Volume 2013, Issue 1 504573
Research Article
Open Access

Positive Periodic Solution of Second-Order Coupled Systems with Singularities

Tiantian Ma

Corresponding Author

Tiantian Ma

School of Mathematical Sciences, Capital Normal University, Beijing 100048, China cnu.edu.cn

Search for more papers by this author
First published: 16 July 2013
Academic Editor: Chuanzhi Bai

Abstract

This paper establishes the existence of periodic solution for a kind of second-order singular nonautonomous coupled systems. Our approach is based on fixed point theorem in cones. Examples are given to illustrate the main result.

1. Introduction

We are concerned with the existence of positive T-periodic solution for the second-order nonautonomous singular coupled systems:
()
where fiC((0, T)×[0, +), [0, +)), ci : (0, T)→[0, +), i = 1,2, are Lebesgue integrable, fi may be singular at t = 0, T and ci can have finitely many singularities.

Singular differential equations or systems arise from many branches of applied mathematics and physics such as gas dynamics, Newtonian fluid mechanics, and nuclear physics, which have been widely studied by many authors (see [17] and references therein). Some classical-tools have been used to study the positive solutions for two point nonperiodic boundary value problems of coupled systems [8, 9]. However, there are few works on periodic solutions of second order nonautonomous singular coupled systems of type (1).

In the recent papers [10, 11], the periodic solutions of singular coupled systems
()
were proved by using some fixed point theorems in cones for completely continuous operators, where a1, a2, c1, c2L1(0, T), f1, f2 ∈ Car([0, T]×(0, +), (0, +)). When the Green′s function Gi(t, s) (i = 1, 2), associated with the periodic boundary problem
()
is nonnegative for all (t, s)∈[0, T]×[0, T] and fi (i = 1,2) satisfies weak singularities
()
where 0 < αi < 1, , , bi ≥ 0, and , bi are strictly positive on some positive measure subsets of (0, T), some sufficient conditions for the existence of periodic solutions of (2) were obtained in [10, 11].
Motivated by the papers [911], we consider the existence of positive T-periodic solution of (1). Owing to the disappearing of the terms ai(t) (i = 1,2) in (2), the methods in [9, 10] are no longer valid. In present paper, we will deal with the periodic solutions of (1) under new conditions. Let k be a constant satisfying 0 < k < π/T. Denote by G(t, s) the Green function of
()
which can be expressed by
()
By a direct computation, we can get
()
Assume fi (i = 1,2) satisfies the following conditions.
  • (H1)

    For t ∈ (0, T), fi(t, 1) > 0 and there exist constants μ1μ2 > 1, λ1λ2 > 1 such that, for any constants 0 ≤ ρi ≤ 1, i = 1,2,

    ()
    ()

  • (H2)

    and

    ()

where , , i = 1,  2.

Remark 1. For any ρ1 ≥ 1, we get from (8) that

()

For any ρ2 ≥ 1, we get from (9) that

()

Typical functions that satisfy (8) or (9) are those taking the form
()
where aikC(0, T), aik(t) > 0, lik > 1, i = 1,2; k = 1,2, …, n.

Definition 2. Supposing that (u, v) ∈ C1[0, T]∩C2(0, T) × C1[0, T]∩C2(0, T) satisfies (1) and u(t) > 0, v(t) > 0 for any t ∈ [0, T], then one says that (u, v) is a C1[0, T] × C1[0, T] positive solution of system (1).

By using fixed point theorem in cones, we are able to prove the following result.

Theorem 3. Assume that (H1), (H2) hold. Then (1) has at least one positive T-periodic solution.

The proof of Theorem 3 will be given in Section 3 of this paper.

2. Preliminaries

Lemma 4 (see [12].)Let X be a Banach space, K a cone in X, Ω1, Ω2 two nonempty bounded open sets in K, . is a completely continuous operator. If

  • (i)

    T(x) ≠ λx, xΩ1, λ > 1,

  • (ii)

    T(x) ≠ λx, xΩ2, 0 < λ < 1, ,

then T has a fixed point in .

Lemma 5. If fi(t, u) satisfy (H1), then, for t ∈ (0, T), fi(t, u) are increasing on u and, for any [α, β]⊂(0, T),

()
uniformly with respect to t ∈ [α, β].

Proof. We only deal with f1. Without loss of generality, let 0 ≤ xy. If y = 0, we get f1(t, x) ≤ f1(t, y). If y ≠ 0, let c0 = x/y, then 0 ≤ c0 ≤ 1. From (8), we get

()
which means f1(t, u) is increasing on u.

Assume u > 1. It follows from (11) that . Thus

()
From (H1), for any [α, β]⊂(0, T), we obtain
()
Therefore
()
uniformly with respect to t ∈ [α, β].

Let X = C[0, T]. We know X is a Banach space with the norm ∥u∥ = max t∈[0,T] | u(t)|. Define the sets
()
It is easy to check that P, Q are cones in X and QP. Throughout this paper, we consider the space X × X. It is easy to see X × X is a Banach space with the norm,
()

We can get the conclusion that P × P, Q × Q are cones in X × X and Q × QP × P.

For any uX, define the function
()
Then the solution of periodic boundary value problem
()
can be expressed by , and the solution of periodic boundary value problem
()
can be expressed by . Obviously, , i = 1, 2. Then
()
From (11), (12), and Lemma 5, we have
()
()
Then, for any fixed (u, v) ∈ P × P, it follows from (H2) that
()
Thus, we can define the operator T : P × PP × P, T(u, v) = (T1u, T2v) by
()
for (u, v) ∈ P × P. Then, we have the following lemma.

Lemma 6. Assuming that (H1), (H2) hold, then T has a fixed point if and only if

()
has one positive T-periodic solution.

Lemma 7. Assuming that (H1), (H2) hold, then T(Q × Q) ⊂ Q × Q and T : Q × QQ × Q is completely continuous.

Proof. For (u, v) ∈ Q × Q, we have

()
Then, we can get
()
which means T(Q × Q) ⊂ Q × Q.

Let BQ × Q be any bounded set. Then there exists a constant N such that, for any (u, v) ∈ B,

()
From (27), we have
()
Let
()
Thus
()
which implies that T(B) is bounded.

Next we prove that T(B) is equicontinuous. For any (u, v) ∈ B, t ∈ [0, T], we know

()
From (25) and (H2), we have
()

Using the same method, we can obtain . Therefore, T(B) is equicontinuous. According to Ascoli-Arzela theorem, T(B) is a relatively compact set.

Next, we prove that T : Q × QQ × Q is continuous. Suppose (un, vn), (u0, v0) ∈ Q × Q, (un, vn)→(u0, v0), n → +, that is, unu0, vnv0, n → +. We know that there exists a constant L > 0 such that

()

We shall prove T(un, vn) → T(u0, v0), n → +, that is,

()

We first deal with T1unT1u0, n → +. Otherwise, there exist ε0 > 0, {tn}∈[0, T] such that |T1un(tn) − T1u0(tn)| ≥ ε0. Without loss of generality, we can assume tnt0 ∈ [0, T]. We know

()

Next, we show T1un(t0) − T1u0(t0) → 0, n. In fact,

()
Let
()
Since
()
and f1 is continuous, we know rn(s) → 0, n. From (24), we know
()
It can be inferred from (8) that
()

Set . Thus, we get

()
From (H2), we know . Using Lebesgue-dominated convergence theorem, we get
()

From (40) and (47), we obtain

()
which is a contradiction. Thus, we know T1unT1u0, n → +. Using the same method, we can obtain T2vnT2v0, n → +. Then
()
which means T : Q × QQ × Q is continuous. Therefore T : Q × QQ × Q is a completely continuous operator.

3. Proof of Theorem 3

We proceed to prove Theorem 3 in two steps.

(1) Let Ω1 = {(u, v) ∈ Q × Q : ∥u∥ < 2M2r1/m, ∥v∥ < 2M2r2/m}. We can get
()
For (u, v) ∈ Ω1, we have the following two cases.

Case I. One has {(u, v) ∈ Q × Q : ∥u∥ = 2M2r1/m, ∥v∥ ≤ 2M2r2/m}. Under this condition, we can get

()
Otherwise, there exist (u0, v0) ∈ Ω1, λ0 > 1 such that T1u0 = λ0u0. As
()
then . By a direct computation, we know u0 = (1/λ0)T1u0 satisfies
()
Then, we get
()

Since , integrating both sides of (54) on [0, T], we get

()
Then
()
That is,
()
which contradicts with (H2).

Case II. One has {(u, v) ∈ Q × Q : ∥u∥ ≤ 2M2r1/m, ∥v∥ = 2M2r2/m}. Under this condition, we can get

()
Otherwise, there exist (u0, v0) ∈ Ω1, λ0 > 1 such that T2v0 = λ0v0. As
()
then . By a direct computation, we know v0 = (1/λ0)T2v0 satisfies
()

For , using the same method as condition I, we obtain

()
which is also a contradiction.

(2) Choose an interval [α, β]⊂(0, T) satisfying βα = T/2. Set . From Lemma 5, there exists R1 > 2M2r/m, r = max {r1, r2} such that
()
Let R ≥ (2M/m)R1 > R1 > 2M2r/m. Define Ω2 = {(u, v) ∈ Q × Q : ∥u∥ < R, ∥v∥ < R}. We can get
()

For (u, v) ∈ Ω2, we have the following two cases.

Case I. One has {(u, v) ∈ Q × Q : ∥u∥ = R, ∥v∥ ≤ R}. Under this condition, we know

()
Thus . Furthermore, we obtain from (62) that
()
On the other hand, for 0 < λ < 1, we know λv(t) < v(t) ≤ R. From the choice of , we get T2vλv.

Case II. One has {(u, v) ∈ Q × Q : ∥u∥ ≤ R, ∥v∥ = R}. Under this condition, we get

()
Thus . From (62), we get
()
For 0 < λ < 1, λu(t) < u(t) ≤ R, from the choice of , we know T1uλu.

Furthermore, we can obtain

()
It implies .

From Lemma 4, we know T has a fixed point in . For , we have the following three cases.

Case 1. One has

()

Case 2. One has

()

Case 3. One has

()

Next, we show Cases 1 and 2 are impossible. In Case 1, we have

()
It follows that . By a direct computation, we know satisfies
()
Then, we get
()
Since , integrating both sides of (74) on [0, T], we get
()
Then
()
That is,
()
which contradicts with (H2). Using the same method, we can prove that Case 2 is also impossible. Therefore, Case 3 is satisfied and T has a fixed point in satisfying
()
Since
()
from Lemma 6, we know satisfy
()

Let . For

()
we obtain
()

This means (u*, v*) is one positive T-periodic solution of (1).

4. Applications of Theorem 3

Finally, we give some examples as the applications of Theorem 3:
()
Choosing , , , , , i = 1,2, and μ1 = 2 > μ2 = 5/4 > 1, λ1 = 2 > λ2 = 5/4 > 1, then (H1) is satisfied. Notice (H2) also holds, since
()
Existence of the positive T-periodic solutions is guaranteed from Theorem 3. We can also consider the following examples and the same result can be obtained:
()
()

Acknowledgments

The author is grateful to the referees for valuable comments and useful remarks on the paper. Research supported by China Postdoctoral Science Foundation (2012M510341), Beijing Natural Science Foundation (1112006) and the Grant of Beijing Education Committee Key Project (KZ20130028031).

      The full text of this article hosted at iucr.org is unavailable due to technical difficulties.