Positive Solutions For A Second-Order, Singular Semipositone Dynamic Boundary Value Problem

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 11

178

Int. J. Dynamical Systems and Differential Equations, Vol. 3, Nos. 1/2, 2011

Positive solutions for a second-order, singular


semipositone dynamic boundary value problem
Rajendra Dahal
Department of Mathematics and Statistics,
Coastal Carolina University,
Conway, SC 29528, USA
E-mail: rdahal@coastal.edu
Abstract: We obtain a sufcient condition for the existence of a positive
solution for a second-order singular semipositone dynamic boundary
value problem by constructing a special cone. As a special case when
T = R, this result includes those of Liu (2007). This result is new in a
general time scale setting. One example is given at the end of this paper
to demonstrate the result.
Keywords: semipositone; cone; time scale; delta and nabla derivatives.
Reference to this paper should be made as follows: Dahal, R. (2011)
Positive solutions for a second-order, singular semipositone dynamic
boundary value problem, Int. J. Dynamical Systems and Differential
Equations, Vol. 3, Nos. 1/2, pp.178188.
Biographical notes: Rajendra Dahal is an Assistant Professor of
Mathematics and Statistics at Coastal Carolina University, Conway, SC,
USA. He received his PhD Degree in Mathematics under the supervision
of Professors Lynn Erbe and Allan Peterson from the University of
Nebraska, Lincoln, NE, USA in August 2009. His research interest lies
in dynamic equations on time scales.

1 Introduction
We consider the following Boundary Value Problem (BVP)

Lx = f (t, x) + g(t, x), t ((a), (b))T


1 x((a)) 1 x[] ((a)) = 0,

2 x((b)) + 2 x[] (b) = 0,

(1.1)

where the operator L is dened by Lx := (p(t)x ) , f (t, x) is continuous, g(t, x)


is Lebesgue -integrable, p > 0 is -differentiable, 1 , 2 0, 1 , 2 > 0, with
1 + 2 > 0, and


(b)

:= 1 2 + 2 1 + 1 2
(a)

1
s > 0.
p(s)

Copyright 2011 Inderscience Enterprises Ltd.

179

Positive solutions for a second-order

Note that x[] (t) := p(t)x (t) is called the quasi -derivative. The domain D of L
is the set of functions x : T R such that x is continuous on [(a), (b)]T , x is
continuous on [(a), b]T , and (p(t)x ) is continuous on [a, b]T . Let a and b be such
that 0 (a) a < a + 1 b < with ((a), (b))T contains at least two points.
Note that f and (p(t)x ) continuous imply g is continuous, and it may change
sign in above BVP. We call this type of problem is semipositone. Semipositone
problems arise in many physical and chemical processes such as in chemical reactor
theory. In applications one is interested in nding positive solutions.
For convenience let us denote c, d, c(t), d(t), e(t) by


(b)

c = 1 + 1
(a)

1
s, d = 2 + 2
p(s)

c(t) = 1 + 1
(a)

(b)

(a)

1
s, d(t) = 2 + 2
p(s)


t

1
s
p(s)
(b)

1
s
p(s)

1
e(t) = c(t) d(t).

(1.2)

Then we have
2 c(t) + 1 d(t) = .
We impose the following conditions throughout this paper:
(H1 )

For any (t, x) ((a), (b))T (0, ), f (t, x) 0, g(t, x) > q(t),
where q(t) 0 is Lebesgue -integrable on ((a), (b))T .

(H2 )

Let f (t, x) p1 (t)q1 (x), g(t, x) p2 (t)q2 (x), where


p1 : ((a), (b))T (0, ), p2 : ((a), (b))T R, q1 : (0, )
b
(0, ), q2 : (0, ) R are continuous with (a) e(s)q(s)s < ,
b
and 0 < (a) e(s)[p1 (s) + p2 (s) + q(s)]s < .
b
There exists r such that r > 2((b) (a)) (a) q(s)s and

(H3 )

r
>
max0|x|r {q1 (x), |q2 (x)|, 1}
(H4 )

e(s) [p1 (s) + p2 (s) + q(s)] s.

(1.3)

(a)

Choose t1 [a, b/2)T and t2 (b/2, b]T so that


f (t, x)
1
> M , where M 1 =
inf
lim x
x
2
(a)<t<(b)

t2

G(t, s)s.

min

t1 <t<t2

t1

2 Preliminary lemmas
We state the following lemma which we will use later in this section.
Lemma 2.1 (Guo and Sun, 1998): Let K be a cone in the real Banach space X,
K K be a completely
be a bounded, open subset of X with 0 , and A :
continuous operator.

180

R. Dahal

(i)

Suppose that Au = u, for all u K, 1. Then i(A, K, K) = 1.

(ii)

K K which is completely continuous


Suppose that there exists B :
such that (a) inf uK ||Bu|| > 0, and
(b) u Au = Bu, 0, u K. Then i(A, K, K) = 0.

Let
X := {x C([(a), (b)]T , R)} with
x
=

sup

|x(t)|,

t[(a),(b)]T

and dene
K := {x X : x(t) e(t)||x||, t [(a), (b)]T },
e(t)
, and e(t) is as dened in (1.2).
where e(t) := (b)(a)
Then one can easily verify that X is a real Banach space, and K is a cone in X.

Greens function: The Greens function G(t, s) for the BVP





= 0, t [(a), (b)]T ,
p(t)x
1 x((a)) 1 x[] ((a)) = 0,

2 x((b)) + 2 x[] (b) = 0,


is given by
1
G(t, s) :=


c(t)d(s), for (a) t s (b);
d(t)c(s), for (a) s t (b);

where , c(t), d(t) are as dened before.


Proof: Recall that the Cauchy function for (p(t)x ) = 0 is given by

x(t, s) =
s

1
.
p( )

Then the Greens function can be written as



u(t, s),
for t s;
G(t, s) =
u(t, s) + x(t, s), for t s;
where, for xed s, u(t, s) is the unique solution of the BVP

(p(t)u (t, s)) = 0,


1 u((a), s) 1 p((a))u ((a), s) = 0,

2 u((b), s) + 2 p(b)u (b, s) = 2 x((b), s) 2 p(b)x (b, s).


Note that all the derivatives are taken with respect to t, and s t.

(2.1)

181

Positive solutions for a second-order


Solving (p(t)u (t, s)) = 0, we get


u(t, s) = c1 (s) + c2 (s)


(a)

1
.
p( )

so that
u (t, s) = c2 (s)

1
.
p(t)

Now by the boundary conditions, we get


1 c1 (s) 1 c2 (s) = 0


 (b)
(b)
1
1
+ 2 c2 (s) = 2
+ 2 .
2 c1 (s) + 2
p( )
(a) p( )
s
If 1 = 0, we get from the former that c1 (s) =
second equation above, we get

1
1 c2 (s).

Substituting this into the


 (b)
1
1
1 2
+ 1 2
c2 (s) =

p( )
s


 (b)
1
1
c1 (s) =
1 2
+ 1 2 .

p( )
s
This leads to


(a) p( )



 t
(b)
1
1
1
=
1 + 1

2
+ 2

p( )
(a) p( )
s

u(t, s) = c1 (s) + c2 (s)

1
c(t)d(s).

Therefore, for t s, G(t, s) =

c(t)d(s).
t 1
For t s, G(t, s) = x(t, s) + u(t, s) = s p(
) + u(t, s).
 (b) 1
Recall = 1 2 + 2 1 + 1 2 (a) p(s) s. Multiplying and dividing the rst
term of the right side by , we get


 s
 (b)
1
1
1
1 2
+ 1 2
+ 1 2 + 1 2 H ,
G(t, s) =

p( )
(a) p( )
t
where H =

 (b)
s

1
p( )

t

(a) p( )

 (b)

(a) p( )

t

1
.
s p( )

182

R. Dahal
Note that
 (b)
 t
1
1
1

H=
p( )
s
(a) p( )
s
s p( )
 s
 t
 (b)
 t
1
1
1
1

p(
)
p(
)
p(
)
p(
)
(a)
s
s
s
 s
 (b)
1
1
=

.
p(
)
p(
)
(a)
t


(b)

p( )

Thus for t s, we have

 s
 (b)
1
1
1
1 2
+ 1 2
+ 1 2
G(t, s)=

p(
)
p(
)
(a)
t

 (b)
 s
1
1

+1 2
p( )
(a) p( )
t



 s
 (b)
1
1
1
=
1 + 1

2 + 2

p( )
(a) p( )
t
=

1
c(s)d(t).

If 1 = 0, we get c2 (s) = 0. We have




(b)

2 c1 (s) = 2


s
(b)

c1 (s) =
s

1
+ 2
p( )

1
2
.
+
p( )
2

Thus,


u(t, s) = c1 (s) + c2 (s)




(a)
(b)

=
s

1
=
1 2

p( )

2
1
+
p( )
2


1 2 + 2
s

(b)

p( )

1
= c(t)d(s).

And,


 (b)
 t
1
1
1
u(t, s) + x(t, s) =
1 2 + 1 2
1 2

p( )
s
s p( )

183

Positive solutions for a second-order


 (b)
1
1

= 1 2 + 2

p( )
t
=

1
c(s)d(t).

Note that
0 G(t, s) G(s, s) =

1
c(s)d(s) = e(s).

(2.2)

Now we set


w(t) :=

(2.3)

G(t, s)q(s)s.
(a)

Let us dene an operator A : K X by






G(t, s) f (s, [u w]+ (s)) + g(s, [u w]+ (s)) + q(s) s. (2.4)

(Au)(t) =
(a)

Lemma 2.2: Assume (H1 ) and (H2 ) hold. Then A : K K is a completely


continuous operator.
Proof: For any x K, and t, t0 , s [(a), (b)]T , we get
c(t)
for t, t0 s;

c(t0 ) ,

c(t)d(s)

d(t0 )c(s) , for t s t0 ;

G(t, s)
=
G(t0 , s)

d(t)

d(t0 ) ,

d(t)c(s)

c(t0 )d(s) ,

for t, t0 s;
for t s t0 ;

c(t)
,

c(t)2

cd ,

d(t)

d ,

d(t)1
cd



d(t) 2 c(t) 1 d(t)
, and e(t) =
,
,
,
where e(t) = min c(t)
c
d
cd
cd
Then t [(a), (b)]T ,


e(t) e(t),

,
e(t)
(b)(a) .

G(t, s)[f (s, [x w]+ (s)) + g(s, [x w]+ (s)) + q(s)]s

(Ax)(t) =
(a)

=
(a)

G(t, s)
G(t0 , s)[f (s, [x w]+ (s))
G(t0 , s)

+g(s, [x w]+ (s)) + q(s)]s


e(t)(Ax)(t0 ) for any t0 [(a), (b)]T .
Ax(t)
Ax

e(t) for any x K, t [(a), (b)]T .

184

R. Dahal

Thus Ax K and hence A(K) K.


By standard arguments using ArzelaAscoli Theorem, we get that A is a
completely continuous operator.

Now from the denitions of A and w, it is easy to prove the following lemma.
Lemma 2.3: u is a positive solution of BVP (1.1) if and only if u > 0 and u + w
is the xed point of the operator A.
Proof: If u + w is the xed point of A, we get


u+w =

G(t, s)[f (s, u) + g(s, u) + q(s)]s.


(a)

Using properties of the Greens function, we get


(p(t)u ) (p(t)w ) = f (t, u) + g(t, u) + q(t)
(p(t)u ) = f (t, u) + g(t, u),
using the denition of w from (2.3).
It is easy to see u satises the boundary conditions. The other direction is
similar.

Lemma 2.4: Assume (H3 ) holds. Let r = {x K :
x
< r}, and r = {x
b
K :
x
= r}, where r > 2((b) (a)) (a) q(s)s. Then i(A, r K, K) = 1.
Proof: Assume that there exists u0 r and 0 1 such that 0 u0 = Au0 . Then
u0 = 10 Au0 , and 0 < 10 1. Since u0 K, we have
u0 (t)
u0

e(t) = r
e(t).
Then since for all t [(a), (b)]T ,


w(t) =

G(t, s)q(s)s
(a)

e(t)

q(s)s
(a)

= e(t)((b) (a))

q(s)s
(a)

u0 (t)
((b) (a))
r

q(s)s,
(a)

we get using (H3 ) that for all t [(a), (b)]T ,



u0 (t) w(t)

((b) (a))
1
r


q(s)s u0 (t)

(a)

1
u0 (t) 0.
2

(2.5)

185

Positive solutions for a second-order


Now for all t [(a), (b)]T ,
1
Au0 (t)
0
 b

G(t, s)[f (s, [u0 w]+ (s)) + g(s, [u0 w]+ (s)) + q(s)]s

u0 (t) =

(a)

G(t, s)[p1 (s)q1 (u0 w) + p2 (s)q2 (u0 w) + q(s)]s


(a)

max {q1 (u), |q2 (u)|, 1}


0ur

G(t, s)[p1 (s) + p2 (s) + q(s)]s


(a)

max {q1 (u), |q2 (u)|, 1}


0ur

e(s)[p1 (s) + p2 (s) + q(s)]s


(a)


u0
= r max {q1 (u), |q2 (u)|, 1}
0ur

e(s)[p1 (s) + p2 (s) + q(s)]s


(a)

which is a contradiction to (1.3). So Au = u for all u r , 1. Then by


Lemma 2.1, i(A, r K, K) = 1.

Lemma 2.5: Assume (H4 ) holds. Then there exists a constant R > r such that
i(A, R K, K) = 0.
Proof: By (H4 ),  > 0, there exists R > r > 0 such that f (t, u) > (M + )u
2R
, where t1 , t2 are as in (H4 ). Then since
for u R . Let R >
mint[t1 ,t2 ]T e(t)
mint[t1 ,t2 ]T e(t) < 1, it follows that R > R > r > 0.
Dene an operator B such that Bu(t) 1, t [(a), (b)]T . Then it is
clear that inf ur K
Bu
> 0, and B : K K is a completely continuous
operator.
We show that u Au = Bu, u R K, 0. Assume otherwise, then
there exist v0 R , 0 0, such that
v0 Av0 = 0 Bv0 .

(2.6)

Let := min{v0 (t), t [t1 , t2 ]T }, where t1 , t2 are dened as dened in (H4 ). Note
that for all t [(a), (b)]T ,
v0 (t)
v0

e(t) = R
e(t).
Then for all t [t1 , t2 ]T , we get



((b) (a)) b
v0 (t) w(t) 1
q(s)s v0 (t)
R
(a)



((b) (a)) b
1
q(s)s v0 (t)
r
(a)

186

R. Dahal
1
v0 (t) 0
2
1
1
R
e(t) R min e(t) > R .
2
2 t[t1 ,t2 ]T

Now for all t [t1 , t2 ]T , by (2.6)


v0 (t) Av0 (t) = 0 Bv0 (t)
 b
v0 (t) =
G(t, s)[f (s, [v0 w]+ (s)) + g(s, [v0 w]+ (s))
(a)

+q(s)]s + 0
 t2

G(t, s)[f (s, [v0 w](s))]s + 0


t1
t2

t1
 t2
t1
t2

G(t, s)(M + )[v0 (s) w(s)]s + 0


1
G(t, s)(M + ) v0 (s)s + 0
2

1
G(t, s)(M + ) s + 0
2
t1

1 t2
= (M + )
G(t, s)s + 0
2 t1

(M + )M 1 + 0 , by (H4 )
> ,
which is a contradiction to := min{v0 (t), t [t1 , t2 ]}. So for all u R
K, 0, we have u Au = Bu. By Lemma 2.1, we get i(A, R K, K) = 0. 

3 Main result
Now we state and prove the main result.
Theorem 3.1: Suppose that (H1 ) (H4 ) hold. Then the BVP (1.1) has at least one
positive solution.
Proof: By Lemmas 2.4, 2.5 and a property of the xed point index
R \ r ) K, K) = i(A, R K, K) i(A, r K, K)
i(A, (
= 0 1 = 1 (= 0).
This implies that A has a xed point z0 such that r <
z0
< R.
Now let v(t) := z0 (t) w(t), t [(a), (b)]T . Since
z0
r, we get



((b) (a)) b
Z0 (t) w(t) 1
q(s)s z0 (t)
r
(a)

1
z0 (t) > 0.
2

187

Positive solutions for a second-order

That is, v > 0. Also we have v + w (= z0 ) is a xed point of A. Then by Lemma 2.3,
v (= z0 w) is a positive solution of BVP (1.1).


4 Examples
The following example illustrates Theorem 3.1.
Example 4.1: Let T =

x +

2
1+x

50 t

1
2n


n=0

{0, 2}. Then we claim the BVP

1
ex +1

= 0, t (0, 2)T ,

(4.1)

u(0) = 0 = u(2)
has a positive solution.
Note that the BVP (4.1) is of the form (1.1) with 1 = 2 = 1, 1 = 2 = 0, b = 1
and
1 + x2
1
, g(t, x) =
f (t, x) =
50 t
t

1
2 x
e +1


.

Also note that f and g have a singularity at t = 0 and that (a) = a = 0.


Take q(t) = 2t , then it easily follows that g(t, x) > q(t). Thus (H1 ) is satised.
Now take
1
1 + x2
1
, p2 (t) = , q2 (x) =
p1 (t) = , q1 (x) =
50
t
t


2

1
ex + 1

Note that


e(s)q(s)s =
(a)

s(2 s) 2
=
2
s

(2 s s3/2 )s



1 1 1 1
1
= 2 1 + + +
2
2 4 2 8


1
1 1 1 1
1 + + +
2
8 4 8 8

2 2
2 2

0.94,
=
2 21 4 21

and


e(s)[p1 (s) + p2 (s) + q(s)], s


(a)

so (H2 ) is also satised.

=
0

(2 s s3/2 )s 0.94,


.

188

R. Dahal
For (H3 ), since 2 2

1
0

2
s


s = 8 1 +

1
2

13.6, we take r = 14. Now

14
r
 1+x2
 3.5
=
max0|x|r {q1 (x), |q2 (x)|, 1}
max0|x|14 50 , 2 ex1+1 , 1
 1
>
e(s)[p1 (s) + p2 (s) + q(s)]s 0.94
0

Since

lim x

f (t, x)
1 + x2
= , (H4 )
= lim x inf
0<t<2 50x t
x
(a)<t<(b)
inf

is

also

satised.
Thus all the hypotheses (H1 ) (H4 ) are satised and hence the existence of a
positive solution of (4.1) is now guaranteed from Theorem 3.1.

References
Agarwal, R. and ORegan, D. (1999) A note on existence of nonnegative solutions to
singular semi-positone problems, Nonlinear Anal., Vol. 36, pp. 615622.
Atici, F.M. and Guseinov, G.Sh. (2002) On Greens functions and positive solutions for
boundary value problems on time scales, J. Comput. Appl. Math., Vol. 141, pp.7599.
Bohner, M. and Peterson, A. (2001) Dynamic Equations on Time Scales: An Introduction
with Applications, Birkhuser, Boston.
Bohner, M. and Peterson, A. (2003) Advances in Dynamic Equations on Time Scales,
Birkhuser, Boston.
Dahal, R. (2009) Positive solutions of semipositone singular Dirichlet dynamic boundary
value problems, Nonlinear Dynamics and Systems Theory, Vol. 9, No. 4, pp.361374.
Erbe, L., Peterson, A. and Weiss, J. (2008) Positive solutions to a singular second order
boundary value problem, Adv. Dyn. Syst. Appl., Vol. 3, No. 1, pp.89105.
Guo, D.J. and Sun, J. (1998) Calculation and application of topological degree,
J. Math. Res. Expro., Vol. 8, pp.469480.
Guo, D.J. and Lakshmikantham, V. (1988) Nonlinear Problems in Abstract Cone, Academic
Press Inc., New York.
Liu, J. (2007) Positive solutions for second-order singular semipositone boundary value
problems, Int. Journal of Math. Analysis, Vol. 1, No. 28, pp.13771381.

You might also like