Applied Mathematics and Computation 217 (2010) 34253428

Contents lists available at ScienceDirect

Applied Mathematics and Computation

journal homepage: www.elsevier.com/locate/amc

Application of reproducing kernel method to third order three-point boundary value problems
Boying Wu 1, Xiuying Li
Department of Mathematics, Harbin Institute of Technology, Harbin, Heilongjiang 150001, PR China

a r t i c l e

i n f o

a b s t r a c t
This paper investigates the analytical approximate solutions of third order three-point boundary value problems using reproducing kernel method. The solution obtained by using the method takes the form of a convergent series with easily computable components. However, the reproducing kernel method can not be used directly to solve third order three-point boundary value problems, since there is no method of obtaining reproducing kernel satisfying three-point boundary conditions. This paper presents a method for solving reproducing kernel satisfying three-point boundary conditions so that reproducing kernel method can be used to solve third order three-point boundary value problems. Results of numerical examples demonstrate that the method is quite accurate and efcient for singular second order three-point boundary value problems. 2010 Elsevier Inc. All rights reserved.

Keywords: Three-point boundary value problem Reproducing kernel method

1. Introduction In this paper, we consider the following third order three-point boundary value problems:

&

Lux  u000 x axu00 x bxu0 x cxux f x; u0 0; ua 0; u1 0;

0 6 x 6 1;

1:1

where a 2 (0, 1), a(x), b(x), c(x), f(x) 2 C[0, 1]. We consider u(0) = 0, u(a) = 0, u(1) = 0 since the boundary conditions u(0) = c0, u(a) = c1, u(1) = c2 can be reduced to u(0) = 0, u(a) = 0, u(1) = 0. Multi-point boundary value problems arise in a variety of applied mathematics and physics. Two-point boundary value problems have been extensively studied in the literature. Also, the existence and multiplicity of solutions of multi-point boundary value problems have been studied by many authors, see [18] and the references therein. However, research for methods for solving multi-point boundary value problems has proceeded very slowly. To the best of our knowledge, there have been no methods for solving multi-point boundary value problems. In this paper, we will apply reproducing kernel method (RKM) presented by Cui and Geng [913], Cui and Lin [14], Cui and Chen [15] and Li and Cui [16] to solve three-point boundary value problem (1.1). The rest of the paper is organized as follows. In the next section, the reproducing kernel satisfying three-point boundary conditions is constructed. The RKM is applied to (1.1) in Section 3. The numerical examples are presented in Section 4. Section 5 ends this paper with a brief conclusion.
Corresponding author.
E-mail addresses: mathwby@hit.edu.cn (B. Wu), xyli1112@sina.com (X. Li). Project supported by the National Natural Science Foundation of China (Grant No. 10771044), the Natural Science Foundation of Heilongjiang (Grant No. A06-05).
1

0096-3003/$ - see front matter 2010 Elsevier Inc. All rights reserved. doi:10.1016/j.amc.2010.09.009

3426

B. Wu, X. Li / Applied Mathematics and Computation 217 (2010) 34253428

2. Construction of RK satisfying three-point boundary conditions By using the methods in [916], it is impossible to obtain reproducing kernel (RK) satisfying three-point boundary conditions of (1.1). In this section, we will make great efforts to ll this gap. First, we construct a reproducing kernel space W4[0, 1] in which every function satises u(0) = u(1) = 0. W4[0, 1] is dened as W4[0, 1] = {u(x)ju(x), u0 (x), u00 (x), u000 (x) are absolutely continuous real value functions, u0000 (x) 2 L2[0, 1], u(0) = 0, u(1) = 0}. The inner product and norm in W4[0, 1] are given, respectively by

uy; v yW 4 u0v 0 u1v 1 u0 0v 0 0 u0 1v 0 1

and

1 0

0000

0000

dy

kukW 4

q u; uW 4 ; u; v 2 W 4 0; 1:

By [916], it is easy to obtain the following theorem. Theorem 2.1. The space W4[0, 1] is a reproducing kernel Hilbert space. That is, there exists k(x, y) 2 W4[0, 1], for any u(y) 2 W4[0, 1] and each xed x 2 [0, 1], y 2 [0, 1], such that uy; kx; yW 4 ux. The reproducing kernel k(x, y) can be denoted by

kx; y

&

hx; y; y 6 x; hy; x; y > x;

2:1

where hx; y 5040 1 xy x5 11 5yy y6 5x4 y2 y x6 y3 2y 10x3 y1 3yx5040 10080y 5040y2 1 5 6 2 7y y x 5040 15132y 10102y2 7y5 2y6 . Next, we construct a reproducing kernel space W 4 0; 1 in which every function satises u(0) = u(a) = u(1) = 0. a W 4 0; 1 is dened as W 4 0; 1 fuxjux; u0 x; u00 x; u000 x are absolutely continuous real value functions, a a u0000 (x) 2 L2[0, 1], u(0) = 0, u(a) = 0, u(1) = 0}. Clearly, W 4 0; 1 is a closed subspace of W4[0, 1]. a The following theorem give the reproducing kernel of W 4 0; 1. a Theorem 2.2. The reproducing kernel ka of W 4 0; 1 is given by a

ka x; y kx; y ka; a1 kx; aka; y:

Proof. Clearly, not all elements of W4[0, 1] vanish at a. This implies that k(a, a) 0. It is easy to see that ka(x, a) = ka(a, y) = 0, and therefore ka x; y 2 W 4 0; 1. a For 8 uy 2 W 4 0; 1, obviously, u(a) = 0, it follows that a

2:2

uy; ka x; yW 4 uy; kx; yW 4 ka; a1 kx; aua ux:

That is, ka(x, y) is of reproducing property. Thus, ka(x, y) is the reproducing kernel of W 4 0; 1 and the proof is complete. h a 3. Application of RKM to (1.1)  In (1.1), it is clear that L : W 4 0; 1 ! W 1 0; 1 is a bounded linear operator. Put ui x kxi ; x and wi(x) = L*ui(x) where a 2  ; x is the RK of W 1 0; 1, L* is the adjoint operator of L. The orthonormal system fw xg1 of W 4 0; 1 can be derived from i kxi i1 2 a GramSchmidt orthogonalization process of fwi xg1 , i1

 wi x

i X k1

bik wk x bii > 0; i 1; 2; . . .:

3:3

By RKM presented in [916], we have the following theorem. Theorem 3.1. For (1.1), if fxi g1 is dense on [0, 1], then fwi xg1 is the complete system of W 2 0; 1 and wi x Ly ka x; yjyxi . i1 i1 2 Theorem 3.2. If fxi g1 is dense on [0, 1] and the solution of (1.1) is unique, then the solution of (1.1) is i1

ux

1 i XX i1 k1

 bik f xk wi x:

3:4

 Put Q Span fwi gn . Clearly, Q & W 4 0; 1. Denotes by Q\ the orthocomplement space of Q in W 4 0; 1, i.e. a a i1 Q fuxjux; Q 0; ux 2 W 4 0; 1g and dene the null space of L by NL fuxjLux 0; ux 2 W 4 0; 1g. It is obvia a ously N(L) = Q\.
?

B. Wu, X. Li / Applied Mathematics and Computation 217 (2010) 34253428

3427

Theorem 3.3. u(x) is the minimal norm solution of (1.1) in W 4 0; 1. a Proof. From Theorem 3.1, one obtains u(x) is the solution of (1.1). Then all solution of system (1.2) is

wx ux w0 x;
where w0(x) 2 Q . Notice that kwxkW 4 kuxkW 4 kw0 xkW 4 . It is easy to see that u(x) is the minimal norm solution of (1.1) and the proof is complete. h
\

3:5

Now, the approximate solution un can be obtained by taking nitely many terms in the series representation of u(x) and

un x

n i XX i1 k1

bik f xk wi x:

4. Numerical examples In this section, two numerical examples are studied to demonstrate the accuracy of the present method. The examples are computed using Mathematica 5.0. Results obtained by the method are compared with the analytical solution of each example and are found to be in good agreement with each other. Example 4.1. Consider the following third order three-point boundary value problem

x p 2 u000 x e2 u00 x 2 sin xex ux f x; 1 u0 0; u 2 0; u1 0;

0 6 x 6 1;

p x where f x 6 e2 3 3x ex x1 3; x 2; x2 sin x. The exact solution is given by ux x x 1 x 1. Using our 2 2 i1 method, taking xi n1 ; i 1; 2; . . . ; n; n 11; 51, the absolute errors jun(x) u(x)j between the approximate solution and exact solution are given in Fig. 1.

Absolute error 610-7 510-7 410-7 3 10-7 210-7 110-7 0.2 0.4 0.6 0.8 1 x

Absolute error 2.510-8 210-8 1.510-8 110-8 510-9 0.2 0.4 0.6 0.8 1 x

Fig. 1. Absolute errors for Example 4.1 (n = 11, n = 51). (The left: n = 11; the right: n = 51.)

Relative error 810-6 610-6 410-6 210-6 0.2 0.4 0.6 0.8 1 x

Relative error 410-7 310-7 210-7 110-7 0.2 0.4 0.6 0.8 1 x

Fig. 2. Relative errors for Example 4.2, (n = 11, n = 51). (The left: n = 11; the right: n = 51.)

3428

B. Wu, X. Li / Applied Mathematics and Computation 217 (2010) 34253428

Example 4.2. Consider the following third order three-point boundary value problem

x p 2 u000 x e2 u00 x 2 sin xex ux f x;

0 6 x 6 1;

u0 0;
x x

u1 3

e ;
x

1 9

1 3

u1 1 e;

p p where f x e e 2 e ; x 2ex ex x2 ; sin x. The exact solution is given by u(x) = x2 + ex. Using our method, taking     i1 xi n1 ; i 1; 2; . . . ; n; n 11; 51, the relative errors un xux between the approximate solution and exact solution are giux ven in Fig. 2.

5. Conclusion In this paper, we apply RKM to third order three-point boundary value problems and obtain approximate solutions with a high degree of accuracy. Therefore, RKM is an accurate and reliable analytical technique for third order three-point boundary value problems. Acknowledgments The authors would like to express their thanks to unknown referees for their careful reading and helpful comments. References
[1] R.P. Agarwal, H.B. Thompson, C.C. Tisdell, Three-point boundary value problems for second-order discrete equations, Computer and Mathematics with Applications 45 (2003) 14291435. [2] R.P. Agarwal, I. Kiguradze, On multi-point boundary value problems for linear ordinary differential equations with singularities, Journal of Mathematical Analysis and Applications 297 (2004) 131151. [3] H.B. Thompson, C. Tisdell, Three-point boundary value problems for second-order, ordinary, differential equation, Mathematical and Computer Modelling 34 (2001) 311318. [4] A. Ya. Lepin, V.D. Ponomarev, On a positive solution of a three-point boundary value problem, Differential Equations 42 (2) (2006) 291293. [5] Y.P. Sun, Positive solutions for third-order three-point nonhomogeneous boundary value problems, Applied Mathematics Letters 22 (1) (2009) 4551. [6] Y.P. Sun, Existence of triple positive solutions for a third-order three-point boundary value problem, Journal of Computational and Applied Mathematics 221 (1) (2008) 194201. [7] V.R.G. Moorti, J.B. Garner, Existence and uniqueness theorems for three-point boundary value problems for third order differential equations, Journal of Mathematical Analysis and Applications 70 (2) (1979) 370385. [8] D.R. Anderson, G. Smyrlis, Solvability for a third-order three-point BVP on time scales, Mathematical and Computer Modelling (2009), doi:10.1016/ j.mcm.2008.11.009. [9] Minggen Cui, Fazhan Geng, Solving singular two-point boundary value problem in reproducing kernel space, Journal of Computational and Applied Mathematics 205 (2007) 615. [10] Fazhan Geng, Minggen Cui, Solving singular nonlinear second-order periodic boundary value problems in the reproducing kernel space, Applied Mathematics and Computation 192 (2007) 389398. [11] Fazhan Geng, Minggen Cui, Solving singular nonlinear two-point boundary value problems in the reproducing kernel space, Journal of the Korean Mathematical Society 45 (3) (2008) 7787. [12] Fazhan Geng, Minggen Cui, Solving a nonlinear system of second order boundary value problems, Journal of Mathematical Analysis and Applications 327 (2007) 11671181. [13] Minggen Cui, Fazhan Geng, A computational method for solving one-dimensional variable-coefcient Burgers equation, Applied Mathematics and Computation 188 (2007) 13891401. [14] Minggen Cui, Yingzhen Lin, A new method of solving the coefcient inverse problem of differential equation, Science in China Series A 50 (4) (2007) 561572. [15] Minggen Cui, Zhong Chen, The exact solution of nonlinear age-structured population model, Nonlinear Analysis: Real World Applications 8 (2007) 10961112. [16] Chunli Li, Minggen Cui, The exact solution for solving a class nonlinear operator equations in the reproducing kernel space, Applied Mathematics and Computation 143 (23) (2003) 393399.