J. Math. Anal. Appl.

311 (2005) 495–505

www.elsevier.com/locate/jmaa

Positive solutions for boundary value problem

of nonlinear fractional differential equation ✩
Zhanbing Bai a,∗,1 , Haishen Lü b
a Department of Applied Mathematics, University of Petroleum, Dong ying 257061, People’s Republic of China
b Department of Applied Mathematics, Hohai University, Nanjing 210098, People’s Republic of China

Received 14 October 2004

Available online 2 April 2005
Submitted by A. Cellina

Abstract
In this paper, we investigate the existence and multiplicity of positive solutions for nonlinear frac-
tional differential equation boundary value problem:
α u(t) + f
t, u(t) = 0, 0 < t < 1,
D0+
u(0) = u(1) = 0,
where 1 < α
2 is a real number, D0+ α is the standard Riemann–Liouville differentiation, and
f : [0, 1] × [0, ∞) → [0, ∞) is continuous. By means of some fixed-point theorems on cone, some
existence and multiplicity results of positive solutions are obtained. The proofs are based upon the
reduction of problem considered to the equivalent Fredholm integral equation of second kind.
 2005 Elsevier Inc. All rights reserved.

Keywords: Fractional differential equation; Boundary value problem; Positive solution; Green’s function;
Fixed-point theorem

✩
This work is sponsored by the National Nature Science Foundation of China (10371006) and the Doctoral
Program Foundation of Education Ministry of China (1999000722).
* Corresponding author.
E-mail addresses: [email protected], [email protected] (Z. Bai), [email protected]
(H. Lü).
1 Present address: College of Information Science and Technology, Shandong University of Science and
Technology, Qingdao 266510, People’s Republic of China.

0022-247X/$ – see front matter  2005 Elsevier Inc. All rights reserved.
doi:10.1016/j.jmaa.2005.02.052
496 Z. Bai, H. Lü / J. Math. Anal. Appl. 311 (2005) 495–505

1. Introduction

Fractional differential equations have been of great interest recently. It is caused both
by the intensive development of the theory of fractional calculus itself and by the appli-
cations of such constructions in various sciences such as physics, mechanics, chemistry,
engineering, etc. For details, see [4,6–8,12,14] and references therein.
It should be noted that most of papers and books on fractional calculus are devoted to
the solvability of linear initial fractional differential equations in terms of special functions
[11,14,15]. Recently, there are some papers deal with the existence and multiplicity of
solution (or positive solution) of nonlinear initial fractional differential equation by the use
of techniques of nonlinear analysis (fixed-point theorems, Leray–Shauder theory, etc.), see
[1–3,5,16,17].
However, there are few papers consider the Dirichlet-type problem for linear ordinary
differential equations of fractional order, see [8,13]. No contributions exist, as far as we
know, concerning the existence and multiplicity of positive solutions of the following prob-
lem:


α
D0+ u(t) + f t, u(t) = 0, 0 < t < 1, (1.1)
u(0) = u(1) = 0, (1.2)
where 1 < α
2 is a real number, α
D0+
is the standard Riemann–Liouville differentiation,
and f : [0, 1] × [0, ∞) → [0, ∞) is continuous.
In this paper, we firstly derive the corresponding Green’s function. Consequently prob-
lem (1.1), (1.2) is deduced to a equivalent Fredholm integral equation of the second kind.
Finally, by the means of some fixed-point theorems, the existence and multiplicity of pos-
itive solutions are obtained.

2. Background materials and preliminaries

For the convenience of the reader, we present here the necessary definitions from frac-
tional calculus theory. These definitions can be found in the recent literature.

Definition 2.1. The fractional integral of order α > 0 of a function y : (0, ∞) → R is given
by
t
1
α
I0+ y(t) = (t − s)α−1 y(s) ds
Γ (α)
0
provided the right side is pointwise defined on (0, ∞).

Definition 2.2. The fractional derivative of order α > 0 of a continuous function

y : (0, ∞) → R is given by
 n  t
1 d y(s)
D0+ y(t) =
α
ds,
Γ (n − α) dt (t − s)α−n+1
0
 Z. Bai, H. Lü / J. Math. Anal. Appl. 311 (2005) 495–505 497

where n = [α] + 1, provided that the right side is pointwise defined on (0, ∞).

Remark 2.1. As a basic example, we quote for λ > −1,

Γ (λ + 1) λ−α
α λ
D0+ t = t ,
Γ (λ − α + 1)
α t α−m = 0, m = 1, 2, . . . , N , where N is the smallest integer greater
giving in particular D0+
than or equal to α.

In fact, for λ > −1,

 n  t
1 d sλ
α λ
D0+ t = ds
Γ (n − α) dt (t − s)α−n+1
0
 n 1
1 d
= t n−α+λ
zλ (1 − z)n−α−1 dz
Γ (n − α) dt
0
 n
Γ (λ + 1) d
= t n−α+λ .
Γ (λ + 1 + n − α) dt
So,
 
Γ (α − m + 1) d n n−m
α α−m
D0+ t = t = 0, for m = 1, 2, . . . , N.
Γ (n − m + 1) dt
From Definition 2.2 and Remark 2.1, we then obtain

Lemma 2.1. Let α > 0. If we assume u ∈ C(0, 1) ∩ L(0, 1), then the fractional deferential
equation
α
D0+ u(t) = 0

has u(t) = C1 t α−1 +C2 t α−2 +· · ·+CN t α−N , Ci ∈ R, i = 1, 2, . . . , N , as unique solutions.

α I α u = u for all u ∈ C(0, 1) ∩ L(0, 1). From Lemma 2.1 we deduce the follow-
As D0+ 0+
ing law of composition.

Lemma 2.2. Assume that u ∈ C(0, 1) ∩ L(0, 1) with a fractional derivative of order α > 0
that belongs to C(0, 1) ∩ L(0, 1). Then
α
I0+ α
D0+ u(t) = u(t) + C1 t α−1 + C2 t α−2 + · · · + CN t α−N ,
for some Ci ∈ R, i = 1, 2, . . . , N .

In the following, we present the Green’s function of fractional differential equation

boundary value problem.
498 Z. Bai, H. Lü / J. Math. Anal. Appl. 311 (2005) 495–505

Lemma 2.3. Given y ∈ C[0, 1] and 1 < α
2, the unique solution of

α
D0+ u(t) + y(t) = 0, 0 < t < 1, (2.1)
u(0) = u(1) = 0, (2.2)
is
1
u(t) = G(t, s)y(s) ds,
0
where

 [t (1−s)]α−1 −(t−s)α−1 , 0
s
t
1,
Γ (α)
G(t, s) = [t (1−s)]α−1 (2.3)
 , 0
t
s
1.
Γ (α)

Proof. We may apply Lemma 2.2 to reduce Eq. (2.1) to an equivalent integral equation
u(t) = I0+
α
y(t) + C1 t α−1 + C2 t α−2 ,
for some C1 , C2 ∈ R. Consequently, the general solution of Eq. (2.1) is
t
(t − s)α−1
u(t) = − y(s) ds + C1 t α−1 + C2 t α−2 .
Γ (α)
0
1
By (2.2), there are C2 = 0, C1 = 0 (1 − s)α−1 y(s) ds/Γ (α). Therefore, the unique solu-
tion of problem (2.1), (2.2) is
t 1
(t − s)α−1 (1 − s)α−1 t α−1
u(t) = − y(s) ds + y(s) ds
Γ (α) Γ (α)
0 0
t 1
[t (1 − s)]α−1 − (t − s)α−1 [t (1 − s)]α−1
= y(s) ds + y(s) ds
Γ (α) Γ (α)
0 t
1
= G(t, s)y(s) ds.
0
The proof is complete. 2

Lemma 2.4. The function G(t, s) defined by Eq. (2.3) satisfies the following conditions:

(1) G(t, s) > 0, for t, s ∈ (0, 1);

(2) There exists a positive function γ ∈ C(0, 1) such that
min G(t, s)  γ (s) max G(t, s) = γ (s)G(s, s), for 0 < s < 1. (2.4)
1/4
t
3/4 0
t
1
 Z. Bai, H. Lü / J. Math. Anal. Appl. 311 (2005) 495–505 499

Proof. Observing the expression of G(t, s), it is clear that G(t, s) > 0 for s, t ∈ (0, 1). In
the following, we consider the existence of γ (s). Firstly, for given s ∈ (0, 1), G(t, s) is de-
creasing with respect to t for s
t and increasing with respect to t for t
s. Consequently,
setting
[t (1 − s)]α−1 − (t − s)α−1 [t (1 − s)]α−1
g1 (t, s) = , g2 (t, s) = ,
Γ (α) Γ (α)
one has

 3
 g1 ( 4 , s), s ∈ (0, 1/4],
min G(t, s) = min{g1 ( 34 , s), g2 ( 14 , s)}, s ∈ [1/4, 3/4],
1/4
t
3/4 
 g ( 1 , s),
2 4 s ∈ [3/4, 1),

g1 ( 34 , s), s ∈ (0, r],
=
g2 ( 14 , s), s ∈ [r, 1),

1
{[ 34 (1 − s)]α−1 − ( 34 − s)α−1 }, s ∈ (0, r],
= Γ (α)
Γ (α) 4α−1 (1 − s) s ∈ [r, 1),
1 1 α−1 ,

where 1/4 < r < 3/4 is the unique solution of the equation
α−1  α−1
3 3 1
(1 − s) − −s = α−1 (1 − s)α−1 .
4 4 4
Specially, r = 0.5 if α = 2; r → 0.5 as α → 2 and r → 0.75 as α → 1.
Secondly, with the use of the monotonicity of G(t, s), we have
1  α−1
max G(t, s) = G(s, s) = s(1 − s) , s ∈ (0, 1).
0
t
1 Γ (α)
Thus, setting
 3
 [ 4 (1−s)]α−1 −( 34 −s)α−1 , s ∈ (0, r],
γ (s) = [s(1−s)]α−1
 1
, s ∈ [r, 1),
(4s)α−1

the proof is complete. 2

Remark 2.2. Clearly, γ (s) → 0 when s → 0 unless that α = 2 (inf0<s<1 γ (s) = 1/4 if
α = 2). Consequently, we cannot acquire a positive constant γ take instead of the role
of positive function γ (s) with 1 < α < 2 in (2.4). In our opinion, it is the key that the
results obtained for fractional differential equations in this paper are weaker than that have
obtained for integer-order differential equations.

Definition 2.3. The map θ is said to be a nonnegative continuous concave functional on a

cone P of a real Banach space E provided that θ : P → [0, ∞) is continuous and


θ tx + (1 − t)y  tθ (x) + (1 − t)θ (y)
for all x, y ∈ P and 0
t
1.
500 Z. Bai, H. Lü / J. Math. Anal. Appl. 311 (2005) 495–505

The following fixed-point theorems are fundamental in the proofs of our main results.

Lemma 2.5 [9]. Let E be a Banach space, P ⊆ E a cone, and Ω1 , Ω2 two bounded open
balls of E centered at the origin with Ω̄1 ⊂ Ω2 . Suppose that A : P ∩ (Ω̄2 \ Ω1 ) → P is a
completely continuous operator such that either

(i) Ax
x, x ∈ P ∩ ∂Ω1 and Ax  x, x ∈ P ∩ ∂Ω2 , or

(ii) Ax  x, x ∈ P ∩ ∂Ω1 and Ax
x, x ∈ P ∩ ∂Ω2

holds. Then A has a fixed point in P ∩ (Ω̄2 \ Ω1 ).

Lemma 2.6 [10]. Let P be a cone in a real Banach space E, Pc = {x ∈ P | x
c}, θ a
nonnegative continuous concave functional on P such that θ (x)
x, for all x ∈ P̄c , and
P (θ, b, d) = {x ∈ P | b
θ (x), x
d}. Suppose A : P̄c → P̄c is completely continuous
and there exist constants 0 < a < b < d
c such that

(C1) {x ∈ P (θ, b, d) | θ (x) > b} = ∅ and θ (Ax) > b for x ∈ P (θ, b, d);
(C2) Ax < a for x
a;
(C3) θ (Ax) > b for x ∈ P (θ, b, c) with Ax > d.

Then A has at least three fixed points x1 , x2 , x3 with

x1  < a, b < θ (x2 ), a < x3  with θ (x3 ) < b.

Remark 2.3. If there holds d = c, then condition (C1) of Lemma 2.6 implies condition
(C3) of Lemma 2.6.

3. Main results

In this section, we impose growth conditions on f which allow us to apply Lemmas 2.5
and 2.6 to establish some results of existence and multiplicity of positive solutions for
problem (1.1), (1.2).
Let E = C[0, 1] be endowed with the ordering u
v if u(t)
v(t) for all t ∈ [0, 1],
and the maximum norm, |u| = max0
t
1 |u(t)|. Define the cone P ⊂ E by
 
P = u ∈ E | u(t)  0 .
Let the nonnegative continuous concave functional θ on the cone P be defined by
 
θ (u) = min u(t).
1/4
t
3/4

Lemma 3.1. Let T : P → E be the operator defined by

1


T u(t) := G(t, s)f s, u(s) ds,
0
then T : P → P is completely continuous.
 Z. Bai, H. Lü / J. Math. Anal. Appl. 311 (2005) 495–505 501

Proof. The operator T : P → P is continuous in view of nonnegativeness and continuity

of G(t, s) and f (t, u).
Let Ω ⊂ P be bounded, i.e., there exists a positive constant M > 0 such that u
M,
for all u ∈ Ω. Let L = max0
t
1, 0
u
M |f (t, u)| + 1, then, for u ∈ Ω, we have
1 1
 

T u(t)
G(t, s)f s, u(s) ds
L G(s, s) ds.
0 0
Hence, T (Ω) is bounded.
On the other hand, given  > 0, setting
  1
1 Γ (α) α−1
δ= ,
2 M
then, for each u ∈ Ω, t1 , t2 ∈ [0, 1], t1 < t2 , and t2 − t1 < δ, one has |T u(t2 ) − T u(t1 )| < .
That is to say, T (Ω) is equicontinuity.
In fact,
 
T u(t2 ) − T u(t1 )
 1 1 


 

=  G(t2 , s)f s, u(s) ds − G(t1 , s)f s, u(s) ds 
 
0 0
t1 1
 
  

= G(t2 , s) − G(t1 , s) f s, u(s) ds + G(t2 , s) − G(t1 , s) f s, u(s) ds
0 t2
t2
 

+ G(t2 , s) − G(t1 , s) f s, u(s) ds
t1
 t1
M

< (1 − s)α−1 t2α−1 − t1α−1 ds
Γ (α)
0
1 t2 

α−1 
α−1 
+ (1 − s) α−1
t2 − t1α−1 ds + (1 − s) α−1
t2 − t1α−1 ds
t2 t1
M
α−1 
< t2 − t1α−1 .
Γ (α)
In the following, we divide the proof into two cases.
Case 1. δ
t1 < t2 < 1.
 

T u(t2 ) − T u(t1 ) < M t α−1 − t α−1
M α − 1 (t2 − t1 )
Γ (α) 2 1 Γ (α) δ 2−α
M

(α − 1)δ α−1
.
Γ (α)
502 Z. Bai, H. Lü / J. Math. Anal. Appl. 311 (2005) 495–505

Case 2. 0
t1 < δ, t2 < 2δ.

 

T u(t2 ) − T u(t1 ) < M t α−1 − t α−1
M t α−1 < M (2δ)α−1
.
Γ (α) 2 1 Γ (α) 2 Γ (α)
By the means of the Arzela–Ascoli theorem, we have T : P → P is completely continuous.
The proof is complete. 2

Denote
 1 −1  3/4 −1
M= G(s, s) ds , N= γ (s)G(s, s) ds .
0 1/4

Theorem 3.1. Let f (t, u) is continuous on [0, 1] × [0, ∞). Assume that there exist two
positive constants r2 > r1 > 0 such that

(H1) f (t, u)
Mr2 , for (t, u) ∈ [0, 1] × [0, r2 ];

(H2) f (t, u)  Nr1 , for (t, u) ∈ [0, 1] × [0, r1 ].

Then problem (1.1), (1.2) has at least one positive solution u such that r1
u
r2 .

Proof. By Lemmas 2.3 and 3.1, we know T : P → P is completely continuous and prob-
lem (1.1), (1.2) has a solution u = u(t) if and only if u solves the operator equation u = T u.
In order to apply Lemma 2.5, we separate the proof into the following two steps.
Step 1. Let Ω2 := {u ∈ P | u < r2 }. For u ∈ ∂Ω2 , we have 0
u(t)
r2 for all
t ∈ [0, 1]. It follows from (H1) that for t ∈ [0, 1],
1 1


T u = max G(t, s)f s, u(s) ds
Mr2 G(s, s) ds = r2 = u.
0
t
1
0 0

Step 2. Let Ω1 := {u ∈ P | u < r1 }. For u ∈ ∂Ω1 , we have 0
u(t)
r1 for all
t ∈ [0, 1]. By assumption (H2), for t ∈ [1/4, 3/4], there is

1 1



T u(t) = G(t, s)f s, u(s) ds  γ (s)G(s, s)f s, u(s) ds
0 0
3/4
 Nr1 γ (s)G(s, s) ds = r1 = u.
1/4

So
T u  u, for u ∈ ∂Ω1 .
Therefore, by (ii) of Lemma 2.5, we complete the proof. 2
 Z. Bai, H. Lü / J. Math. Anal. Appl. 311 (2005) 495–505 503

Example 3.1. Consider the problem

3/2 sin t
D0+ u(t) + u2 + + 1 = 0, 0 < t < 1, (3.1)
4
u(0) = u(1) = 0. (3.2)
√
A simple computation showed M = 4/ π ≈ 2.25676, N ≈ 13.6649. Choosing r1 = 1/14,
r2 = 1, we have
sin t
f (t, u) = 1 + + u2
2.2107
Mr2 , for (t, u) ∈ [0, 1] × [0, 1],
4
sin t
f (t, u) = 1 + + u2  1  N r1 , for (t, u) ∈ [0, 1] × [0, 1/14].
4
With the use of Theorem 3.1, problem (3.1), (3.2) has at least one solution u such that
1/14
u
1.

Theorem 3.2. Suppose f (t, u) is continuous on [0, 1] × [0, ∞) and there exist constants
0 < a < b < c such that the following assumptions hold:

(A1) f (t, u) < Ma, for (t, u) ∈ [0, 1] × [0, a];

(A2) f (t, u)  N b, for (t, u) ∈ [1/4, 3/4] × [b, c];
(A3) f (t, u)
Mc, for (t, u) ∈ [0, 1] × [0, c].

Then, the boundary value problem (1.1), (1.2) has at least three positive solutions u1 , u2 ,
and u3 with
     
max u1 (t) < a, b < min u2 (t) < max u2 (t)
c,
0
t
1 1/4
t
3/4 0
t
1
   
a < max u3 (t)
c, min u3 (t) < b.
0
t
1 1/4
t
3/4

Proof. We show that all the conditions of Lemma 2.4 are satisfied.
If u ∈ P̄c , then u
c. Assumption (A3) implies f (t, u(t))
Mc for 0
t
1. Con-
sequently,
 1  1

 


T u = max  G(t, s)f s, u(s) ds 
G(s, s)f s, u(s) ds
0
t
1 
0 0
1

G(s, s)Mc ds
c.
0

Hence, T : P̄c → P̄c . In the same way, if u ∈ P̄a , then assumption (A1) yields f (t, u(t)) <
Ma, 0
t
1. Therefore, condition (C2) of Lemma 2.6 is satisfied.
To check condition (C1) of Lemma 2.6, we choose u(t) = (b + c)/2, 0
t
1. It is
easy to see that u(t) = (b + c)/2 ∈ P (θ, b, c), θ (u) = θ ((b + c)/2) > b, consequently, {u ∈
P (θ, b, c) | θ (u) > b} = ∅. Hence, if u ∈ P (θ, b, c), then b
u(t)
c for 1/4
t
3/4.
From assumption (A2), we have f (t, u(t))  N b for 1/4
t
3/4. So
504 Z. Bai, H. Lü / J. Math. Anal. Appl. 311 (2005) 495–505

1
 

θ (T u) = min (T u)(t)  γ (s)G(s, s)f s, u(s) ds
1/4
t
3/4
0
3/4
> γ (s)G(s, s)Nb ds = b,
1/4

i.e.,
θ (T u) > b, for all u ∈ P (θ, b, c).
This shows that condition (C1) of Lemma 2.6 is also satisfied.
By Lemma 2.6 and Remark 2.3, the boundary value problem (1.1), (1.2) has at least
three positive solutions u1 , u2 , and u3 satisfying
   
max u1 (t) < a, b < min u2 (t),
0
t
1 1/4
t
3/4
   
a < max u3 (t), min u3 (t) < b.
0
t
1 1/4
t
3/4

The proof is complete. 2

Example 3.3. Consider the problem

3/2
D0+ u(t) + f (t, u) = 0, 0 < t < 1, (3.3)
u(0) = u(1) = 0, (3.4)
where

t
+ 14u2 , for u
1,
f (t, u) = 20
13 + 20
t
+ u, for u > 1.
√
We have M = 4/ π ≈ 2.25676, N ≈ 13.6649. Choosing a = 1/10, b = 1, c = 12, there
hold
t
f (t, u) = + 14u2
0.19
Ma ≈ 0.225, for (t, u) ∈ [0, 1] × [0, 1/10],
20
t
f (t, u) = 13 + + u  14.05  N b ≈ 13.7, for (t, u) ∈ [1/4, 3/4] × [1, 12],
20
t
f (t, u) = 13 + + u
25.05
Mc ≈ 27.1, for (t, u) ∈ [0, 1] × [0, 12].
20
With the use of Theorem 3.2, problem (3.3), (3.4) has at least three positive solutions u1 ,
u2 and u3 with
     
max u1 (t) < 1/10, 1 < min u2 (t) < max u2 (t)
12,
0
t
1 1/4
t
3/4 0
t
1
   
1/10 < max u3 (t)
12, min u3 (t) < 1.
0
t
1 1/4
t
3/4
 Z. Bai, H. Lü / J. Math. Anal. Appl. 311 (2005) 495–505 505

References

[1] A. Babakhani, V.D. Gejji, Existence of positive solutions of nonlinear fractional differential equations,
J. Math. Anal. Appl. 278 (2003) 434–442.
[2] D. Delbosco, Fractional calculus and function spaces, J. Fract. Calc. 6 (1996) 45–53.
[3] D. Delbosco, L. Rodino, Existence and uniqueness for a nonlinear fractional differential equation, J. Math.
Anal. Appl. 204 (1996) 609–625.
[4] A.M.A. El-Sayed, Nonlinear functional differential equations of arbitrary orders, Nonlinear Anal. 33 (1998)
181–186.
[5] V.D. Gejji, A. Babakhani, Analysis of a system of fractional differential equations, J. Math. Anal. Appl. 293
(2004) 511–522.
[6] A.A. Kilbas, O.I. Marichev, S.G. Samko, Fractional Integral and Derivatives (Theory and Applications),
Gordon and Breach, Switzerland, 1993.
[7] A.A. Kilbas, J.J. Trujillo, Differential equations of fractional order: methods, results and problems I, Appl.
Anal. 78 (2001) 153–192.
[8] A.A. Kilbas, J.J. Trujillo, Differential equations of fractional order: methods, results and problems II, Appl.
Anal. 81 (2002) 435–493.
[9] M.A. Krasnosel’skii, Positive Solutions of Operator Equations, Noordhoff, Groningen, 1964.
[10] R.W. Leggett, L.R. Williams, Multiple positive fixed points of nonlinear operators on ordered Banach spaces,
Indiana Univ. Math. J. 28 (1979) 673–688.
[11] K.S. Miller, Fractional differential equations, J. Fract. Calc. 3 (1993) 49–57.
[12] K.S. Miller, B. Ross, An Introduction to the Fractional Calculus and Fractional Differential Equations, Wi-
ley, New York, 1993.
[13] A.M. Nakhushev, The Sturm–Liouville problem for a second order ordinary differential equation with frac-
tional derivatives in the lower terms, Dokl. Akad. Nauk SSSR 234 (1977) 308–311.
[14] I. Podlubny, Fractional Differential Equations, Mathematics in Science and Engineering, Academic Press,
New York, 1999.
[15] I. Podlubny, The Laplace transform method for linear differential equations of the fractional order, Inst.
Expe. Phys., Slov. Acad. Sci., UEF-02-94, Kosice, 1994.
[16] S.Q. Zhang, The existence of a positive solution for a nonlinear fractional differential equation, J. Math.
Anal. Appl. 252 (2000) 804–812.
[17] S.Q. Zhang, Existence of positive solution for some class of nonlinear fractional differential equations,
J. Math. Anal. Appl. 278 (2003) 136–148.