Professional Documents
Culture Documents
No part of this publication may be reproduced or transmitted in any form or by any means,
electronic or mechanical, including photocopying, recording, or any information storage and
retrieval system, without permission in writing from the publisher. Details on how to seek
permission, further information about the Publisher’s permissions policies and our
arrangements with organizations such as the Copyright Clearance Center and the Copyright
Licensing Agency can be found at our website: www.elsevier.com/permissions.
This book and the individual contributions contained in it are protected under copyright by the
Publisher (other than as may be noted herein).
Notices
Knowledge and best practice in this field are constantly changing. As new research and
experience broaden our understanding, changes in research methods, professional practices, or
medical treatment may become necessary.
Practitioners and researchers must always rely on their own experience and knowledge in
evaluating and using any information, methods, compounds, or experiments described herein.
In using such information or methods they should be mindful of their own safety and the
safety of others, including parties for whom they have a professional responsibility.
To the fullest extent of the law, neither the Publisher nor the authors, contributors, or editors,
assume any liability for any injury and/or damage to persons or property as a matter of
products liability, negligence or otherwise, or from any use or operation of any methods,
products, instructions, or ideas contained in the material herein.
ISBN: 978-0-12-803652-5
focused on the existence, multiplicity, and nonexistence of positive solutions for some
classes of systems of nonlinear Riemann–Liouville fractional differential equations
with parameters or without parameters, subject to coupled Riemann–Stieltjes inte-
gral boundary conditions, and for which the nonlinearities are nonsingular or singular
functions. A system of fractional equations with sign-changing nonsingular or sin-
gular nonlinearities and integral boundary conditions is also investigated. In each
chapter, various examples are presented which support the main results.
Central to the results of each chapter are applications of the Guo–Krasnosel’skii
fixed point theorem for nonexpansive and noncontractive operators on a cone (The-
orem 1.1.1). Unique to applications of the fixed point theorem is the novel rep-
resentation of the Green’s functions, which ultimately provide almost a checklist
for determining conditions for which positive solutions exist relative to given non-
linearities. In the proof of many of the main results, applications are also made
of the Schauder fixed-point theorem (Theorem 1.6.1), the nonlinear alternative of
Leray–Schauder type (Theorem 2.5.1), and some theorems from the fixed point index
theory (Theorems 1.3.1–1.3.3).
There have been other books in the past on positive solutions for boundary value
problems, but in spite of the area receiving much attention, there have been no new
books recently. This monograph provides a springboard for other researchers to em-
ulate the authors’ methods. The audience for this book includes the family of math-
ematical and scientific researchers in boundary value problems for which positive
solutions are important, and in addition, the monograph can serve as a great source
for topics to be studied in graduate seminars.
Johnny Henderson
Rodica Luca
About the authors
We are grateful to all the anonymous referees for carefully reviewing early drafts of
the manuscript. We also express our thanks to Glyn Jones, the Mathematics publisher
at Elsevier, and to Steven Mathews, the Editorial project manager for Mathematics and
Statistics at Elsevier, for their support and encouragement of us during the preparation
of this book; and to Poulouse Joseph, the project manager for book production at
Elsevier, for all of his work on our book.
The work of Rodica Luca was supported by a grant of the Romanian National
Authority for Scientific Research, CNCS-UEFISCDI, project number PN-II-ID-PCE-
2011-3-0557.
Systems of second-order ordinary
differential equations with 1
integral boundary conditions
where the above integrals are Riemann–Stieltjes integrals. The boundary conditions
above include multipoint and integral boundary conditions and the sum of these in a
single framework.
We give sufficient conditions on f and g and on the parameters λ and μ such that
positive solutions of problem (S)–(BC) exist. By a positive solution of problem (S)–
(BC) we mean a pair of functions (u, v) ∈ C 2 ([0, 1]) × C2 ([0, 1]) satisfying (S) and
(BC) with u(t) ≥ 0, v(t) ≥ 0 for all t ∈ [0, 1] and (u, v) = (0, 0). The case in which
the functions H1 , H2 , K1 , and K2 are step functions—that is, the boundary conditions
(BC) become multipoint boundary conditions
⎧
⎪
m
n
⎪
⎪ αu(0) − βa(0)u
(0) = a u(ξ ), γ u(1) + δa(1)u
(1) = bi u(ηi ),
⎪
⎨ i i
i=1 i=1
⎪
⎪
r
l
⎪
⎪ α̃v(0) − β̃c(0)v
(0) = ci v(ζi ), γ̃ v(1) + δ̃c(1)v (1) = di v(ρi ),
⎩
i=1 i=1
(BC1 )
where m, n, r, l ∈ N, (N = {1, 2, . . .})—was studied in Henderson and Luca (2013g).
System (S) with a(t) = 1, c(t) = 1, b(t) = 0, and d(t) = 0 for all t ∈ [0, 1],
f (t, u, v) = f̃ (u, v), and g(t, u, v) = g̃(u, v) (denoted by (S1 )) and (BC1 ) was inves-
tigated in Henderson and Luca (2014e). Some particular cases of the problem from
Henderson and Luca (2014e) were studied in Henderson and Luca (2012e) (where in
(BC1 ), ai = 0 for all i = 1, . . . , m, ci = 0 for all i = 1, . . . , r, γ = γ̃ = 1, and
δ = δ̃ = 0—denoted by (BC2 )), in Luca (2011) (where in (S1 ), f̃ (u, v) = f̃ (v) and
g̃(u, v) = g̃(u)—denoted by (S2 )—and in (BC2 ) we have n = l, bi = di , and ηi = ρi
for all i = 1, . . . , n, α = α̃, and β = β̃—denoted by (BC3 )), in Henderson et al.
(2008b) (problem (S2 )–(BC3 ) with α = α̃ = 1 and β = β̃ = 0), and in Henderson
and Ntouyas (2008b) and Henderson et al. (2008a) (system (S2 ) with the boundary
conditions u(0) = 0, u(1) = αu(η), v(0) = 0, v(1) = αv(η), η ∈ (0, 1), and
0 < α < 1/η, or u(0) = βu(η), u(1) = αu(η), v(0) = βv(η), and v(1) = αv(η)). In
Henderson and Ntouyas (2008a), the authors investigated system (S2 ) with the bound-
ary conditions αu(0) − βu (0) = 0, γ u(1) + δu (1) = 0, αv(0) − βv (0) = 0, and
γ v(1) + δv (1) = 0, where α, β, γ , δ ≥ 0 and α + β + γ + δ > 0.
In the proof of our main results, we shall use the Guo–Krasnosel’skii fixed point
theorem (see Guo and Lakshmikantham, 1988a), which we present now:
Theorem 1.1.1. Let X be a Banach space and let C ⊂ X be a cone in X. Assume
1 and 2 are bounded open subsets of X with 0 ∈ 1 ⊂ 1 ⊂ 2 , and let A :
C ∩ ( 2 \ 1 ) → C be a completely continuous operator (continuous, and compact—
that is, it maps bounded sets into relatively compact sets) such that either
(1) Au ≤ u , u∈ C∩∂ 1, and Au ≥ u , u∈C∩∂ 2, or
(2) Au ≥ u , u∈ C∩∂ 1, and Au ≤ u , u∈C∩∂ 2.
1 1
1
G1 (t, s) = g1 (t, s) + ψ(t) φ(s) dH2 (s) + φ(t) τ1 − ψ(s) dH2 (s)
1 0 0
1 1
1
× g1 (τ , s) dH1 (τ ) + ψ(t) τ1 − φ(s) dH1 (s)
0 1 0
1 1
+ φ(t) ψ(s) dH1 (s) g1 (τ , s) dH2 (τ ),
0 0
(1.5)
4 Boundary Value Problems for Systems of Differential, Difference and Fractional Equations
and ψ and φ are the functions defined by (1.3) and (1.4), respectively.
Proof. Because τ1 = 0, the functions ψ and φ are two linearly independent solu-
tions of the equation (a(t)u (t)) − b(t)u(t) = 0. Then the general solution of (1.1) is
u(t) = Aψ(t) + Bφ(t) + u0 (t), with A, B ∈ R, and u0 is a particular solution of (1.1).
We shall determine u0 by the method of variation of constants—namely, we shall look
for two functions C(t) and D(t) such that u0 (t) = C(t)ψ(t) + D(t)φ(t) is a solution of
(1.1). The derivatives of C(t) and D(t) satisfy the system
⎧
⎨ C (t)ψ(t) + D (t)φ(t) = 0, t ∈ (0, 1),
⎩ C (t)a(t)ψ (t) + D (t)a(t)φ (t) = −y(t), t ∈ (0, 1).
The above system has the determinant d0 = −τ1 = 0, and the solution of the above
system is C (t) = − τ11 φ(t)y(t) and D (t) = τ11 ψ(t)y(t). Then we choose C(t) =
1 t
t φ(s)y(s) ds and D(t) = ψ(s)y(s) ds. We deduce that the general solution
1 1
τ1 τ1 0
of (1.1) is
ψ(t) 1 φ(t) t
u(t) = φ(s)y(s) ds + ψ(s)y(s) ds + Aψ(t) + Bφ(t).
τ1 t τ1 0
Then we obtain
1
u(t) = g1 (t, s)y(s) ds + Aψ(t) + Bφ(t),
0
where g1 is defined in (1.6).
By using condition (1.2), we conclude
⎧
⎪
⎪ 1 1 1 1
⎪
⎪ α ψ(0)φ(s)y(s) ds + Aψ(0) + Bφ(0) − βa(0) φ(s)ψ (0)y(s) ds
⎪
⎪ τ1 0 τ1 0
⎪
⎪
⎪
⎪ 1 1
⎪
⎪
⎪
⎨ + Aψ (0) + Bφ (0) = g 1 (s, τ )y(τ ) dτ + Aψ(s) + Bφ(s) dH1 (s),
0 0
⎪
⎪
⎪ 1 1 1 1
⎪
⎪ γ φ(1)ψ(s)y(s) ds + Aψ(1) + Bφ(1) + δa(1) φ (1)ψ(s)y(s) ds
⎪
⎪ τ1 0 τ1 0
⎪
⎪
⎪
⎪ 1 1
⎪
⎪
⎩ + Aψ (1) + Bφ (1) = g1 (s, τ )y(τ ) dτ + Aψ(s) + Bφ(s) dH2 (s),
0 0
or
Systems of second-order ordinary differential equations 5
⎧ 1
⎪
⎪
⎪
⎪ A −αψ(0) + βa(0)ψ (0) + ψ(s) dH1 (s)
⎪
⎪
⎪
⎪
0
⎪
⎪ 1
⎪
⎪
⎪
⎪ + B −αφ(0) + βa(0)φ (0) + φ(s) dH1 (s)
⎪
⎪
⎪
⎪
0
⎪
⎪
⎪
⎪ α 1 βa(0) 1
⎪
⎪
⎪ = φ(s)ψ(0)y(s) ds − φ(s)ψ (0)y(s) ds
⎪
⎪ τ1 0 τ1 0
⎪
⎪ 1 1
⎪
⎪
⎪
⎪ − g1 (s, τ )y(τ ) dτ dH1 (s),
⎪
⎨ 0 0
⎪
⎪
1
⎪
⎪ A γ ψ(1) + δa(1)ψ (1) − ψ(s) dH2 (s)
⎪
⎪
⎪
⎪ 0
⎪
⎪
⎪
⎪ 1
⎪
⎪ + B γ φ(1) + δa(1)φ (1) − φ(s) dH2 (s)
⎪
⎪
⎪
⎪ 0
⎪
⎪
⎪
⎪ γ 1 δa(1) 1
⎪
⎪ =− φ(1)ψ(s)y(s) ds − φ (1)ψ(s)y(s) ds
⎪
⎪
⎪
⎪ τ1 0 τ1 0
⎪
⎪ 1 1
⎪
⎪
⎪
⎪
⎩ + g1 (s, τ )y(τ ) dτ dH2 (s).
0 0
Therefore, we obtain
⎧ 1 1
⎪
⎪
⎪
⎪ A ψ(s) dH (s) + B −τ + φ(s) dH (s)
⎪
⎪
1 1 1
⎪
⎪ 0
1 1
0
⎪
⎪
⎪
⎪
⎪
⎨ =− g1 (s, τ )y(τ ) dτ dH1 (s),
0 0
1 1 (1.7)
⎪
⎪
⎪
⎪ A τ1 − ψ(s) dH2 (s) + B − φ(s) dH2 (s)
⎪
⎪
⎪
⎪ 0 0
⎪
⎪ 1 1
⎪
⎪
⎪
⎩ = g1 (s, τ )y(τ ) dτ dH2 (s).
0 0
The above system with the unknowns A and B has the determinant
1 1
1 = τ1 − ψ(s) dH2 (s) τ1 − φ(s) dH1 (s)
0 0
1 1
− ψ(s) dH1 (s) φ(s) dH2 (s) .
0 0
By using the assumptions of this lemma, we have 1 = 0. Hence, system (1.7) has a
unique solution—namely,
6 Boundary Value Problems for Systems of Differential, Difference and Fractional Equations
1 1 1
1
A= φ(s) dH2 (s) g1 (s, τ )y(τ ) dτ dH1 (s)
1 0 0 0
1 1 1
+ τ1 − φ(s) dH1 (s) g1 (s, τ )y(τ ) dτ dH2 (s) ,
0 0 0
1 1 1
1
B= ψ(s) dH1 (s) g1 (s, τ )y(τ ) dτ dH2 (s)
1 0 0 0
1 1 1
+ τ1 − ψ(s) dH2 (s) g1 (s, τ )y(τ ) dτ dH1 (s) .
0 0 0
Then the solution of problem (1.1)–(1.2) is
1
u(t) = g1 (t, s)y(s) ds
0
ψ(t) 1 1 1
+ φ(s) dH2 (s) g1 (s, τ )y(τ ) dτ dH1 (s)
1 0 0 0
1 1 1
+ τ1 − φ(s) dH1 (s) g1 (s, τ )y(τ ) dτ dH2 (s)
0 0 0
φ(t) 1 1 1
+ ψ(s) dH1 (s) g1 (s, τ )y(τ ) dτ dH2 (s)
1 0 0 0
1 1 1
+ τ1 − ψ(s) dH2 (s) g1 (s, τ )y(τ ) dτ dH1 (s) .
0 0 0
Therefore, we deduce
1
u(t) = g1 (t, s)y(s) ds
0
1 1
1
+ ψ(t) φ(s) dH2 (s) + φ(t) τ1 − ψ(s) dH2 (s)
1 0 0
1 1 1
1
× g1 (s, τ ) dH1 (s) y(τ ) dτ + ψ(t) τ1 − φ(s) dH1 (s)
0 0 1 0
1 1 1
+ φ(t) ψ(s) dH1 (s) g1 (s, τ ) dH2 (s) y(τ ) dτ
0 0 0
1
= g1 (t, s)y(s) ds
0
1 1
1
+ ψ(t) φ(s) dH2 (s) + φ(t) τ1 − ψ(s) dH2 (s)
1 0 0
1 1 1
1
× g1 (τ , s) dH1 (τ ) y(s) ds + ψ(t) τ1 − φ(s) dH1 (s)
0 0 1 0
1 1 1
+ φ(t) ψ(s) dH1 (s) g1 (τ , s) dH2 (τ ) y(s) ds.
0 0 0
Systems of second-order ordinary differential equations 7
1
So, the solution u of (1.1)–(1.2) is u(t) = 0 G1 (t, s)y(s) ds, t ∈ [0, 1], where G1 is
given in (1.5).
Now, we introduce the following assumptions:
(A1) a ∈ C1 ([0, 1], (0, ∞)), b ∈ C([0, 1], [0, ∞)).
(A2) α, β, γ , δ ∈ [0, ∞), with α + β > 0 and γ + δ > 0.
(A3) If b(t) ≡ 0, then α + γ > 0.
(A4) H1 , H2 : [0, 1] → R are nondecreasing functions.
1 1
(A5) τ1 − 0 φ(s) dH1 (s) > 0, τ1 − 0 ψ(s) dH2 (s) > 0, and 1 > 0.
Lemma 1.1.2 (Atici and Guseinov, 2002). Let (A1) and (A2) hold. Then
(a) the function ψ is nondecreasing on [0, 1], ψ(t) ≥ 0 for all t ∈ [0, 1] and ψ(t) > 0
on (0, 1];
(b) the function φ is nonincreasing on [0, 1], φ(t) ≥ 0 for all t ∈ [0, 1] and φ(t) > 0
on [0, 1).
Lemma 1.1.3 (Atici and Guseinov, 2002). Let (A1) and (A2) hold.
(a) If b(t) is not identically zero, then τ1 > 0.
(b) If b(t) is identically zero, then τ1 > 0 if and only if α + γ > 0.
Lemma 1.1.4. Let (A1)–(A3) hold. Then the function g1 given by (1.6) has the
following properties:
(a) g1 is a continuous function on [0, 1] × [0, 1].
(b) g1 (t, s) ≥ 0 for all t, s ∈ [0, 1], and g1 (t, s) > 0 for all t, s ∈ (0, 1).
(c) g1 (t, s) ≤ g1 (s, s) for all t, s ∈ [0, 1].
(d) For any σ∈ (0, 1/2), we have mint∈[σ ,1−σ ] g1 (t, s) ≥ ν1 g1 (s, s) for all s ∈ [0, 1], where
φ(1−σ ) ψ(σ )
ν1 = min φ(0) , ψ(1) .
For the proof of Lemma 1.1.4 (a) and (b), see Atici and Guseinov (2002), and for
the proof of Lemma 1.1.4 (c) and (d), see Ma and Thompson (2004).
Lemma 1.1.5. Let (A1)–(A5) hold. Then the Green’s function G1 of problem (1.1)–
(1.2) is continuous on [0, 1] × [0, 1] and satisfies G1 (t, s) ≥ 0 for all (t, s) ∈ [0, 1] ×
[0, 1]. Moreover, if y ∈ C(0, 1) ∩ L1 (0, 1) satisfies y(t) ≥ 0 for all t ∈ (0, 1), then the
solution u of problem (1.1)–(1.2) satisfies u(t) ≥ 0 for all t ∈ [0, 1].
Proof. By using the assumptions of this lemma, we deduce G1 (t, s) ≥ 0 for all
(t, s) ∈ [0, 1] × [0, 1], and so u(t) ≥ 0 for all t ∈ [0, 1].
Lemma 1.1.6. Assume that (A1)–(A5) hold. Then the Green’s function G1 of prob-
lem (1.1)–(1.2) satisfies the following inequalities:
(a) G1 (t, s) ≤ J1 (s), ∀ (t, s) ∈ [0, 1] × [0, 1], where
J1 (s) = g1 (s, s)
1 1 1
1
+ ψ(1) φ(s) dH2 (s) + φ(0) τ1 − ψ(s) dH2 (s) g1 (τ , s) dH1 (τ )
1 0 0 0
1 1 1
1
+ ψ(1) τ1 − φ(s) dH1 (s) + φ(0) ψ(s) dH1 (s) g1 (τ , s) dH2 (τ ).
1 0 0 0
1 1
1
× g1 (τ , s) dH1 (τ ) + ψ(σ ) τ1 − φ(s) dH1 (s)
0 1 0
1 1
+ φ(1 − σ ) ψ(s) dH1 (s) g1 (τ , s) dH2 (τ )
0 0
1 ψ(σ ) 1 φ(1 − σ )
= ν1 g1 (s, s) + ψ(1) φ(s) dH2 (s) + φ(0)
1 ψ(1) 0 φ(0)
1 1
× τ1 − ψ(s) dH2 (s) g1 (τ , s) dH1 (τ )
0 0
1 ψ(σ ) 1
+ ψ(1) τ1 − φ(s) dH1 (s)
1 ψ(1) 0
φ(1 − σ ) 1 1
+ φ(0) ψ(s) dH1 (s) g1 (τ , s) dH2 (τ ) ≥ ν1 J1 (s).
φ(0) 0 0
Lemma 1.1.7. Assume that (A1)–(A5) hold and let σ ∈ (0, 1/2). If y ∈ C(0, 1) ∩
L1 (0, 1), y(t) ≥ 0 for all t ∈ (0, 1), then the solution u(t), t ∈ [0, 1], of problem
(1.1)–(1.2) satisfies the inequality inft∈[σ ,1−σ ] u(t) ≥ ν1 supt ∈[0,1] u(t ).
Proof. For σ ∈ (0, 1/2), t ∈ [σ , 1 − σ ], and t ∈ [0, 1] we have
1 1 1
u(t) = G1 (t, s)y(s) ds ≥ ν1 J1 (s)y(s) ds ≥ ν1 G1 (t , s)y(s) ds = ν1 u(t ),
0 0 0
and so inft∈[σ ,1−σ ] u(t) ≥ ν1 supt ∈[0,1] u(t ).
We can also formulate results similar to those in Lemmas 1.1.1–1.1.7 for the bound-
ary value problem
Systems of second-order ordinary differential equations 9
f (t, u, v) g(t, u, v)
f0i = lim inf min , gi0 = lim inf min ,
u+v→0+ t∈[σ ,1−σ ] u + v u+v→0+ t∈[σ ,1−σ ] u + v
10 Boundary Value Problems for Systems of Differential, Difference and Fractional Equations
s f (t, u, v) g(t, u, v)
f∞ = lim sup max , gs∞ = lim sup max ,
u+v→∞ t∈[0,1] u+v u+v→∞ t∈[0,1] u+v
f (t, u, v) g(t, u, v)
i
f∞ = lim inf min , gi∞ = lim inf min .
u+v→∞ t∈[σ ,1−σ ] u+v u+v→∞ t∈[σ ,1−σ ] u+v
In the definitions of the extreme limits above, the variables u and v are
nonnegative.
By using the Green’s functions G1 and G2 from Section 1.1.2 (Lemma 1.1.1), we
can write our problem (S)–(BC) equivalently as the following nonlinear system of
integral equations:
⎧ 1
⎪
⎪
⎪
⎨ u(t) = λ G1 (t, s)p(s)f (s, u(s), v(s)) ds, 0 ≤ t ≤ 1,
0
⎪
⎪
⎪
1
⎩ v(t) = μ G2 (t, s)q(s)g(s, u(s), v(s)) ds, 0 ≤ t ≤ 1.
0
We consider the Banach space X = C([0, 1]) with the supremum norm u =
supt∈[0,1] |u(t)| and the Banach space Y = X×X with the norm (u, v) Y = u + v .
We define the cone P ⊂ Y by
P = (u, v) ∈ Y; u(t) ≥ 0, v(t) ≥ 0, ∀ t ∈ [0, 1] and
inf (u(t) + v(t)) ≥ ν (u, v) Y ,
t∈[σ ,1−σ ]
where ν = min{ν1 , ν2 }, and ν1 and ν2 are the constants defined in Section 1.1.2
(Lemma 1.1.4) with respect to the above σ .
For λ, μ > 0, we introduce the operators Q1 , Q2 : Y → X and Q : Y → Y
defined by
1
Q1 (u, v)(t) = λ G1 (t, s)p(s)f (s, u(s), v(s)) ds, 0 ≤ t ≤ 1,
0
1
Q2 (u, v)(t) = μ G2 (t, s)q(s)g(s, u(s), v(s)) ds, 0 ≤ t ≤ 1,
0
and Q(u, v) = (Q1 (u, v), Q2 (u, v)), (u, v) ∈ Y. The solutions of our problem (S)–(BC)
coincide with the fixed points of the operator Q.
Lemma 1.1.8. If (I1)–(I6) hold, then Q : P → P is a completely continuous
operator.
Proof. Let (u, v) ∈ P be an arbitrary element. Because Q1 (u, v) and Q2 (u, v)
satisfy problem (1.1)–(1.2) for y(t) = λp(t)f (t, u(t), v(t)), t ∈ [0, 1], and problem
(1.8)–(1.9) for h(t) = μq(t)g(t, u(t), v(t)), t ∈ [0, 1], respectively, then by Lemma
1.1.7, we obtain
Systems of second-order ordinary differential equations 11
Hence, we conclude
inf [Q1 (u, v)(t)+Q2 (u, v)(t)] ≥ inf Q1 (u, v)(t) + inf Q2 (u, v)(t)
t∈[σ ,1−σ ] t∈[σ ,1−σ ] t∈[σ ,1−σ ]
≥ν1 Q1 (u, v) + ν2 Q2 (u, v)
≥ν (Q1 (u, v), Q2 (u, v)) Y = ν Q(u, v) Y.
By (I1)–(I6) and Lemma 1.1.5, we obtain Q1 (u, v)(t) ≥ 0, Q2 (u, v)(t) ≥ 0 for all
t ∈ [0, 1], and so we deduce that Q(u, v) ∈ P. Hence, we get Q(P) ⊂ P. By using
standard arguments, we can easily show that Q1 and Q2 are completely continuous,
and then Q is a completely continuous operator.
1−σ 1 1−σ
We denote A = σ J1 (s)p(s) ds, B = 0 J1 (s)p(s) ds, C = σ J2 (s)q(s) ds,
1
and D = 0 J2 (s)q(s) ds.
i , gi ∈ (0, ∞) and numbers α , α ≥ 0 and α̃ , α̃ > 0 such that
First, for f0s , gs0 , f∞ ∞ 1 2 1 2
α1 + α2 = 1 and α̃1 + α̃2 = 1, we define the numbers L1 , L2 , L3 , L4 , L2 , and L4 by
α1 α̃1 α2 α̃2
L1 = , L2 = s , L3 = , L4 = ,
νν1 f∞ A
i f0 B νν2 gi∞ C gs0 D
1 1
L2 = s , L4 = s .
f0 B g0 D
Theorem 1.1.2. Assume that (I1)–(I6) hold, and α1 , α2 ≥ 0 and α̃1 , α̃2 > 0 such
that α1 + α2 = 1 and α̃1 + α̃2 = 1.
i , gi ∈ (0, ∞), L < L , and L < L , then for each λ ∈ (L , L ) and μ ∈
(1) If f0s , gs0 , f∞ ∞ 1 2 3 4 1 2
(L3 , L4 ) there exists a positive solution (u(t), v(t)), t ∈ [0, 1] for (S)–(BC).
(2) If f0s = 0, gs0 , f∞ i , gi ∈ (0, ∞), and L < L , then for each λ ∈ (L , ∞) and μ ∈ (L , L )
∞ 3 4 1 3 4
there exists a positive solution (u(t), v(t)), t ∈ [0, 1] for (S)–(BC).
(3) If gs0 = 0, f0s , f∞ i , gi ∈ (0, ∞), and L < L , then for each λ ∈ (L , L ) and μ ∈ (L , ∞)
∞ 1 2 1 2 3
there exists a positive solution (u(t), v(t)), t ∈ [0, 1] for (S)–(BC).
(4) If f0s = gs0 = 0 and f∞ i , gi ∈ (0, ∞), then for each λ ∈ (L , ∞) and μ ∈ (L , ∞) there
∞ 1 3
exists a positive solution (u(t), v(t)), t ∈ [0, 1] for (S)–(BC).
(5) If {f0s , gs0 , f∞
i ∈ (0, ∞), gi = ∞} or {f s , gs , gi ∈ (0, ∞), f i = ∞} or {f s , gs ∈ (0, ∞),
∞ 0 0 ∞ ∞ 0 0
i = gi = ∞}, then for each λ ∈ (0, L ) and μ ∈ (0, L ) there exists a positive solution
f∞ ∞ 2 4
(u(t), v(t)), t ∈ [0, 1] for (S)–(BC).
(6) If {f0s = 0, gs0 , f∞ i ∈ (0, ∞), gi = ∞} or {f s = 0, f i = ∞, gs , gi ∈ (0, ∞)} or {f s = 0,
∞ 0 ∞ 0 ∞ 0
g0 ∈ (0, ∞), f∞ = gi∞ = ∞}, then for each λ ∈ (0, ∞) and μ ∈ (0, L4 ) there exists a
s i
Proof. We consider the above cone P ⊂ Y and the operators Q1 , Q2 , and Q. Be-
cause the proofs of the above cases are similar, in what follows we shall prove one of
them—namely, case (2). So, we suppose f0s = 0, gs0 , f∞i , gi ∈ (0, ∞), and L < L .
∞ 3 4
α1 α2
Let λ ∈ (L1 , ∞) and μ ∈ (L3 , L4 )—that is, λ ∈ i A, ∞
νν1 f∞
and μ ∈ , 1
νν2 gi∞ C gs0 D
.
We choose α̃2 ∈ (μgs0 D, 1). Let α̃1 = 1 − α̃2 and let ε > 0 be a positive number such
that ε < min{f∞i , gi } and
∞
α1 α2 α̃1 α̃2
≤ λ, ≤ μ, ≥ λ, ≥ μ.
νν1 (f∞ − ε)A
i νν2 (g∞ − ε)C
i εB (gs0 + ε)D
By using (I6) and the definitions of f0s and gs0 , we deduce that there exists R1 > 0
such that f (t, u, v) ≤ ε(u + v) and g(t, u, v) ≤ (gs0 + ε)(u + v) for all t ∈ [0, 1] and
u, v ≥ 0, with 0 ≤ u + v ≤ R1 . We define the set 1 = {(u, v) ∈ Y, (u, v) Y < R1 }.
Now let (u, v) ∈ P ∩ ∂ 1 —that is, (u, v) ∈ P with (u, v) Y = R1 or equivalently
u + v = R1 . Then u(t) + v(t) ≤ R1 for all t ∈ [0, 1], and by Lemma 1.1.6, we
obtain
1
Q1 (u, v)(t) ≤ λ J1 (s)p(s)f (s, u(s), v(s)) ds
0
1
≤λ J1 (s)p(s)ε(u(s) + v(s)) ds
0
1
≤ λε J1 (s)p(s)( u + v ) ds
0
= λεB (u, v) Y ≤ α̃1 (u, v) Y, ∀ t ∈ [0, 1].
Therefore, Q1 (u, v) ≤ α̃1
(u, v) Y . In a similar manner, we conclude
1
Q2 (u, v)(t) ≤ μ J2 (s)q(s)g(s, u(s), v(s)) ds
0
1
≤μ J2 (s)q(s)(gs0 + ε)(u(s) + v(s)) ds
0
1
≤ μ(gs0 + ε) J2 (s)q(s)( u + v ) ds
0
= μ(gs0 + ε)D (u, v) Y ≤ α̃2 (u, v) Y, ∀ t ∈ [0, 1].
Therefore, Q2 (u, v) ≤ α̃2 (u, v) Y .
Then for (u, v) ∈ P ∩ ∂ 1,we deduce
for all t ∈ [σ , 1 − σ ].
Then, by Lemma 1.1.6, we conclude
1
Q1 (u, v)(σ ) ≥ λν1 J1 (s)p(s)f (s, u(s), v(s)) ds
0
1−σ
≥ λν1 J1 (s)p(s)f (s, u(s), v(s)) ds
σ
1−σ
i
≥ λν1 J1 (s)p(s)(f∞ − ε)(u(s) + v(s)) ds
σ
i
≥ λν1 (f∞ − ε)Aν (u, v) Y ≥ α1 (u, v) Y.
α̃1 , α̃2 > 0 such that α1 + α2 = 1 and α̃1 + α̃2 = 1, we define the numbers L̃1 ,
L̃2 , L̃3 , L̃4 , L̃2 , and L̃4 by
14 Boundary Value Problems for Systems of Differential, Difference and Fractional Equations
α1 α̃1 α2 α̃2
L̃1 = , L̃2 = , L̃3 = , L̃4 = ,
νν1 f0i A s B
f∞ νν2 gi0 C gs∞ D
1 1
L̃2 = s B
, L̃4 = .
f∞ gs∞ D
Theorem 1.1.3. Assume that (I1)–(I6) hold, and α1 , α2 ≥ 0 and α̃1 , α̃2 > 0 such
that α1 + α2 = 1 and α̃1 + α̃2 = 1.
s , gs ∈ (0, ∞), L̃ < L̃ , and L̃ < L̃ , then for each λ ∈ (L̃ , L̃ ) and μ ∈
(1) If f0i , gi0 , f∞ ∞ 1 2 3 4 1 2
(L̃3 , L̃4 ) there exists a positive solution (u(t), v(t)), t ∈ [0, 1] for (S)–(BC).
s ∈ (0, ∞), gs = 0, and L̃ < L̃ , then for each λ ∈ (L̃ , L̃ ) and μ ∈ (L̃ , ∞)
(2) If f0i , gi0 , f∞ ∞ 1 2 1 2 3
there exists a positive solution (u(t), v(t)), t ∈ [0, 1] for (S)–(BC).
s = 0, and L̃ < L̃ , then for each λ ∈ (L̃ , ∞) and μ ∈ (L̃ , L̃ )
(3) If f0i , gi0 , gs∞ ∈ (0, ∞), f∞ 3 4 1 3 4
there exists a positive solution (u(t), v(t)), t ∈ [0, 1] for (S)–(BC).
(4) If f0i , gi0 ∈ (0, ∞) and f∞ s = gs = 0, then for each λ ∈ (L̃ , ∞) and μ ∈ (L̃ , ∞) there
∞ 1 3
exists a positive solution (u(t), v(t)), t ∈ [0, 1] for (S)–(BC).
(5) If {f0i = ∞, gi0 , f∞ s , gs ∈ (0, ∞)} or {f i , f s , gs ∈ (0, ∞), gi = ∞} or {f i = gi = ∞,
∞ 0 ∞ ∞ 0 0 0
f∞ , g∞ ∈ (0, ∞)}, then for each λ ∈ (0, L̃2 ) and μ ∈ (0, L̃4 ) there exists a positive solution
s s
Proof. We consider the above cone P ⊂ Y and the operators Q1 , Q2 , and Q. Be-
cause the proofs of the above cases are similar, in what follows we shall prove one
of them—namely, the second case of (6). So, we suppose f0i , f∞
s ∈ (0, ∞), gi = ∞,
0
and gs∞ = 0. Let λ ∈ (0, L̃2 )—that is, λ ∈ 0, f s1B —and μ ∈ (0, ∞). We choose
∞
α1 > 0, α1 < min{λνν1 f0i A, 1}, and α̃1 ∈ (λf∞
s B, 1). Let α = 1−α and α̃ = 1− α̃ ,
2 1 2 1
and let ε > 0 be a positive number such that ε < f0i and
By using (I6) and the definitions of f0i and gi0 , we deduce that there exists R3 > 0
such that f (t, u, v) ≥ (f0i − ε)(u + v) and g(t, u, v) ≥ 1ε (u + v) for all u, v ≥ 0, with
0 ≤ u + v ≤ R3 , and t ∈ [σ , 1 − σ ]. We denote 3 = {(u, v) ∈ Y, (u, v) Y < R3 }.
Let (u, v) ∈ P with (u, v) Y = R3 —that is, u + v = R3 . Because u(t) +
v(t) ≤ u + v = R3 for all t ∈ [0, 1], then, by using Lemma 1.1.6, we obtain
Systems of second-order ordinary differential equations 15
1−σ
Q1 (u, v)(σ ) ≥ λν1 J1 (s)p(s)f (s, u(s), v(s)) ds
σ
1−σ
≥ λν1 J1 (s)p(s)(f0i − ε)(u(s) + v(s)) ds
σ
1−σ
≥ λνν1 (f0i − ε) J1 (s)p(s)( u + v ) ds
σ
1.1.4 Examples
Let a(t) = 1, b(t) = 4, c(t) = 1, d(t) = 1, p(t) = 1, and q(t) = 1 for all t ∈ [0, 1],
α = 1, β = 3, γ = 1, δ = 1, α̃ = 3, β̃ = 2, γ̃ = 1, δ̃ = 3/2,
⎧
⎨ 0, t ∈ [0, 1/3),
H1 (t) = t2 , H2 (t) = 7/2, t ∈ [1/3, 2/3),
⎩
11/2, t ∈ [2/3, 1],
0, t ∈ [0, 1/2),
K1 (t) = K2 (t) = t3 .
4/3, t ∈ [1/2, 1],
1 1 1
0 u(s) dH2 (s) = 2 u 3 +2u 3 , 0 u(s) dK1 (s) = 43 u 0 u(s) dH1 (s) =
7 1 2 1
Then 2 ,
1 1 1 2
2 0 su(s) ds, and 0 u(s) dK2 (s) = 3 0 s u(s) ds.
We consider the second-order differential system
u (t) − 4u(t) + λf (t, u(t), v(t)) = 0, t ∈ (0, 1),
(S)
v (t) − v(t) + μg(t, u(t), v(t)) = 0, t ∈ (0, 1),
Systems of second-order ordinary differential equations 17
1
7 1 2
1 := τ1 − ψ(s) dH2 (s) = τ1 − ψ + 2ψ
0 2 3 3
1
= (21 e − 5)/(4 e ) − 2
4 2
s(1 + 3 e4−4s )/(4 e2−2s ) ds ≈ 36.83556247 > 0,
0
1 1 1
3 := ψ(s) dH1 (s) = 2 sψ(s) ds = 2 s(7 e4s + 5)/(4 e2s ) ds ≈ 7.71167043,
0 0 0
1
7 1 2
4 := φ(s) dH2 (s) = φ + 2φ
0 2 3 3
1 = 1 2 − 3 4 ≈ 284.07473844 > 0.
The functions g1 and J1 are given by
1 φ(t)ψ(s), 0 ≤ s ≤ t ≤ 1,
g1 (t, s) = or
τ1 φ(s)ψ(t), 0 ≤ t ≤ s ≤ 1,
⎧
⎪ (1 + 3 e4−4t )(7 e4s + 5)
⎪
⎪ , 0 ≤ s ≤ t ≤ 1,
4 e2 ⎨ 16 e2−2t+2s
g1 (t, s) =
21 e4 − 5 ⎪
⎪
⎩ (1 + 3 e
⎪ 4−4s )(7 e4t + 5)
, 0 ≤ t ≤ s ≤ 1,
16 e2−2s+2t
18 Boundary Value Problems for Systems of Differential, Difference and Fractional Equations
1
1
J1 (s) = g1 (s, s) + (4 ψ(1) + 1 φ(0)) × 2 τ g1 (τ , s) dτ
1 0
1 7 1 2
+ (2 ψ(1) + 3 φ(0)) g1 , s + 2g1 ,s
1 2 3 3
s 1
1
= g1 (s, s) + (4 ψ(1) + 1 φ(0)) × 2 τ g1 (τ , s) dτ + τ g1 (τ , s) dτ
1 0 s
1 7 1 2
+ (2 ψ(1) + 3 φ(0)) g1 , s + 2g1 ,s
1 2 3 3
⎧ s 1
⎪
⎪ 1 1
⎪
⎪ φ(s)ψ(s)+ (4 ψ(1)+1 φ(0)) × 2 φ(s) τ ψ(τ ) dτ + ψ(s) τ φ(τ ) dτ
⎪
⎪ τ1 1
⎪
⎪
0 s
⎪
⎪
⎪
⎪ 1 7 1 2 1
⎪
⎪ + (2 ψ(1)+3 φ(0)) φ ψ(s)+2φ ψ(s) , 0 ≤ s < ,
⎪
⎪ 1
⎪
⎪
2 3 3 3
⎪
⎪ s 1
⎪
⎪ 1 1
⎪
⎪ φ(s)ψ(s)+ (4 ψ(1)+1 φ(0)) × 2 φ(s) τ ψ(τ ) dτ +ψ(s) τ φ(τ ) dτ
⎨
τ1 1 0 s
=
⎪
⎪ 1 7 1 2 1 2
⎪
⎪ + (2 ψ(1) + 3 φ(0)) φ(s)ψ + 2φ ψ(s) , ≤s< ,
⎪
⎪
⎪
⎪ 1 2 3 3 3 3
⎪
⎪ s 1
⎪
⎪
⎪
⎪ 1 1
⎪
⎪ φ(s)ψ(s)+ (4 ψ(1)+1 φ(0)) × 2 φ(s) τ ψ(τ ) dτ +ψ(s) τ φ(τ ) dτ
⎪
⎪ τ1 1
⎪
⎪
0
s
⎪
⎪
⎪
⎪ 1 7 1 2 2
⎩ + (2 ψ(1) + 3 φ(0)) φ(s)ψ + 2φ(s)ψ , ≤ s ≤ 1.
1 2 3 3 3
The functions ψ̃ and φ̃ from Section 1.1.2 are the solutions of the following
problems:
⎧ ⎧
⎨ ψ̃ (t) − ψ̃(t) = 0, 0 < t < 1, ⎨ φ̃ (t) − φ̃(t) = 0, 0 < t < 1,
⎪
⎩ ⎪
⎩ φ̃(1) = 3 ,
ψ̃(0) = 2, ψ̃ (0) = 3, φ̃ (1) = −1.
2
We obtain ψ̃(t) = 5 e2 e−1 e 25 e2 +1
2t 2−2t
t and φ̃(t) = 1+5
4 e1−t
for all t ∈ [0, 1], τ2 = 4e ,
1 1
˜ 1 := τ2 −
ψ̃(s) dK2 (s) = τ2 − 3 s2 ψ̃(s) ds
0 0
1
2
= (25 e + 1)/(4 e) − 3 s (5 e − 1)/(2 es ) ds ≈ 11.93502177 > 0,
2 2s
0
1
˜ 4 1
2 := τ2 − φ̃(s) dK1 (s) = τ2 − φ̃
0 3 2
= (25 e2 + 1)/(4 e) − (1 + 5 e)/(3 e1/2 ) ≈ 14.13118562 > 0,
1
4 1
˜ 3 :=
ψ̃(s) dK1 (s) = ψ̃ = (10 e − 2)/(3 e1/2 ) ≈ 5.09138379,
0 3 2
Systems of second-order ordinary differential equations 19
1 1 1
˜ 4 :=
φ̃(s) dK2 (s) = 3 s2 φ̃(s) ds = 3 s2 (1 + 5 e2−2s )/(4 e1−s ) ds ≈ 1.83529455,
0 0 0
˜ 1
2 = ˜ 3
˜2 − ˜ 4 ≈ 159.31181898 > 0.
1
1 ˜ ˜ 3 φ̃(0)) × 3
+ (2 ψ̃(1) + τ 2 g2 (τ , s) dτ
2 0
1 4 1
= g2 (s, s) + ˜ 4 ψ̃(1) +
( ˜ 1 φ̃(0)) g2 ,s
2 3 2
s 1
1
+ ˜ 2 ψ̃(1) +
( ˜ 3 φ̃(0)) × 3 τ 2 g2 (τ , s) dτ + τ 2 g2 (τ , s) dτ
2 0 s
⎧
⎪ 1 1 4 1
⎪
⎪ φ̃(s)ψ̃(s) + ˜ 4 ψ̃(1) +
˜ 1 φ̃(0) × φ̃ ψ̃(s)
⎪
⎪ τ2 2 3 2
⎪
⎪
⎪
⎪
⎪
⎪ 1 ˜ ˜ 3 φ̃(0)
⎪
⎪ + 2 ψ̃(1) +
⎪
⎪ 2
⎪
⎪
⎪
⎪
⎪
⎪ ×3 φ̃(s)
s 2 1 2
⎪
⎨ 0 τ ψ̃(τ ) dτ + ψ̃(s) s τ φ̃(τ ) dτ , 0 ≤ s < 12 ,
=
⎪
⎪ 1 1
˜ 4 ψ̃(1) +
4
˜ 1 φ̃(0) × φ̃(s)ψ̃
1
⎪
⎪ φ̃(s)ψ̃(s) +
⎪
⎪ τ2 2 3 2
⎪
⎪
⎪
⎪
⎪
⎪ 1
˜ 2 ψ̃(1) +
˜ 3 φ̃(0)
⎪
⎪ +
⎪
⎪ 2
⎪
⎪
⎪
⎪ s
⎪
⎪
1 1
⎩ ×3 φ̃(s) τ ψ̃(τ ) dτ + ψ̃(s)
2
τ φ̃(τ ) dτ
2
, ≤ s ≤ 1.
0 s 2
20 Boundary Value Problems for Systems of Differential, Difference and Fractional Equations
α2 (1 + 3 e4 )(5 e2 − 1) α2
L3 = , L4 = .
e3/2 (1 + 3 e)(5 e1/2 − 1)p2 (q2 − 1)C (q2 + 1)D
The conditions L1 < L2 and L3 < L4 become
1
u(t) = Q1 (u, v)(t) = λ G1 (t, s)p(s)f (s, u(s), v(s)) ds
0
1 1
≤λ J1 (s)p(s)f (s, u(s), v(s)) ds ≤ λM1 J1 (s)p(s)(u(s) + v(s)) ds
0 0
1
≤ λM1 ( u + v ) J1 (s)p(s) ds = λM1 B (u, v) Y, ∀ t ∈ [0, 1].
0
Therefore, we conclude
1
u ≤ λM1 B (u, v) Y < λ0 M1 B (u, v) Y = (u, v) Y.
2
In a similar manner, we have
1
v(t) = Q2 (u, v)(t) = μ G2 (t, s)q(s)g(s, u(s), v(s)) ds
0
1 1
≤μ J2 (s)q(s)g(s, u(s), v(s)) ds ≤ μM2 J2 (s)q(s)(u(s) + v(s)) ds
0 0
1
≤ μM2 ( u + v ) J2 (s)q(s) ds = μM2 D (u, v) Y, ∀ t ∈ [0, 1].
0
Therefore, we conclude
1
v ≤ μM2 D (u, v) Y < μ0 M2 D (u, v) Y = (u, v) Y.
2
Hence, (u, v) Y = u + v < 12 (u, v) Y + 12 (u, v) Y = (u, v) Y , which is
a contradiction. So, the boundary value problem (S)–(BC) has no positive solution.
1
u(σ ) = Q1 (u, v)(σ ) = λ G1 (σ , s)p(s)f (s, u(s), v(s)) ds
0
1−σ
≥λ G1 (σ , s)p(s)f (s, u(s), v(s)) ds
σ
1−σ
≥ λm1 G1 (σ , s)p(s)(u(s) + v(s)) ds
σ
1−σ
≥ λm1 ν1 J1 (s)p(s)ν( u + v ) ds = λm1 νν1 A (u, v) Y.
σ
Therefore, we deduce
u ≥ u(σ ) ≥ λm1 νν1 A (u, v) Y > λ̃0 m1 νν1 A (u, v) Y = (u, v) Y,
Therefore, we deduce
v ≥ v(σ ) ≥ μm2 νν2 C (u, v) Y > μ̃0 m2 νν2 C (u, v) Y = (u, v) Y,
and so
(u, v) Y = u + v ≥ λm1 νν1 A (u, v) Y + μm2 νν2 C (u, v) Y
1 1
=
(u, v) Y + (u, v) Y = (u, v) Y ,
2 2
which is a contradiction. Therefore, the boundary value problem (S)–(BC) has no
positive solution.
1.2.2 An example
Example 1.2.1. We consider the first example from Section 1.1.4—that is, problem
(S)–(BC) with f and g given in Example 1.1.1. Because f0s = eq1 , f∞ s = ep (q + 1),
1 1
g0 = q2 + 1, and g∞ = p2 (q2 + 1) are finite, we can apply Theorem 1.2.1. Using
s s
the same values for p1 , q1 , p2 , and q2 as in Example 1.1.1 from Section 1.1.4—that
is, p1 = 128, q1 = 2, p2 = 117, and q2 = 2, we deduce
f (t, u, v) [p1 (u+v)+1](q1 +sin v)
M1 = sup max = e sup ≈ 1043.82022212,
u,v≥0 t∈[0,1] u + v u,v≥0 u+v +1
We recall now some theorems concerning the fixed point index theory. Let E be
a real Banach space, P ⊂ E a cone, “≤” the partial ordering defined by P, and θ the
zero element in E. For > 0, let B = {u ∈ E, u < } be the open ball of radius
centered at 0, and let its boundary be ∂B = {u ∈ E, u = }. The proofs of our
results are based on the following fixed point index theorems:
Theorem 1.3.1 (Amann, 1976). Let A : B̄ ∩ P → P be a completely continuous
operator which has no fixed point on ∂B ∩ P. If Au ≤ u for all u ∈ ∂B ∩ P,
then i(A, B ∩ P, P) = 1.
Theorem 1.3.2 (Amann, 1976). Let A : B̄ ∩ P → P be a completely continuous
operator. If there exists u0 ∈ P \ {θ} such that u − Au = λu0 for all λ ≥ 0 and
u ∈ ∂B ∩ P, then i(A, B ∩ P, P) = 0.
Theorem 1.3.3 (Zhou and Xu, 2006). Let A : B̄ ∩ P → P be a completely contin-
uous operator which has no fixed point on ∂B ∩ P. If there exists a linear operator
L : P → P and u0 ∈ P \ {θ} such that
(1) u0 ≤ Lu0 and (2) Lu ≤ Au, ∀ u ∈ ∂B ∩ P,
then i(A, B ∩ P, P) = 0.
We shall suppose that assumptions (A1)–(A5) from Section 1.1.2 hold, and we
shall use in our main results from this section Lemmas 1.1.1–1.1.7 from
Section 1.1.2.
⎧
⎪
⎪
1 1
⎪
⎨ u(t) = G1 (t, s)f s, G2 (s, τ )g(τ , u(τ )) dτ ds, t ∈ [0, 1],
0 0
⎪ 1
⎪
⎪
⎩ v(t) = G2 (t, s)g(s, u(s)) ds, t ∈ [0, 1].
0
We consider the Banach space X = C([0, 1]) with the supremum norm · , and
define the cone P ⊂ X by P = {u ∈ X, u(t) ≥ 0, ∀ t ∈ [0, 1]}.
We also define the operator A : P → X by
1 1
(Au)(t) = G1 (t, s)f s, G2 (s, τ )g(τ , u(τ )) dτ ds, t ∈ [0, 1],
0 0
Proof. For σ from (H6), we define the cone P0 = {u ∈ P ; inft∈[σ ,1−σ ] u(t) ≥
ν u }, where ν = min{ν1 , ν2 }, and ν1 and ν2 are defined in Section 1.1.2. From our
assumptions and Lemma 1.1.7, we obtain B (P ) ⊂ P0 and C (P ) ⊂ P0 . Now we
1
consider the function u0 (t) = 0 G1 (t, s) ds = (B y0 )(t), t ∈ [0, 1], where y0 (t) = 1
for all t ∈ [0, 1], and the set
M = {u ∈ P ; there exists λ ≥ 0 such that u = Au + λu0 }.
We shall show that M ⊂ P0 and that M is a bounded subset of X. If u ∈ M, then
there exists λ ≥ 0 such that u(t) = (Au)(t) + λu0 (t), t ∈ [0, 1]. Hence, we have
u(t) = (Au)(t) + λ(B y0 )(t) = B (Fu(t)) + λ(B y0 )(t) = B (Fu(t) + λy0 (t)) ∈ P0 ,
1
u ≤ inf u(t), ∀ u ∈ M. (1.15)
ν t∈[σ ,1−σ ]
From (1) of assumption (H6), we deduce that there exist C1 , C2 > 0 such that
f (t, u) ≥ C1 up − C2 , ∀ (t, u) ∈ [σ , 1 − σ ] × [0, ∞). (1.16)
If i
f̃∞ ∈ (0, ∞), then we obtain
f (t, u)
∀ ε > 0 ∃ δε > 0 such that ∀ u ≥ δε we have ≥ f̃∞
i
− ε, ∀ t ∈ [σ , 1 − σ ].
up
We choose ε ∈ (0, f̃∞ i ), and we denote C = f̃ i − ε > 0. Then from the above
1 ∞
relation, we deduce that there exists δ0 > 0 such that f (t, u) ≥ C1 up for all u ≥ δ0 and
t ∈ [σ , 1 − σ ]. Because f (t, u) ≥ 0 ≥ C1 up − C2 for all u ∈ [0, δ0 ] and t ∈ [σ , 1 − σ ],
p
with C2 = C1 δ0 , then we obtain relation (1.16).
i = ∞—that is, lim f (t,u)
If f̃∞ u→∞ inft∈[σ ,1−σ ] up = ∞—then we obtain
f (t, u)
∀ε > 0 ∃ δε > 0 such that ∀u ≥ δε we have ≥ ε, ∀ t ∈ [σ , 1−σ ].
up
For ε = 1, we deduce that there exists δ̃0 > 0 such that f (t, u) ≥ up for all u ≥ δ̃0 and
t ∈ [σ , 1 − σ ]. Because f (t, u) ≥ 0 ≥ up − C2 for all u ∈ [0, δ̃0 ] and t ∈ [σ , 1 − σ ],
p
with C2 = δ̃0 , then we obtain relation (1.16) with C1 = 1. Therefore, in both cases,
we obtain inequality (1.16).
From (2) of assumption (H6), we have limu→∞ inft∈[σ ,1−σ ] g(t,u)
u1/p
= ∞—that is,
g(t, u)
∀ε > 0 ∃ uε > 0 such that ∀ u ≥ uε we have ≥ ε, ∀ t ∈ [σ , 1−σ ].
u1/p
p 1−σ
We consider ε0 = 2/(C1 ν1 ν2 m1 m2 ) > 0, where m1 = σ J1 (τ ) dτ > 0 and
1−σ
m2 = σ (J2 (τ ))p dτ > 0. Then from the above relation, we deduce that there
exists u0 > 0 such that (g(t, u))p ≥ ε0 u for all (t, u) ∈ [σ , 1 − σ ] × [u0 , ∞). For
p/q0
1−σ 1 1 1
≥ G1 (t, s) C1 (G2 (s, τ )g(τ , u(τ ))) dτ p
dτ ds − C2 J1 (s) ds
σ 0 0 0
1−σ 1−σ
≥ C1 G1 (t, s) (G2 (s, τ ))p (g(τ , u(τ )))p dτ ds − C4 , ∀ t ∈ [0, 1],
σ σ
1
where q0 = p/(p − 1) and C4 = C2 0 J1 (s) ds.
The above inequality is also valid for p = 1. Therefore, for u ∈ P and p ∈ (0, 1]
we have
1−σ 1−σ
(Au)(t) ≥ C1 G1 (t, s) (G2 (s, τ ))p (g(τ , u(τ )))p dτ ds−C4 , ∀ t ∈ [0, 1].
σ σ
(1.18)
Next, for u ∈ M and t ∈ [σ , 1 − σ ], by using Lemma 1.1.6 and relations (1.17) and
(1.18), we obtain
u(t) = (Au)(t) + λu0 (t) ≥ (Au)(t)
1−σ 1−σ
≥ C1 G1 (t, s) (G2 (s, τ )) (g(τ , u(τ ))) dτ ds − C4
p p
σ σ
1−σ 1−σ
p
≥ C1 ν1 ν2 J1 (s) ds (J2 (τ ))p (ε0 u(τ ) − C3 ) dτ − C4
σ σ
1−σ 1−σ
p
≥ C1 ν1 ν2 ε0 J1 (s) ds (J2 (τ ))p dτ inf u(τ ) − C5
σ σ τ ∈[σ ,1−σ ]
=2 inf u(τ ) − C5 ,
τ ∈[σ ,1−σ ]
p
where C5 = C4 + C3 C1 ν1 ν2 m1 m2 > 0.
Hence, inft∈[σ ,1−σ ] u(t) ≥ 2 inft∈[σ ,1−σ ] u(t) − C5 , and so
inf u(t) ≤ C5 , ∀ u ∈ M. (1.19)
t∈[σ ,1−σ ]
From relations (1.15) and (1.19) we obtain u ≤ C5 /ν for all u ∈ M—that is, M
is a bounded subset of X.
Moreover, there exists a sufficiently large R1 (R1 ≥ 1) such that
u = Au + λu0 , ∀ u ∈ ∂BR1 ∩ P , ∀ λ ≥ 0.
From Theorem 1.3.2 we deduce that the fixed point index of the operator A is
i(A, BR1 ∩ P , P ) = 0. (1.20)
Next, from (1) of assumption (H7) we have
f (t, u)
∀ ε > 0 ∃ uε > 0 such that ≤ f̃0s + ε, ∀ (t, u) ∈ [0, 1] × (0, uε ].
uq1
For ε = 1 there exists ũ0 > 0 such that f u(t,u)
q1 ≤ (f̃0 + 1) for all (t, u) ∈ [0, 1] × (0, ũ0].
s
q
If ũ0 ≥ 1, then the last inequality is true for every u ∈ (0, 1]. If ũ0 < 1, then uq1 > ũ01
30 Boundary Value Problems for Systems of Differential, Difference and Fractional Equations
and f (t, u) ≤ m̃ for all u ∈ (ũ0 , 1] and t ∈ [0, 1], with m̃ > 0. Therefore, for u ∈ (ũ0 , 1]
and t ∈ [0, 1] we obtain f u(t,u)
q1 ≤ q1 . So, we deduce
m̃
ũ0
f (t, u) m̃
≤ M0 , ∀ (t, u) ∈ [0, 1] × (0, 1], where M0 = max f̃0s + 1, q1 > 0,
uq1 ũ0
and using also (H5), we obtain
f (t, u) ≤ M0 uq1 , ∀ (t, u) ∈ [0, 1] × [0, 1]. (1.21)
From (2) of assumption (H7) and from (H5) we have
∀ε > 0 ∃ δε > 0 such that g(t, u) ≤ εuq2 , ∀ (t, u) ∈ [0, 1] × [0, δε ].
q 1
We consider ε1 = min 1/M2 , (1/(2M0 M1 M2 1 ))1/q1 > 0, where M1 = 0 J1 (s) ds >
1
0 and M2 = 0 J2 (s) ds > 0. Then we deduce that there exists r1 ∈ (0, 1) such that
g(t, u) ≤ ε1 uq2 , ∀ (t, u) ∈ [0, 1] × [0, r1 ]. (1.22)
Hence, by (1.22), for any u ∈ B̄r1 ∩ P and t ∈ [0, 1] we obtain
1 1
G2 (t, s)g(s, u(s)) ds ≤ ε1 J2 (s)(u(s))q2 ds ≤ ε1 M2 u q2
≤ 1. (1.23)
0 0
Therefore, by (1.21) and (1.23), we deduce that for any u ∈ B̄r1 ∩ P and t ∈ [0, 1]
1 1 q1
(Au)(t) ≤ M0 G1 (t, s) G2 (s, τ )g(τ , u(τ )) dτ ds
0 0
q q q q
≤ M0 ε11 M1 M2 1 u q1 q2
≤ M0 ε11 M1 M2 1 u ≤ u /2.
This gives us Au ≤ u /2 for all u ∈ ∂Br1 ∩ P . From Theorem 1.3.1 we conclude
that the fixed point index of A is
i(A, Br1 ∩ P , P ) = 1. (1.24)
Combining (1.20) and (1.24), we obtain
i(A, (BR1 \ B̄r1 ) ∩ P , P ) = i(A, BR1 ∩ P , P ) − i(A, Br1 ∩ P , P ) = −1.
Hence, we deduce that A has at least one fixed point u1 ∈ (BR1 \ B̄r1 ) ∩ P —that is,
r1 < u1 < R1 . Let v1 (t) = 0 G2 (t, s)g(s, u1 (s)) ds. Then (u1 , v1 ) ∈ P × P is
1
a solution of (S )–(BC). By using (H5), we also have v1 > 0. If we suppose that
v1 (t) = 0 for all t ∈ [0, 1], then by using (H5), we have f (s, v1 (s)) = f (s, 0) = 0 for
all s ∈ [0, 1]. This implies u1 (t) = 0 for all t ∈ [0, 1], which contradicts u1 > 0.
The proof of Theorem 1.3.4 is completed.
Theorem 1.3.5. Assume that (H1)–(H5) hold. If the functions f and g also satisfy
the following conditions (H8) and (H9), then problem (S )–(BC) has at least one
positive solution (u(t), v(t)), t ∈ [0, 1]:
Systems of second-order ordinary differential equations 31
[u0 , ∞). Because for u ∈ [0, u0 ] and t ∈ [0, 1] there exists C7 > 0 such that f (t, u) ≤
C7 , we obtain relation (1.25) with C6 = f̃∞s + 1 > 0.
(Au)(t) ≤ 2α1 C6 M1 M2α1 ε2α1 u +2α1 C6 M1 M2α1 C8α1 +M1 C7 =: ( u ), ∀ t ∈ [0, 1].
(1.27)
Because lim u →∞ ( u )/ u = 1/2, there exists a large R2 (R2 ≥ 1) such that
3
( u ) ≤ u , ∀ u ∈ P , u ≥ R2 . (1.28)
4
Hence, from (1.27) and (1.28) we deduce Au ≤ 34 u for all u ∈ ∂BR2 ∩ P , and
from Theorem 1.3.1 we have that the fixed point index of the operator A is
i(A, BR2 ∩ P , P ) = 1. (1.29)
On the other hand, from (H9) (1) we deduce that there exist positive constants
C9 > 0 and ũ1 > 0 such that
f (t, u) ≥ C9 u, ∀ (t, u) ∈ [σ , 1 − σ ] × [0, ũ1 ]. (1.30)
If f̃0i ∈ (0, ∞), then we obtain
f (t, u)
∀ε > 0 ∃ uε > 0 such that ≥ f̃0i −ε, ∀ (t, u) ∈ [σ , 1−σ ]×(0, uε ].
u
We choose ε > 0 such that f̃0i − ε > 0; we denote C9 = f̃0i − ε. Then there exists
ũ2 > 0 such that f (t, u) ≥ C9 u for all (t, u) ∈ [σ , 1 − σ ] × [0, ũ2]. We obtain relation
(1.30) with ũ1 = ũ2 .
If f̃0i = ∞, then limu→0+ inft∈[σ ,1−σ ] f (t,u)
u = ∞, and we have
f (t, u)
∀ε > 0 ∃ uε > 0 such that ≥ ε, ∀ (t, u) ∈ [σ , 1 − σ ] × (0, uε ].
u
For ε = 1 there exists û2 such that f (t, u) ≥ u for all (t, u) ∈ [σ , 1 − σ ] × [0, û2]. We
obtain relation (1.30) with C9 = 1 and ũ1 = û2 .
From (H9) (2) we have g̃i0 = limu→0+ inft∈[σ ,1−σ ] g(t,u) u = ∞, and so
g(t, u)
∀ε > 0 ∃ uε > 0 such that ≥ ε, ∀ (t, u) ∈ [σ , 1 − σ ] × (0, uε ].
u
1−σ
For ε = C0 /C9 with C0 = 1/(ν1 ν2 m1 m3 ) > 0 and m3 = σ J2 (τ ) dτ , we deduce
that there exists û1 > 0 such that g(t, u) ≥ C
C9 u for all (t, u) ∈ [σ , 1 − σ ] × [0, û1 ].
0
From (1.31), (1.32), and Lemma 1.1.6, we deduce that for any u ∈ B̄r2 ∩ P we
have
1−σ 1−σ
(Au)(t) ≥ G1 (t, s)f s, G2 (s, τ )g(τ , u(τ )) dτ ds
σ σ
1−σ 1−σ
≥ C9 G1 (t, s) G2 (s, τ )g(τ , u(τ )) dτ ds
σ σ
1−σ 1−σ
≥ C0 G1 (t, s) G2 (s, τ )u(τ ) dτ ds
σ σ
1−σ 1−σ
≥ C0 ν2 G1 (t, s) J2 (τ )u(τ ) dτ ds =: (Lu)(t), t ∈ [0, 1],
σ σ
By (H5), we also deduce that v1 > 0. This completes the proof of Theorem 1.3.5.
34 Boundary Value Problems for Systems of Differential, Difference and Fractional Equations
Theorem 1.3.6. Assume that (H1)–(H5) hold. If the functions f and g also satisfy
conditions (H6) and (H9) and the following condition (H10), then problem (S )–(BC)
has at least two positive solutions (u1 (t), v1 (t)), (u2 (t), v2 (t)), t ∈ [0, 1]:
(H10) For each t ∈ [0, 1], f (t, u) and g(t, u) are nondecreasing with respect to u, and there
exists a constant N > 0 such that
1
N
f t, m0 g(s, N) ds < , ∀ t ∈ [0, 1],
0 m 0
where m0 = max{K̃1 , K̃2 }, K̃1 = maxs∈[0,1] J1 (s), K̃2 = maxs∈[0,1] J2 (s), and J1 and J2
are defined in Section 1.1.2.
Proof. From Section 1.1.2, we have 0 ≤ G1 (t, s) ≤ J1 (s) ≤ K̃1 and G2 (t, s) ≤
J2 (s) ≤ K̃2 for all (t, s) ∈ [0, 1] × [0, 1]. By using (H10), for any u ∈ ∂BN ∩ P , we
obtain
1 1
(Au)(t) ≤ G1 (t, s)f s, K̃2 g(τ , u(τ )) dτ ds
0 0
1 1
≤ J1 (s)f s, m0 g(τ , N) dτ ds, ∀ t ∈ [0, 1].
0 0
Then
1 1 1
N N K̃1
Au ≤ J1 (s)f s, m0 g(τ , N) dτ ds < J1 (s) ds ≤ ≤ N.
0 0 m0 0 m0
So, Au < u for all u ∈ ∂BN ∩ P .
By Theorem 1.3.1, we conclude that the fixed point index of the operator A is
i(A, BN ∩ P , P ) = 1. (1.36)
On the other hand, from (H6), (H9), and the proofs of Theorems 1.3.4 and 1.3.5,
we know that there exists a sufficiently large R1 > N and a sufficiently small r2 with
0 < r2 < N such that
i(A, BR1 ∩ P , P ) = 0, i(A, Br2 ∩ P , P ) = 0. (1.37)
From relations (1.36) and (1.37) we obtain
with ui (t) ≥ 0, vi (t) ≥ 0 for all t ∈ [0, 1] and ui > 0, vi > 0, i = 1, 2. The proof
of Theorem 1.3.6 is completed.
1.3.3 Examples
In this section, we shall present some examples which illustrate our results.
Example 1.3.1. Let f (t, u) = (at + 1)uθ and g(t, u) = (bt + 1)uζ for all t ∈ [0, 1]
and u ∈ [0, ∞), with θ > 1/2 and ζ > 2, and p = 1/2, q1 = 1/2, q2 = 2, a, b ≥ 0,
and σ = 1/4. Then assumptions (H6) and (H7) are satisfied; we have f̃∞ i = ∞,
K2 (t) = t3 , f (t, x) = a(xα̂ + xβ̂ ), and g(t, x) = b(xγ̂ + xδ̂ ) for all t ∈ [0, 1], x ∈
1
[0, ∞), with a, b > 0, α̂ > 1, β̂ < 1, γ̂ > 2, and δ̂ < 1. Then 0 u(s) dH2 (s) =
1 1 1
2 u 3 + 2u 3 0 u(s) dK1 (s) = 0 u(s) dH1 (s) = 2
7 1 2 4 1
, 3u 2 , 0 su(s) ds, and
1 1 2
0 u(s) dK2 (s) =3 0 s u(s) ds.
We consider the second-order differential system
u (t) − 4u(t) + a(v α̂ (t) + v β̂ (t)) = 0, t ∈ (0, 1),
(S )
v (t) − v(t) + b(uγ̂ (t) + uδ̂ (t)) = 0, t ∈ (0, 1),
with the boundary conditions
36 Boundary Value Problems for Systems of Differential, Difference and Fractional Equations
⎧ 1
⎪
⎪ 7 1 2
⎨ u(0) − 3u (0) = 2 su(s) ds, u(1) + u (1) = u + 2u ,
0 2 3 3
⎪
⎪ 4 1 3 1
⎩ 3v(0) − 2v (0) = v , v(1) + v (1) = 3 s2 v(s) ds.
3 2 2 0
(BC )
By using the expressions for J1 and J2 given in Section 1.1.4, we obtain K̃1 =
maxs∈[0,1] J1 (s) ≈ 3.04247954 and K̃2 = maxs∈[0,1] J2 (s) ≈ 1.05002303. Then m0 =
max{K̃1 , K̃2 } = K̃1 . The functions f (t, u) and g(t, u) are nondecreasing with respect
to u for any t ∈ [0, 1], and for p = 1/2 and σ ∈ (0, 1/2) fixed, assumptions (H6) and
(H9) are satisfied; we have f̃∞i = ∞, g̃i = ∞, f̃ i = ∞, and g̃i = ∞.
∞ 0 0
We take N = 1, and then 0 g(s, 1) ds = 2b and f (t, 2bm0 ) = a[(2bm0 )α̂ +
1
β̂+1
(2bm0 )β̂ ]. If a[mα̂+1 α̂ β̂
0 (2b) + m0 (2b) ] < 1, assumption (H10) is satisfied. For
5/2 3/2
example, if α̂ = 3/2, β̂ = 1/2, b = 1/2, and a < 1/(m0 + m0 ) (e.g., a ≤ 0.04),
then the above inequality is satisfied. By Theorem 1.3.6, we deduce that problem
(S )–(BC ) has at least two positive solutions.
This problem is a generalization of the problem studied in Henderson and Luca (2013i),
where in (S ) we have a(t) = 1, c(t) = 1, b(t) = 0, and d(t) = 0 for all t ∈ (0, 1)
(denoted by (S̃ )), and α = α̃ = 1, β = β̃ = 0, γ = γ̃ = 1, δ = δ̃ = 0, H1 and
K1 are constant functions, and H2 and K2 are step functions. Problem (S̃ )–(BC) also
generalizes the problem investigated in Liu et al. (2007), where the authors studied the
existence of positive solutions for system (S̃ ) with the boundary conditions u(0) = 0,
u(1) = αu(η), v(0) = 0, and v(1) = αv(η) with η ∈ (0, 1), 0 < αη < 1.
We shall suppose that assumptions (A1)–(A5) from Section 1.1.2 hold, and we
shall use in our main results from this section Lemmas 1.1.1–1.1.7 from
Section 1.1.2.
The pair of functions (u, v) ∈ (C([0, 1]) ∩ C2 (0, 1))2 is a solution for our problem
(S )–(BC)if and only if (u, v) ∈ (C([0, 1]))2 is a solution for the nonlinear integral
equations
⎧ 1 1
⎪
⎪
⎨ u(t) =
⎪ G1 (t, s)f s, G2 (s, τ )g(τ , u(τ )) dτ ds, t ∈ [0, 1],
0 0
⎪
⎪
⎪
1
⎩ v(t) = G2 (t, s)g(s, u(s)) ds, t ∈ [0, 1].
0
We consider again the Banach space X = C([0, 1]) with the supremum norm ·
and the cone P ⊂ X by P = {u ∈ X, u(t) ≥ 0, ∀ t ∈ [0, 1]}.
We also define the operator D: P → X by
1 1
D(u)(t) = G1 (t, s)f s, G2 (s, τ )g(τ , u(τ )) dτ ds.
0 0