,
53:213-229.
[5] Martinson, W. , Barton, P. (2000), “A differentiation index for partial differential-
algebraic equations”, SIAM J. Sci. Comput., Vol. 21, 6:2295-2315.
[6] Martinson, W. , Barton, P. (2002), “Index and characteristic analysis of linear PDAE
systems”, SIAM J. Sci. Comput., Vol. 24, 3:905-923.
[7] Chen, C.K. , Ho, S.H. (1999), “Solving partial differential equations by two-dimensional
differential transform method”, Applied Mathematics and Computation,106:171-179.
[8] Hassan, I.H. (2002), “On solving some eigenvalue problems by using a differential
transformation”, Applied Mathematics and Computation, 127:1-22.
[9] Jang, M.J. , Chen, C.L. , Liy, L.C. (2000), “On solving the initial-value problems using the
differential transformation method”, Applied Mathematics and Computation, 115:145-
160.
[10] Liu, H. and Song, Y. (2007), “Differential transform method applied to high index
differential-algebraic equations”, Applied Mathematics and Computation, 184:748-753.
[11] X Yang, Y. Liu, S. Bai, (2006). “ A numerical solution of second-order linear partial
differential equations by differential transform”, Applied Mathematics and Computation
173 ;792-802.
[12] C.Kesan, (2003) ,“ Chebyshev polynomial solutions of second-order linear partial dif-
ferential equations”, Applied Mathematics and Computation 134 ;109–124.
[13] Bayram, M., Çelik, E., (2003), “Arbitrary Order Numerical Method for Solving
Differential- Algebraic Equations by Padé Series”, Applied Mathematics and Compu-
tation. 137;57-65.