A GALERKIN METHOD OF BOUNDARY VALUE PROBLEMS )

advertisement
IJMMS 29:6 (2002) 361–369
PII. S0161171202005550
http://ijmms.hindawi.com
© Hindawi Publishing Corp.
A GALERKIN METHOD OF O(h2 ) FOR SINGULAR
BOUNDARY VALUE PROBLEMS
G. K. BEG and M. A. EL-GEBEILY
Received 8 May 2000 and in revised form 2 August 2000
We describe a Galerkin method with special basis functions for a class of singular twopoint boundary value problems. The convergence is shown which is of O(h2 ) for a certain
subclass of the problems.
2000 Mathematics Subject Classification: 65L10.
1. Introduction. We consider the class of singular two-point boundary value problems:
1
− (pu ) + f (x, u) = 0, 0 < x < 1,
p
(1.1)
u(1) = 0.
(pu ) 0+ = 0,
We assume that the real-valued function p satisfies
p ≥ 0,
1
x
p −1 ∈ L1loc (0, 1],
p −1 ∈ L1p (0, 1),
p −1 ∉ L1loc [0, α)
1 1
that is,
0
x
for any α > 0,
1
ds p(x)dx < ∞.
p(s)
(1.2)
(1.3)
Note that (1.3) is clearly satisfied when p is an increasing function on (0, 1). We also
assume that f (x, u) is continuous in u such that for any real u, f (·, u) ∈ L∞
p (0, 1),
q(u, v, x) ≡
f (x, u) − f (x, v)
≥0
u−v
for − ∞ < u, v < ∞, u ≠ v.
(1.4)
The singular two-point boundary value problems of the form (1.1) occur frequently
in many applied problems, for example, in the study of electrohydrodynamics [9], in
the theory of thermal explosions [4], in the separation of variables in partial differential equations [11]; see also [1]. There is a considerable literature on the numerical
methods for the singular boundary value problems. Special finite difference methods were considered in Chawla et al. [5]. The Galerkin method for singular problems
was considered in Ciarlet et al. [6], Eriksson et al. [7], Jesperson [8]. Ciarlet et al. [6]
assumed that p(x) > 0 on (0, 1), p ∈ C 1 (0, 1), and p −1 ∈ L1 (0, 1). In this paper, we
address the problem with p −1 ∉ L1 (0, 1), and we assume that p ≥ 0, p −1 ∈ L1loc (0, 1);
see (1.2) and (1.3). We investigate a Galerkin method with the same special patch functions considered by Ciarlet et al. [6] and we show that the method is of O(h2 ) when
362
G. K. BEG AND M. A. EL-GEBEILY
p is an increasing function on (0, 1). The linear case with more general settings was
considered in [2] and a nonlinear case was considered in [3]. The special case considered here requires a different approach to establish its order of convergence and to
obtain the optimal order of convergence h2 under an easily checked condition on p;
namely that p is increasing on [0, 1].
2. Preliminaries. Let I = (0, 1) and H = L2p (I) denote the weighted Hilbert space
with the inner product
u, vH =
u(x)v(x)p(x)dx.
(2.1)
I
Also let V be the Hilbert space consisting of functions u ∈ L2p (I) which are locally
absolutely continuous on I, u(1) = 0, and u ∈ L2p (I). The inner product on the space
V is defined by
u, vV =
u (x)v (x)p(x)dx.
(2.2)
I
The variational formulation of the problem (1.1) now follows:
Find u ∈ V such that
a(u, v) = 0
∀v ∈ V ,
(2.3)
where
1
a(u, v) ≡ u, vV +
f x, u(x) v(x)p(x)dx.
(2.4)
0
It can be shown [3] that (1.1) and (2.3) have unique absolutely continuous (in [0, 1])
solutions and that the weak solution of (2.3) coincides with the strong solution of (1.1).
3. The Galerkin approximation and convergence results. Let π : 0 = x0 < x1 <
· · · < xN+1 = 1 be a mesh on the interval [0, 1] and, for i = 1, 2, . . . , N, define the patch
functions

−


ri (x) if xi−1 ≤ x ≤ xi ,

ri (x) = ri+ (x) if xi ≤ x ≤ xi+1 ,




0
otherwise,
(3.1)
where
r1− (x) = 1,
x
ri− (x)
=
1/p(s) ds
x
xi−1
,
i
xi−1 1/p(s) ds
xi+1 1/p(s) ds
x
+
,
ri (x) = xi+1 1/p(s) ds
xi
i = 2, 3, . . . , N,
i = 1, 2, . . . , N.
(3.2)
A GALERKIN METHOD OF O(h2 ) FOR SINGULAR BOUNDARY . . .
363
Define the discrete subspace VN of V by
N
VN = span ri i=1 .
(3.3)
The discrete version of the weak problem (2.3) reads:
Find uG ∈ VN such that
a uG , vN = 0 ∀vN ∈ VN .
(3.4)
Note that (3.4) has a unique solution uG ∈ AC[0, 1]. It follows from (2.3) and (3.4)
that
u − uG , vN
1
V
+
0
f (x, u) − f x, uG u − uG vN p = 0.
u − uG
(3.5)
Let q(x)
be the unique function (because u and uG are unique) defined by
 f x, u(x) − f x, uG (x)



,
u(x) − uG (x)
q(x)
≡


0,
u(x) ≠ uG (x)
(3.6)
u(x) = uG (x).
We assume that f is such that
1
Cq :=
0
1
q(x)
x
ds
p(x)dx < ∞.
p(s)
(3.7)
This is the case for example if f satisfies a Lipschitz condition in its second argument
(see (1.3)). We can now state our results on the convergence of the Galerkin solution
uG to the weak solution u of (2.3).
Theorem 3.1. The following relation holds:
G
u − u ≤ 1 + 4Cq f ·, u(·) πN ,
∞
∞
(3.8)
where (πN ) is given by
πN = max
0≤i≤N
xi+1 xi+1
xi
s
1
dt p(s)ds.
p(t)
(3.9)
Corollary 3.2. If p is increasing then the method is O(h2 ) where
h = max xi+1 − xi .
0≤i≤N
(3.10)
Remark 3.3. The absolute continuity of the solution u and the continuity of f
imply that f (·, u(·))
∞ < ∞ in the above expression for the error.
364
G. K. BEG AND M. A. EL-GEBEILY
4. Proof of the results. Let
uG (x) =
N
αi ri (x)
(4.1)
i=1
be the Galerkin approximation and uI be the VN -interpolant of the solution u given
by
uI (x) =
N
ui ri (x),
(4.2)
i=1
where ui = u(xi ) and ri is given by (3.1), i = 1, . . . , N. We note here that uI is the
orthogonal projection of u with respect to the inner product · , ·V :
u − uI , vN
V
=0
(4.3)
for all vN ∈ VN . The following relation is also easily checked (using (3.5) and (4.3))
G
u − uG , vN p ,
u − uI , vN V = q
(4.4)
for all vN ∈ VN . We have the following lemma.
Lemma 4.1. The following relation holds:
u − uI ≤ f ·, u(·) πN .
∞
∞
(4.5)
Proof. For any x ∈ [xi , xi+1 ], i = 0, 1, . . . , N
u(x) − uI (x) ≤
xi+1
g(s)
xi
xi+1
s
dt
p(s)ds,
p(t)
(4.6)
where g(s) = −f (s, u(s)). To see this we consider two cases: i = 0 and i ≥ 1.
For i = 0, that is, for x ∈ [0, x1 ] we have
u(x) − uI (x) = u(x) − u x1
=
=
x1
x
x1
x
x
≤
=
1
p(s)
ds
p(s)
s
g(t)p(t)dt
0
x
0
g(s)p(s)
0
x1
0
x1
s
g(s)
x1
s
x1
x1
dt
ds
p(t)
x1
x1
dt
dt
g(s)p(s)
ds +
ds
p(t)
p(t)
x
s
g(s)p(s)ds +
g(s)p(s)
x
dt
p(s)ds.
p(t)
s
(4.7)
A GALERKIN METHOD OF O(h2 ) FOR SINGULAR BOUNDARY . . .
It can be shown, using the fact
[xi , xi+1 ], i = 1, . . . , N,
N
i=1 ri (x)
365
= 1 and integrating by parts, that for x ∈
u(x) − uI (x)
xi+1 xi+1
dt
dt
−
g(s)p(s)ds + ri+1 (x)
g(s)p(s)ds
=
p(t)
xi
xi p(t)
x
s
xi+1
s
ds/p(s) x
x
= xi+1
dt/p(t) g(s)p(s)ds
ds/p(s) xi
xi
xi
x
xi+1 xi+1
dt
xi ds/p(s)
+ xi+1
g(s)p(s)ds
p(t)
ds/p(s)
x
s
x
ri+ (x)
≤
≤
=
i
xi+1
x
x
x s
ds
p(s)
g(s)p(s)ds +
x
xi
g(s)p(s)
xi+1
xi
s
xi+1
g(s)
xi
xi+1
s
dt
ds +
p(t)
xi+1 xi+1
x
s
xi+1 xi+1
x
s
dt g(s)p(s)ds
p(t)
dt g(s)p(s)ds
p(t)
dt
p(s)ds
p(t)
(4.8)
The result thus follows.
Proof of Theorem 3.1. In (4.4) taking vN = ri for i = 1, . . . , N, we obtain
uG − uI , ri
V
u − uG , ri p ,
= q
(4.9)
which can be written as
N
j , ri p αj − uj = q
u − uI , ri p .
rj , ri V + qr
(4.10)
j=1
This gives the system
(A + Q )e = d,
(4.11)
where A = (aij ) = (ri , rj V ) is a symmetric and tridiagonal matrix given by
1
,
(1/p(s))ds
x1
a11 = x2
1
1
+ xi+1
,
(1/p(s))ds
(1/p(s))ds
xi−1
xi
aii = xi
ai,i+1 = − xi+1
xi
1
,
(1/p(s))ds
i = 1, . . . , N − 1,
i = 2, . . . , N,
(4.12)
366
G. K. BEG AND M. A. EL-GEBEILY
j , ri p ), e = (ei ) = (αi − ui ), and d = (di ) is given by
Q =(qij ) = (qr
x2
x2
x1
x h(s)p(s) s (dt/p(t))ds
x2
h(s)p(s)ds + 1
d1 =
x0
x1 dt/p(t)
s
x
xi+1
xi
h(s)p(s) s i+1 (dt/p(t))ds
xi−1 h(s)p(s) xi−1 (dt/p(t))ds
xi
+
,
di =
xi
xi+1
dt/p(t)
xi−1 dt/p(t)
xi
(4.13)
i > 1,
where h(s) stands for q(s)(u(s)
− uI (s)). Now A is an M-matrix, qij ≥ 0 (see (1.4)),
qij < −aij (i ≠ j) for sufficiently small mesh size and therefore, A + Q is an M-matrix
with (A + Q )−1 ≤ A−1 (see Ortega [10]). Thus |e| ≤ A−1 |d|. The inverse of the matrix
A, denoted by B = (bij ), can be explicitly written as
 1

ds


if i ≤ j,


 xj p(s)
bij = (4.14)

1

ds



if
i
≥
j.

xi p(s)
Therefore,
N
e i ≤
bij dj j=1
=
≤
i 1
N 1
ds ds dj +
dj xi p(s)
xj p(s)
j=1
j=i+1
N 1
j=1
xj
(4.15)
ds dj .
p(s)
We see that
x2
x2 x1
1
1
1
ds ds
ds x1 h(s)p(s) s (dt/p(t))ds
d1 ≤
h(s)p(s)ds +
x2
x1 p(s)
x1 p(s) x0
x1 p(s)
x1 dt/p(t)
x2 x2
1
x1
x2
ds
ds x1 h(s)p(s) s (dt/p(t))ds
=
h(s) p(s)ds +
x2
x1 p(s) x0
x1 p(s)
x1 dt/p(t)
x2
x2 1
ds x1 h(s)p(s) s (dt/p(t))ds
x2
+
x2 p(s)
x1 dt/p(t)
x1
x2
x2
1
ds
dt
h(s)p(s)ds +
h(s)p(s)
ds
≤
p(t)
x1
s
x1 p(s) x0
x2
1
ds
h(s)p(s)ds
+
x2 p(s) x1
x1
x2
1
1
ds
dt
h(s)p(s)ds +
h(s)p(s)
ds
=
x1 p(s) x0
s p(t)
x1
x2
1
1
x1
dt
dt
h(s)p(s)
h(s)p(s)
ds +
ds.
≤
s p(t)
s p(t)
x0
x1
(4.16)
A GALERKIN METHOD OF O(h2 ) FOR SINGULAR BOUNDARY . . .
367
Also for j = 2, . . . , N, by a similar approach, we have
1
xj
1
ds dj ≤
p(s)
ds
p(s)
xj
1
+
≤
xj
xj
xj
h(s)p(s)ds
xj−1
ds
p(s)
xi+1 h(s)p(s) xi+1 dt/p(t) ds
xi
s
xi+1
dt/p(t)
xi
h(s)p(s)
xj−1
1
s
dt
ds +
p(t)
xj+1
(4.17)
h(s)p(s)
xj
1
s
dt
ds.
p(t)
Substituting these two inequalities in (4.15) we obtain
e i ≤
xN
h(s)p(s)
1
x0
1
≤2
h(s)p(s)
0
1
=2
s
1
0
s
dt
ds +
p(t)
xN+1
h(s)p(s)
x1
1
s
dt
ds
p(t)
dt
ds
p(t)
(4.18)
q(s)
u(s) − uI (s) p(s)
1
s
dt
ds.
p(t)
Thus using (3.7), we have
max αi − ui ≤ 2Cq u − uI ∞ .
(4.19)
G
u − uI ≤ 2 max αi − ui .
∞
(4.20)
u − uG ≤ u − uI + uG − uI ∞
∞
∞
I
≤ u − u ∞ + 2 max ui − αi (4.21)
1≤i≤N
It can be shown that
1≤i≤N
Therefore,
1≤i≤N
≤ 1 + 4Cq u − uI ∞ .
The result thus follows from Lemma 4.1.
5. Example. In this section we give examples which are solved by the Galerkin
method just described above with equal mesh size h. We then compare the results
with the actual solutions.
Example 5.1. We consider the boundary value problem
−
1
xu + eu = 0,
x
0 < x < 1,
u (0) = u(1) = 0.
(5.1)
368
G. K. BEG AND M. A. EL-GEBEILY
√
The exact solution is known: u(x) = 2 ln((1 + β)/(1 + βx 2 )), β = −5 + 2 6. It is seen
that uG − u
∞ = 0.188845 × 10−2 for h = 0.1 and uG − u
∞ = 0.189 × 10−4 for
h = 0.01. According to the Corollary 3.2 the method is O(h2 ) which is reflected in
these results.
Example 5.2. We consider the equation
−
β2 x 2β−2 u β(α + β − 1)x β−2
1 α e =
x u + α
x
5 4 + xβ
4 + xβ
α + x u 0 = 0,
u(1) = 0.
(5.2)
The exact solution is u = ln 5 − ln(4 + x β ). The following results were obtained:
Table 5.1
α
β
h
uG − u
∞
0.5
2
0.02
1.0299×10−4
0.5
2
0.01
2.6147×10−5
1.0
2
0.02
9.9647×10−5
1.0
2
0.01
2.4913×10−5
2.0
6
0.02
3.4133×10−4
2.0
6
0.01
8.6170×10−5
Remark 5.3. Our method does not differentiate between 0 < α < 1 and α ≥ 1 as is
the case in many articles in the literature.
Acknowledgement. The authors acknowledge the excellent research facilities
available at King Fahd University of Petroleum and Minerals, Saudi Arabia.
References
[1]
[2]
[3]
[4]
[5]
[6]
[7]
[8]
[9]
W. F. Ames, Nonlinear Ordinary Differential Equations in Transport Process, Mathematics
in Science and Engineering, vol. 42, Academic Press, New York, 1968.
G. K. Beg and M. A. El-Gebeily, A Galerkin method for singular two point linear boundary
value problems, Arab. J. Sci. Eng. Sect. C Theme Issues 22 (1997), no. 2, 79–98.
, A Galerkin method for nonlinear singular two point boundary value problems,
Arab. J. Sci. Eng. Sect. A Sci. 26 (2001), no. 2, 155–165.
P. L. Chambre, On the solution of the Poisson-Boltzman equation with the application to
the theory of thermal explosions, J. Chem. Phys 20 (1952), 1795–1797.
M. M. Chawla, S. McKee, and G. Shaw, Order h2 method for a singular two-point boundary
value problem, BIT 26 (1986), no. 3, 318–326.
P. G. Ciarlet, F. Natterer, and R. S. Varga, Numerical methods of high-order accuracy for
singular nonlinear boundary value problems, Numer. Math. 15 (1970), 87–99.
K. Eriksson and V. Thomée, Galerkin methods for singular boundary value problems in
one space dimension, Math. Comp. 42 (1984), no. 166, 345–367.
D. Jespersen, Ritz-Galerkin methods for singular boundary value problems, SIAM J. Numer. Anal. 15 (1978), no. 4, 813–834.
J. B. Keller, Electrohydrodynamics. I. The equilibrium of a charged gas in a container, J.
Rational Mech. Anal. 5 (1956), 715–724.
A GALERKIN METHOD OF O(h2 ) FOR SINGULAR BOUNDARY . . .
[10]
[11]
369
J. M. Ortega and W. C. Rheinboldt, Iterative Solution of Nonlinear Equations in Several
Variables, Academic Press, New York, 1970.
S. V. Parter, Numerical methods for generalized axially symmetric potentials, J. Soc. Indust.
Appl. Math. Ser. B Numer. Anal. 2 (1965), 500–516.
G. K. Beg and M. A. El-Gebeily: Department of Mathematical Sciences, King Fahd University of Petroleum and Minerals, Dhahran 31261, Saudi Arabia
Download