Applied Mathematics and Computation 174 (2006) 180–187

www.elsevier.com/locate/amc

Cubic splines method for solving

fourth-order obstacle problems
Eisa A. Al-Said a, Muhammad Aslam Noor b,
Themistocles M. Rassias c,*
a
Department of Mathematics, College of Science, King Saud University, P.O. Box 2455,
Riyadh 11451, Saudi Arabia
b
Mathematics Department, COMSATS Institute of Information Technology,
Islamabad, Pakistan
c
Department of Mathematics, National Technical University of Athens,
Zografou Campus, GR-15780 Athens, Greece

Abstract

In this paper, we develop a new cubic spline method for computing approximate
solution of a system of fourth-order boundary value problems associated with obstacle,
unilateral and contact problems. It is shown that the present method is of order two and
gives approximations which are better than those produced by some other collocation,
finite difference and spline methods. Numerical examples are presented to illustrate the
applicability of the new method.

2005 Elsevier Inc. All rights reserved.

Keywords: Variational inequalities; System of differential equations; Convergence; Cubic splines;

Obstacle problems

*
Corresponding author.
E-mail addresses: eisasaid@ksu.edu.sa (E.A. Al-Said), noormaslam@hotmail.com, noor@
mscs.dal.ca (M.A. Noor), trassias@math.ntua.gr (T.M. Rassias).

0096-3003/$ - see front matter
2005 Elsevier Inc. All rights reserved.
doi:10.1016/j.amc.2005.03.022
 E.A. Al-Said et al. / Appl. Math. Comput. 174 (2006) 180–187 181

1. Introduction

It is well-known that a wide class of unrelated problems arising industrial,

regional, economics, telecommunication, computer network, image reconstruc-
tion, pure and applied sciences can be studied in the general and unified frame-
work of Variational inequalities, see [2–6,9–13]. We note that the finite
difference methods including splines techniques cannot be applied directly to
find the approximate solutions of variational inequalities. However. if the
obstacle function is known, then one can be characterized the variational
inequalities associated with the fourth-order obstacle problems by a system of
fourth-order differential equations using the penalty function technique. The
penalty function technique is due to Lewy and Stampacchia [7], which they used
to study the regularity of the solution, whereas Noor and his coworkers used
this technique to solve the system of differential equations numerically. The
main computational advantage of this technique is its simple applicability for
solving system of differential equations. Despite their importance, little atten-
tion has been given to develop numerical methods for solving such type of sys-
tems of differential equations associated with obstacle, free and moving
problems. In recent years, Al-Said and Noor [2,3], Khalifa and Noor [5] and
Noor and Al-Said [11,12] have used such types of penalty function in solving
a class of contact problems in elasticity in conjunction with collocation, finite
difference and spline techniques.
For the purpose of numerical experience, we consider an example of an elas-
tic beam lying over an elastic obstacle. The formulation and the approximation
of the elastic beam is very simple. However, it should be pointed out that the
kind of numerical problems which occur for more complicated system will be
the same. To convey an idea, we consider a system of fourth-order boundary
value problem of the type
8
< f ðxÞ;
> a 6 x 6 c;
ðivÞ
u ¼ gðxÞuðxÞ þ f ðxÞ þ r; c 6 x 6 d; ð1:1aÞ
>
:
f ðxÞ; d6x6b
with the boundary and continuity conditions
uðaÞ ¼ uðbÞ ¼ a1 ; u00 ðaÞ ¼ u00 ðbÞ ¼ a2 ;
ð1:1bÞ
uðcÞ ¼ uðdÞ ¼ b1 ; u00 ðcÞ ¼ u00 ðdÞ ¼ b2 ;

where f and g are continuous functions on [a, b] and [c, d], respectively. The
parameters r, ai, i = 1,2 and bi, i = 1, 2 are real constants. The possibility of
using collocation method with quintic spline as a basis functions for solving
(1.1) was discussed by Khalifa and Noor [5]. Their numerical results indicated
that the quintic spline collocation method produced second order approxima-
tions. After this, Al-Said and Noor [2] used a second order finite difference
182 E.A. Al-Said et al. / Appl. Math. Comput. 174 (2006) 180–187

method to compute second order approximations for the solution of (1.1). For
related results, see [3,4,11–13] and the references therein. In the present paper,
we develop a new cubic spline method for solving the boundary value problem
(1.1) over the whole interval [a, b]. In Section 2, we derive the numerical method
and briefly discuss its error analysis. Section 3 is devoted to the numerical
experiments and comparison with other methods.

2. Cubic spline method

For simplicity, we first develop the cubic spline method for solving the
fourth-order boundary value problem
uðivÞ ¼ gðxÞu þ f ðxÞ;
uðcÞ ¼ uðdÞ ¼ b1 ; ð2:1Þ
u00 ðcÞ ¼ u00 ðdÞ ¼ b2
then we use it to solve (1.1) in Section 3. For this purpose we divide the interval
[c, d ] into n equal subintervals using the grid points xi = c + ih, i = 0, 1, 2, . . . , n,
x0 = c, xn+1 = d and h ¼ d
c nþ1
, where n is a positive integer.
Consider the problem of constructing a cubic spline S(x) satisfying the inter-
ð2Þ
polation conditions Sðxi
12 Þ ¼ uðxi
12 Þ, for i = 1, 2, . . . , n. Also, let M i
12 ¼ S i
1 for
2
i = 1, 2, . . . , n. Now for the differential equation in (2.1) at the midknots xi
12 we
may have
ð4Þ
M i
32
2M i
12 þ M iþ12 ¼ h2 ui
1 þ Oðh6 Þ. ð2:2Þ
2

The parameters of cubic spline S(x) satisfy the consistency relation

24 h i
M i
32 þ 22M i
12 þ M iþ12 ¼ 2 S i
32
2S i
12 þ S iþ12 ð2:3Þ
h
for i = 2, 3, . . . , n
1, see Al-Said [1].
It follows from (2.2) and (2.3) that
1h i h2
ð4Þ
M i
12 ¼ 2 ui
32
2ui
12 þ uiþ12
ui
1 þ Oðh6 Þ ð2:4Þ
h 24 2

for i = 2, 3, . . . , n
1.
Now from (2.3) and (2.4) we have the consistency relations
h4 h ð4Þ ð4Þ ð4Þ
i
ui
52
4ui
32 þ 6ui
12
4uiþ12 þ uiþ32 ¼ ui
3 þ 22ui
1 þ uiþ1 þ ti ð2:5Þ
24 2 2 2

for i = 3, 4, . . . , n
2.
Eq. (2.5) form a linear system of n
4 equations in the n unknowns ui
12 ,
i = 1, 2, . . . , n. So we need four more equations two at each end of the integra-
 E.A. Al-Said et al. / Appl. Math. Comput. 174 (2006) 180–187 183

tion range. Using quartic spline functions we have the following relations, see
[15,8]:
9
10ui
12
5uiþ12 þ uiþ32 ¼ 6ui
1
54 h2 u00i
1 >
>
h i >
>
>
>
1 4 ð4Þ ð4Þ ð4Þ
þ 384 h 154ui
1 þ 75uiþ1 þ uiþ3 ; for i ¼ 1; >
>
>
>
2 2 2
>
>
1

5ui
32 þ 6ui
12
4uiþ12 þ uiþ32 ¼
2ui
2
4 h ui
2 2 >
>
>
>
h i >
>
>
þ 384 h 75ui
3 þ 230ui
1 þ 76uiþ1 þ uiþ3 ; for i ¼ 2; >
4 ð4Þ ð4Þ ð4Þ ð4Þ
1 =
2 2 2 2
ð2:6Þ
ui
52
4ui
32 þ 6ui
12
5uiþ12 ¼
2uiþ1
14 h2 u00iþ1 >
>
>
>
>
>
þ 384 h ½ui
5 þ 76ui
3 þ 230ui
1 þ 75uiþ1 ; for i ¼ n
1; >
4 ð4Þ ð4Þ ð4Þ ð4Þ
1 >
>
2 2 2 2 >
>
>
>
5 2 00
ui
52
5ui
32 þ 10ui
12 ¼ 6ui
4 h ui >
>
>
>
h i >
>
1 4 ð4Þ ð4Þ ð4Þ >
;
þ 384 h ui
5 þ 75ui
3 þ 154ui
1 ; for i ¼ n.
2 2 2

The local truncation errors ti related to the consistency relations (2.5), and
(2.6) are given by
8
>
41 6 ð6Þ
>
> h ui
1 þ Oðh7 Þ for i ¼ 1
>
> 2304
>
>
>
>
473 6 ð6Þ
>
> h ui
2 þ Oðh7 Þ for i ¼ 2
>
> 11520
>
<
1 6 ð6Þ
ti ¼ h ui þ Oðh7 Þ for 3 6 i 6 n
2 ð2:7Þ
>
> 8
>
>
>
>
473 6 ð6Þ
>
> h uiþ1 þ Oðh7 Þ for i ¼ n
1
>
> 11520
>
>
>
:
41 h6 uið6Þ þ Oðh7 Þ
>
for i ¼ n.
2304
Now, using a standard convergence analysis, see, for example Usmani [15],
it can be shown that our method is a second order convergent process, that is
kek  O(h2), where e = (ei) is the discretisation error.

3. Applications and numerical results

To illustrate the application of the spline method developed in the previous

sections, we consider the fourth-order obstacle boundary value problem of
finding u such that
9
uðivÞ P f ðxÞ; on X ¼ ½
1; 1 >
>
>
u P wðxÞ; on X ¼ ½
1; 1 =
; ð3:1Þ
½uðivÞ
f ðxÞ½u
wðxÞ ¼ 0 on X ¼ ½
1; 1 >
>
>
;
uð
1Þ ¼ uð1Þ ¼ 0; u00 ð
1Þ ¼ u00 ð1Þ ¼ 0;
184 E.A. Al-Said et al. / Appl. Math. Comput. 174 (2006) 180–187

where f is a given force acting on the beam and w(x) is the elastic obstacle. Eq.
(3.1) describes the equilibrium configuration of an elastic beam, pulled at the
ends and lying over an elastic obstacle. We study problem (3.1) in the frame-
work of variational inequality approach. To do so, we define the set K as
K ¼ fv :2 H 20 ðXÞ : v P w on Xg ð3:2Þ
which is a closed convex set in H 20 ðXÞ, where H 20 ðXÞ is a Sobolev space, which is
in fact a Hilbert space. It can be easily shown that the energy functional asso-
ciated with the obstacle problem (3.1) is
Z

d4 v
1
I½v ¼ 4

2f ðxÞ vðxÞ dx; for all v 2 H 20 ðXÞ

1 dx
Z 1  2 2 Z 1
dv
¼ 2
dx
2 f ðxÞvðxÞ dx

1 dx
1

¼ aðv; vÞ
2hf ; vi; ð3:3Þ

where
Z   
1
d2 u d2 v
aðu; vÞ ¼ dx ð3:4Þ

1 dx2 dx2
and
Z 1
h f ; vi ¼ f ðxÞvðxÞ dx. ð3:5Þ

1

It can be easily proved that the form a(u, v) defined by (3.4) is bilinear, sym-
metric and positive (in fact, coercive) and the functional f defined by (3.5) is a
linear continuous functional. It is well known [6,9,10] that the minimum u of
the functional I [v] defined by (3.3) on the closed convex set K in H 20 ðXÞ can
be characterized by the variational inequality

aðu; v
uÞ P hf ; v
ui for all v 2 K. ð3:6Þ

Thus we conclude that the obstacle problem (3.1) is equivalent to solving the
variational inequality problem (3.6). This equivalence has been used to study
the existence of a unique solution of (3.1), see [6,9,10]. Now using the idea
of Lewy and Stampacchia [7], the problem (3.6) can be written as
uðivÞ þ mfu
wgðu
wÞ ¼ f ; ð3:7Þ
where w is the obstacle function and m(t) is a penalty function defined by


4; tP0
mðtÞ ¼ ð3:8Þ
0; t < 0.
 E.A. Al-Said et al. / Appl. Math. Comput. 174 (2006) 180–187 185

We assume that the obstacle function w(x) is defined by

(

1 for
1 6 x 6
12 and 12 6 x 6 1;
wðxÞ ¼ 1 4 ð3:9Þ
4
for
12 6 x 6 12 .

From (3.7)–(3.9), we obtain the following system of equations:

(
ðivÞ
f; for
1 6 x 6
12 and 12 6 x 6 1;
u ¼ ð3:10Þ
1
4u þ f ; for
12 6 x 6 12

with the boundary conditions

   
1 1
uð
1Þ ¼ u
¼u ¼ uð1Þ ¼ 0;
2 2
    ð3:11Þ
1 1
u00 ð
1Þ ¼ u00
¼ u00 ¼ u00 ð1Þ ¼ 0
2 2
and the conditions of continuity of u and u00 at x ¼
12 and 12.

Example 4.1. In this example, we consider the system of differential equation

(3.10) when f = 0, namely,
(
ðivÞ
0 for
1 6 x 6
12 and 12 6 x 6 1;
u ¼ ð3:12Þ
1
4u for
12 6 x 6 12

with the boundary conditions

   
1 1
uð
1Þ ¼ u
¼u ¼ uð1Þ ¼ 0;
2 2
    ð3:13Þ
1 1
u00 ð
1Þ ¼
u00
¼ u00 ¼
u00 ð1Þ ¼
;
2 2
where
! 0. The analytical solution for this boundary value problem is
8  2 3 3 2 13 1

<
3 x
2 x
12 x
4

> for
1 6 x 6
12 ;
uðxÞ ¼ 0.5
/13 ½/1 sin x sinh x þ /2 cos x cosh x for
12 6 x 6 12 ;
>
:  2 3 3 2 13 

3 x þ 2 x
12 x þ 14
for 1
2
6 x 6 1;
ð3:14Þ
where /1 ¼ sin 12 sinh 12, /2 ¼ cos 12 cosh 12 and /3 ¼ cos 1 þ cosh 1.

Example 4.2. For f = 1, the problem (3.10) becomes

(
ðivÞ
1 for
1 6 x 6
12 and 12 6 x 6 1;
u ¼ ð3:15Þ
2
4u for
12 6 x 6 12
186 E.A. Al-Said et al. / Appl. Math. Comput. 174 (2006) 180–187

Table 1
Observed maximum errors for Example 4.1 with
= 10
6
h Our method Central diff. [2] Quar-spline [3] Colloc-quintic [5] Cub-spline [14]
1
8 7.2 · 10
6 1.4 · 10
4 1.3 · 10
5 3.0 · 10
4 1.9 · 10
5
1
16 2.2 · 10
6 3.6 · 10
5 3.2 · 10
6 7.0 · 10
5 4.8 · 10
6
1
32 5.7 · 10
7 8.9 · 10
6 8.1 · 10
7 1.4 · 10
5 1.3 · 10
6

Table 2
Observed maximum errors for Example 4.2
h Our method Central diff. [2] Quar-spline [3] Cub-spline [14]

6
5
5
1
12 7.8 · 10 6.2 · 10 1.2 · 10 8.4 · 10
6
1
24 1.9 · 10
6 1.6 · 10
5 2.8 · 10
6 2.2 · 10
6
1
48 4.9 · 10
7 3.9 · 10
6 6.9 · 10
7 5.4 · 10
7

with the boundary conditions (3.11). The analytical solution for this boundary
value problem is
81 4 1 3 1 2 3 1
< 24 x þ 8 x þ 8 x þ 64 x þ 192
> for
1 6 x 6
12 ;
uðxÞ ¼ 0.5
/13 ½/1 sin x sinh x þ /2 cos x cosh x for
12 6 x 6 12 ;
>
:1 4 1 3 1 2 3 1
24
x
8 x þ 8 x
64 x þ 192 for 12 6 x 6 1;
ð3:16Þ
where /1, /2 and /3 are as defined in Example 4.1.
The problems (3.12) and (3.15) were solved over the whole interval [
1, 1]
using the spline method developed in Section 2 with a variety of h and
values.
The observed maximum errors in absolute values are given in Tables 1 and 2.
From the these tables we can notice that if the stepsize h is reduced by a fac-
tor 1/2, then the errors are approximately reduced by a factor 1/4. Thus, the
numerical results confirm that our present method is a second-order conver-
gent process as predicted in Section 3. We also used the cubic spline method
developed by Papamichael and Worsey [14] to solve the problems and the
numerical results are also listed in Tables 1 and 2. These problems were also
solved by Al-Said and Noor [2,3] using finite difference and quartic spline
methods and problem (3.12) was solved by Khalifa and Noor [5] using collo-
cation method with quintic B-spline as basis functions. Some of their results
are also given in Tables 1 and 2. From these tables we may notice that our pres-
ent method gives better results than the others.

References

[1] E.A. Al-Said, Cubic spline method for solving two-point boundary value problems, Korean J.
Comput. Appl. Math. 5 (1998) 669–680.
 E.A. Al-Said et al. / Appl. Math. Comput. 174 (2006) 180–187 187

[2] E.A. Al-Said, M.A. Noor, Computational methods for fourth-order obstacle boundary value
problems, Comm. Appl. Nonlinear Anal. 2 (1995) 73–83.
[3] E.A. Al-Said, M.A. Noor, Quartic spline method for solving fourth-order obstacle boundary
value problems, J. Comput. Appl. Math. 143 (2002) 107–116.
[4] E.A. Al-Said, M.A. Noor, D. Kaya, K. Al-Khalid, Finite difference method for solving
fourth-order obstacle problems, Int. J. Comput. Math. 81 (2004) 741–748.
[5] A.K. Khalifa, M.A. Noor, Quintic splines solutions of a class of contact problems, Math.
Comput. Modell. 13 (1990) 51–58.
[6] N. Kikuchi, J.T. Oden, Contact Problems in Elasticity, SIAM Publishing Co., Philadelphia,
1988.
[7] H. Lewy, G. Stampacchia, On the regularity of the solutions of the variational inequalities,
Comm. Pure Appl. Math. 22 (1969) 153–188.
[8] G. Micula, S. Micula, Handbook of Splines, Kluwer Academic Publishers, Dordrecht,
Holland, 1999.
[9] M. Aslam Noor, Some recent advances in variational inequalities Part I. Basic concepts, New
Zealand J. Math. 26 (1997) 53–80.
[10] M. Aslam Noor, Some recent advances in variational inequalities, Part II, Other concepts,
New Zealand J. Math. 26 (1997) 229–255.
[11] M.A. Noor, E.A. Al-Said, Fourth-order obstacle problems, in: T.M. Rassias, H.M. Srivastava
(Eds.), Analytic and Geometric Inequalities and Applications, Kluwer Academic Publishers,
Dordrecht, Holland, 1999, pp. 277–300.
[12] M.A. Noor, E.A. Al-Said, Numerical solutions of fourth-order variational inequalities, Int. J.
Comput. Math. 75 (2000) 107–116.
[13] M.A. Noor, S.I. Tirmizi, Numerical methods for unilateral problems, J. Comput. Appl. Math.
16 (1986) 387–395.
[14] N. Papamichael, A.J. Worsey, A cubic spline method for the solution of linear fourth-order
two-point boundary value problem, J. Comput. Appl. Math. 7 (1981) 187–189.
[15] R.A. Usmani, The use of quartic splines in the numerical solution of fourth-order boundary
value problem, J. Comput. Appl. Math. 44 (1992) 187–199.