Math. Nachr. 284, No. 11–12, 1472 – 1482 (2011) / DOI 10.1002/mana.

200710145

On the existence of positive periodic solutions of systems of second

order differential equations
Xiong Li∗1 and Ziheng Zhang∗∗2
1
School of Mathematical Sciences, Beijing Normal University, Laboratory of Mathematics and Complex
Systems, Ministry of Education, Beijing 100875, P. R. China
2
Department of Mathematics, Tianjin Polytechnic University, Tianjin 300160, P. R. China

Received 15 August 2007, revised 20 May 2010, accepted 5 June 2010

Published online 31 May 2011

Key words Positive periodic solution, Green’s functions, fixed point theorem
MSC (2010) 34B18, 34B27, 47H10
In this paper, we establish several criteria for the existence, multiplicity, nonexistence of positive periodic solu-
tions of the following system

ẍ + A(t)x = λf (t, x),

by combining some new properties of Green’s function together with Krasnosel’skiı̆ fixed point theorem on the
compression and expression of cones.


c 2011 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

1 Introduction
The existence of periodic solutions of the second order equation for the periodic boundary problem

ẍ = f (t, x),
(1.1)
x(0) = x(1), ẋ(0) = ẋ(1),

and its generalization have attracted much attention. Two of the most common techniques to approach this prob-
lem are: (1) the derivation of a priori bounds for the possible solutions and then the application of topological
degree arguments [9] and (2) the theory of upper and lower solutions [3]. These techniques can be interconnected
and have been proved to be very strong and fruitful, and became very popular in this research area.
On the other hand, some fixed point theorems in cones for completely continuous operators have been exten-
sively employed in studying the existence of positive solutions, especially in connection the separated boundary
value problems after the early works [4], [5]. However, for the periodic problem, it is difficult to find many ref-
erences, and only very recent papers [8], [10]–[12] are known to us. The reason for this contrast may be due to
the fact that it is more difficult to analyze the sign of the Green’s function for the corresponding linear periodic
problems. In the paper [12], Torres succeeded in overcoming this difficulty by using a new Lp -anti-maximum
principle developed in [13] and he obtained some new existence results to problem (1.1).
In [6], the authors studied the perturbation of the Hill equation

ẍ + a(t)x = f (t, x), (1.2)

where f (t, x) has a repulsive singularity near x = 0 and f (t, x) is superlinear near x = +∞, and a(t) ∈ Λ+ (see
Lemma 2.1 for this notation). Under some additional assumptions about growth condition on the nonlinearity f ,
the authors obtained two different positive solutions to Equation (1.2) for the periodic boundary value condition.

∗ Corresponding author: e-mail: xli@bnu.edu.cn, Phone: +(86)(10)58807735, Fax: +(86)(10)58808202

∗∗ e-mail: zhangziheng@tjpu.edu.cn, Phone: +(86)15900357728


c 2011 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
Math. Nachr. 284, No. 11–12 (2011) / www.mn-journal.com 1473

In [8], the following second order singular differential equations


ẍ + a1 (t)x = f1 (t, x, y),
(1.3)
ẍ + a2 (t)x = f2 (t, x, y)

are studied, where the nonlinear terms f1 (t, x, y), f2 (t, x, y) have singularity near (0, 0). Suppose that
ai (t) ∈ Λ+ ∪ Λ− with i = 1, 2 (see also Lemma 2.1 for these notations) and some conditions on the growth of
f1 and f2 , then Equation (1.3) has at least two different positive periodic solutions.
More recently, the following second order differential equation [10] are considered

ẍ + m2 x = λG(t)F (x),
(1.4)
x(0) = x(2π), ẋ(0) = ẋ(2π),
 
where m ∈ 0, 12 is a constant, x = (x1 , . . . , xn )T , G(t) = diag[g1 (t), . . . , gn (t)], F (x) = (f1 (x), . . . , fn (x))T
and λ > 0 is a positive parameter. The authors discussed the existence, multiplicity, nonexistence of positive
periodic solutions of problem (1.4) by Krasnosel’skiı̆ fixed point theorem on the compression and expression of
cones.
Such method also has been employed to prove analogous results for the existence, multiplicity, nonexistence of
positive periodic solutions of boundary value condition of delay differential equations and difference equations,
see [1], [2], [14], [15] and the references therein.
Motivated by the previous works, in this paper we consider the existence, multiplicity, nonexistence of positive
solutions to the periodic boundary problem

ẍ + A(t)x = λf (t, x),
(1.5)
x(0) = x(1), ẋ(0) = ẋ(1),

where x = (x1 , . . . , xn )T , A(t) = diag[a1 (t), . . . , an (t)], f (t, x) = [f1 (t, x), . . . , fn (t, x)]T and λ > 0 is a
positive parameter. n n
Let R = (−∞, ∞), R+ = [0, ∞), Rn+ = i=1 R+ , and for any u = (u1 , . . . , un ) ∈ Rn+ , u = i=1 |ui |.
Throughout this paper, we always assume
(H1 ): ai (t) ∈ Λ+ , i = 1, . . . , n;
(H2 ): fi (t, x) : [0, 1] × Rn+ → [0, ∞) is continuous, fi (t, x) > 0 if x > 0, i = 1, . . . , n.
In order to state our results we introduce the following notations
fi (t, u) fi (t, u)
fi0 = lim , fi∞ = lim , u ∈ Rn+ , i = 1, . . . , n, uniformly in t;
u →0 u u →∞ u
and denote by
 
F0 = max fi0 ,
i=1,...,n
F∞ = max {fi∞ } ,
i=1,...,n
i0 = the number of zeros in the set {F0 , F∞ } ,
i∞ = the number of infinities in the set {F0 , F∞ } .
Now we are in a position to state our main results.
Theorem 1.1 Assume that (H1 ) and (H2 ) hold.
(a) If F0 = 0 and F∞ = ∞, then for all λ > 0 problem (1.5) has one positive solution.
(b) If F0 = ∞ and F∞ = 0, then for all λ > 0 problem (1.5) has one positive solution.
Theorem 1.2 Assume that (H1 ) and (H2 ) hold.
(a) If i0 = 1 or 2, then there exists λ0 > 0 such that for all λ > λ0 problem (1.5) has i0 positive solutions.
(b) If i∞ = 1 or 2, then there exists λ0 > 0 such that for all 0 < λ < λ0 problem (1.5) has i∞ positive
solutions.
www.mn-journal.com 
c 2011 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
1474 X. Li and Z. Zhang: Positive periodic solutions

(c) If i∞ = 0, then there exists λ0 > 0 such that for all 0 < λ < λ0 problem (1.5) has no positive solution.
(d) If i0 = 0, then there exists λ0 > 0 such that for all λ > λ0 problem (1.5) has no positive solution.
Theorem 1.3 Assume that (H1 ) and (H2 ) hold, i0 = i∞ = 0. If
1 1
<λ< ,
σP max {F0 , F∞ } N min {F0 , F∞ }
then problem (1.5) has one positive periodic solution, where σ, P , N are defined in Section 2.
Let us remark that systems (1.3) and (1.4) are special cases of system (1.5), so our results generalize the
corresponding results in [8] and [10].
The remaining part of this paper is organized as follows. In Section 2, we transform problem (1.5) into a fixed
point problem and establish some lemmas. In Section 3, we apply the fixed point theorem to give the proof of
Theorems 1.1–1.3.

2 Preliminaries
We denote by G(t, s) the Green’s function associated with the following problem

ẍ + a(t)x = h(t),
(2.1)
x(0) = x(1), ẋ(0) = ẋ(1).
In [6], the authors assumed that
(A): the Green function G(t, x) is positive for all (t, s) ∈ [0, 1] × [0, 1].
In other words, the (strict) anti-maximum principle holds for (2.1). In this case, the solution of (2.1) is given
by
1
x(t) = (Lh)(t) := G(t, s)h(s) ds. (2.2)
0

In order to guarantee the positivity of G(t, s), it was proved recently in [12] that if a(t) satisfies a  0 then the
positivity of G(t, s) is equivalent to

λ1 (a) > 0, (2.3)

where the notation a  0 means that a(t) ≥ 0 for all t ∈ [0, 1] and a(t) > 0 for t in a subset of positive measure.
The notation λ1 (a) denotes the first anti-periodic eigenvalue of

ẍ + (λ + a(t))x = 0 (2.4)

subject to the anti-periodic boundary condition

x(0) = −x(1), ẋ(0) = −ẋ(1). (2.5)

Some classes of potentials a(t) for (A) to hold have been found recently in [13]. To describe these, we use
.q to denote the usual Lq -norm over (0, 1) for any given q ∈ [1, +∞]. The conjugate exponent of q is denoted
by q ∗ : 1q + q1∗ = 1. Let K(q) denote the best Sobolev constant in the following inequality:

Cu2q ≤ u̇22 for all u ∈ H01 (0, 1).

The explicit formula for K(q) is

⎧
⎪  1−2/q  Γ 1 2
⎨1 2 (q )
, if 1 ≤ q < ∞,
K(q) = q 2 + q Γ ( 12 + q1 ) (2.6)
⎪
⎩4, if q = ∞,

where Γ is the Gamma function.


c 2011 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.mn-journal.com
Math. Nachr. 284, No. 11–12 (2011) / www.mn-journal.com 1475

Let us define the sets of functions

 
Λ− := a ∈ L1 (0, 1) : a(t) ≺ 0 ,
 
Λ+ := a ∈ L1 (0, 1) : a(t)  0, aq ≤ K(2q ∗ ) for some 1 ≤ q ≤ +∞ .

Lemma 2.1 ([12]) Assume that a(t) ∈ Λ− , then G(t, s) < 0 for all (t, s) ∈ [0, 1] × [0, 1]; Assume that
a(t) ∈ Λ+ , then G(t, s) > 0 for all (t, s) ∈ [0, 1] × [0, 1].
Remark 2.2 If q = 1, the condition aq ≤ K(2q ∗ ) can be rewritten as a1 ≤ K(∞) = 4 by the celebrated
stability criterion of Lyapunov. In the case q = ∞, the condition aq ≤ K(2q ∗ ) reads as a∞ ≤ K(2) = π 2 ,
which is a well-known criterion for the anti-maximum principle yet used in the related literature. In this case,
aq ≤ K(2q ∗ ) can be rewritten as a(t) ≺ π 2 .
The proof of our main results will be a consequence of the application of the the following well-known fixed
point theorem of compression and expression of cones [7].
Lemma 2.3 ([7], p. 148) Let X be a Banach space and P ⊂ X be a cone. Assume that Ω1 , Ω2 are open
subsets of X with 0 ∈ Ω1 , Ω̄1 ⊂ Ω2 , and let

T : P ∩ (Ω̄2 \Ω1 ) −→ P

be a completely continuous operator such that either

(i) T u ≥ u, u ∈ P ∩ ∂Ω1 and T u ≤ u, u ∈ P ∩ ∂Ω2 ; or
(ii) T u ≤ u, u ∈ P ∩ ∂Ω1 and T u ≥ u, u ∈ P ∩ ∂Ω2 .
Then T has a fixed point in P ∩ (Ω̄2 \Ω1 ).
In this paper we choose X as the Banach space C[0, 1]n and for u = (u1 , . . . , un ) ∈ X,


n
u = sup |ui (t)|.
i=1 t∈[0,1]

For u ∈ X or Rn+ , u denotes the norm of u in X or Rn+ , respectively.

Under the hypothesis (H1 ), we denote by Gi (t, s) the Green’s function of the following equation

ẍ + ai (t)x = hi (t),
x(0) = x(1), ẋ(0) = ẋ(1),

and
Ai
Ai = min Gi (t, s) > 0, Bi = max Gi (t, s) > 0, 1 > σi = > 0.
0≤t,s≤1 0≤t,s≤1 Bi

Then we can define K be the cone as follows

 
K= u = (u1 , . . . , un ) ∈ X : ui (t) ≥ 0, t ∈ [0, 1], and min ui (t) ≥ σi sup |ui (t)|, i = 1, . . . , n .
t∈[0,1] t∈[0,1]

 Ωr = {u ∈ K : u < r} and ∂Ωr = {u ∈ K : u = r}. Let Tλ : K → X be a map with the
For r > 0, set
components Tλ1 , . . . , Tλn , which are defined by
1
Tλi u(t) = λGi (t, s)fi (s, u(s)) ds, 0 ≤ t ≤ 1.
0

Lemma 2.4 Assume that (H1 ) and (H2 ) hold. Then Tλ : K → K is compact and continuous.

www.mn-journal.com 
c 2011 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
1476 X. Li and Z. Zhang: Positive periodic solutions

P r o o f. Let u ∈ K, then, for i = 1, . . . , n,

1 1
i
Tλ u(t) = λ Gi (t, s)fi (s, u(s)) ds ≥ λAi fi (s, u(s)) ds
0 0
1 1
Ai
≥ λ max Gi (t, s)fi (t, u(s)) ds = λσi max Gi (t, s)fi (t, u(s)) ds
Bi 0 0≤t,s≤1 0 0≤t,s≤1

≥ σi sup Tλi u(t).

0≤t≤1

Thus Tλi (K) → K. Since fi is continuous and Gi is bounded and continuous, it is easy to verify that Tλi is
compact and continuous. 

We remark that u ∈ K is a fixed point of Tλ if and only if u is a positive solution of problem (1.5). Let

A = min {Ai } , σ = min {σi } , B = max {Bi } , P = σA, N = nB.

1≤i≤n 1≤i≤n 1≤i≤n

Then we have
Lemma 2.5 Assume that (H1 ) and (H2 ) hold. Let u = (u1 (t), . . . , un (t)) ∈ K and η > 0, if there exists a
component fi of f such that

n
fi (t, u(t)) ≥ η ui (t), t ∈ [0, 1],
i=1

then Tλ u ≥ λP ηu.

P r o o f. From the definition of Tλ u, it follows that
1 1 
n
Tλ u ≥ max Tλi u(t) ≥ λAi fi (s, u(s)) ds ≥ λAi η ui (t) ds
0≤t≤1 0 0 i=1
1 
n
≥ ληAi σi sup |ui (t)| ds ≥ ληAσu = ληP u.
0 i=1 0≤t≤1 

Lemma 2.6 Assume that (H1 ) and (H2 ) hold. For any r > 0, u = (u1 (t), . . . , un (t)) ∈ ∂Ωr , if there exists
 > 0 such that such that

n
fi (t, u(t)) ≤  ui (t), t ∈ [0, 1], i = 1, . . . , n,
i=1

then Tλ u ≤ λN u.

P r o o f. From the definition of Tλ u, for u ∈ ∂Ωr , we have
n 1 n 1 
n
Tλ u ≤ λ Bi fi (s, u(s)) ds ≤ λ Bi ε ui (s) ds
i=1 0 i=1 0 i=1
n
≤ λε Bi u ≤ λεN u.
i=1 

It is easy to see that the following lemmas can be proved in the similar manners as in Lemmas 2.5 and 2.6, so
we omit the details.
Lemma2.7 Assume that (H1 ) and (H2 ) hold. If u ∈ ∂Ωr , r > 0, then Tλ u ≥ λ Pσ mr , where
mr = min fi (t, u) : u ∈ Rn+ , σr ≤ u ≤ r, t ∈ [0, 1], i = 1, . . . , n > 0.
Lemma 2.8 Assume that (H1 ) and (H2 ) hold. If u ∈ ∂Ωr , r > 0, then Tλ u ≥ λN Mr , where
Mr = max fi (t, u) : u ∈ Rn+ , σr ≤ u ≤ r, t ∈ [0, 1], i = 1, . . . , n > 0.


c 2011 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.mn-journal.com
Math. Nachr. 284, No. 11–12 (2011) / www.mn-journal.com 1477

3 The proofs of main results

3.1 The proof of Theorem 1.1
The proof of Part (a) F0 = 0 implies that fi0 = 0, i = 1, . . . , n. Therefore we can choose r1 > 0 such that

fi (t, u) ≤ εu, u ∈ Rn+ , u ≤ r1 , i = 1, . . . , n, t ∈ [0, 1],

where the constant ε > 0 satisfies

λεN < 1.

Therefore by Lemma 2.6 we have

Tλ u ≤ λN u < u, u ∈ ∂Ωr 1 . (3.1)

On the other hand, since F∞ = ∞, there is a component fi of f such that fi∞ = ∞. Therefore, there exists a
constant H > r1 > 0 such that

fi (t, u) ≥ ηu, u = (u1 , . . . , un ) ∈ Rn+ , u ≥ H, t ∈ [0, 1],

where η > 0 is chosen such that

λP η > 1.
 
Set r2 = max 2r1 , Hσ . If u = (u1 , . . . , un ) ∈ ∂Ωr 2 , then


n
min ui (t) ≥ σu = σr2 ≥ H,
0≤t≤1
i=1

which implies that

fi (t, u(t)) ≥ ηui (t), t ∈ [0, 1].

It follows from Lemma 2.5 that

Tλ u ≥ λP ηu > u, u ∈ ∂Ωr 2 . (3.2)

Now according to (3.1) and (3.2), by Lemma 2.3 we get the fixed point of Tλ on Ωr 2 \Ωr 1 .
The proof of Part (b) If F0 = ∞, there is a component fi of f such that fi0 = ∞. Therefore, there exists
r1 > 0 such that

fi (t, u) ≥ ηu, u = (u1 , . . . , un ) ∈ Rn+ , u ≤ r1 , t ∈ [0, 1],

where η > 0 is chosen such that

λP η > 1.

If u = (u1 , . . . , un ) ∈ ∂Ωr 1 , then


n
fi (t, u(t)) ≥ η ui (t), t ∈ [0, 1],
i=1

which implies, according to Lemma 2.5, that

Tλ u ≥ λP ηu > u, u ∈ ∂Ωr 1 . (3.3)

www.mn-journal.com 
c 2011 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
1478 X. Li and Z. Zhang: Positive periodic solutions

Next we determine Ωr 2 . Since F∞ = 0, we have fi∞ = 0, i = 1, . . . , n. Therefore, there exists H > r1 > 0
such that

fi (t, u) ≤ εu, u = (u1 , . . . , un ) ∈ Rn+ , u ≥ H, t ∈ [0, 1],

where ε satisfies

λεN < 1.
 
Set r2 = max 2r1 , Hσ . If u = (u1 , . . . , un ) ∈ ∂Ωr 2 , then


n
min ui (t) ≥ σu = σr2 ≥ H, u ∈ ∂Ωr 2 ,
0≤t≤1
i=1

which implies that


n
fi (t, u(t)) ≤ ε ui (t), t ∈ [0, 1].
1

Thus, by Lemma 2.6, the following inequality

Tλ u ≤ λN u < u, u ∈ ∂Ωr 2 (3.4)

holds. Now, combining (3.3) and (3.4), by Lemma 2.3 Tλ has a fixed point on Ωr 2 \Ωr 1 . 
This method can be applied to the case that f (t, x) has singularity near 0. We assume

(H2 ): fi (t, x) : [0, 1] × Rn+ \ {0} → (0, ∞) is continuous, and has a singularity near x = 0, i = 1, . . . , n.
Then, we can obtain the analogous result.
 
Corollary 3.1 Assume that (H1 ) and H2 hold. If F∞ = 0, then problem (1.5) has one positive solution.
P r o o f. In fact, since fi (t, u), i = 1, . . . , n have singularity near u = 0, then

fi (t, u)
lim = ∞, i = 1, . . . , n,
u →0 u
that is F0 = ∞. By part (b) of Theorem 1.1 we have finished the proof. 

3.2 The proof of Theorem 1.2

The proof of Part (a) Let r1 = 1, according to Lemma 2.7, we have

σ
Tλ u > u, u ∈ ∂Ωr 1 , λ > λ0 := .
m1 P
If F0 = 0, then fi0 = 0, i = 1, . . . , n. Therefore, there exists r2 < r1 such that

fi (t, u) ≤ εu, u = (u1 , . . . , un ) ∈ Rn+ , u ≤ r2 , t ∈ [0, 1],

where the constant ε satisfies

λεN < 1.

If u = (u1 , . . . , un ) ∈ ∂Ωr 2 , then


n
fi (t, u(t)) ≤ ε ui (t), t ∈ [0, 1],
1


c 2011 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.mn-journal.com
Math. Nachr. 284, No. 11–12 (2011) / www.mn-journal.com 1479

which implies, according to Lemma 2.6, that

Tλ u ≤ λN u < u, u ∈ ∂Ωr 2 . (3.5)

If F∞ = 0, then fi∞ = 0, i = 1, . . . , n. Therefore, there exists H > r1 > 0 such that

fi (t, u) ≤ εu, u = (u1 , . . . , un ) ∈ Rn+ , u ≥ H, t ∈ [0, 1],

where ε is chosen such that

λN ε < 1.
 
Set r3 = max 2r1 , Hσ . If u = (u1 , . . . , un ) ∈ ∂Ωr 3 , then


n
min ui (t) ≥ σu ≥ H, u ∈ ∂Ωr 3 ,
0≤t≤1
i=1

which implies that


n
fi (t, u(t)) ≤ ε ui (t), t ∈ [0, 1].
i=1

It follows from Lemma 2.6 that

Tλ u ≤ λN εu > u, u ∈ ∂Ωr 3 . (3.6)

Thus, in view of Lemma 2.3, Tλ has a fixed point on Ωr 1 \Ωr 2 or Ωr 3 \Ωr 1 according to F0 = 0 or F∞ = 0,
respectively. Consequently, problem (1.5) has one positive solution for λ > λ0 .
If F0 = F∞ = 0, it is easy to see from the above proof that Tλ has a fixed point u1 (t) ∈ Ωr 1 \Ωr 2 and a fixed
point u2 (t) ∈ Ωr 3 \Ωr 1 , such that

r2 < u1  < r1 < u2  < r3 .

Therefore problem (1.5) has two different positive solutions.

The proof of Part (b) Let r1 = 1, according to Lemma 2.8, we have
1
Tλ u < u, u ∈ ∂Ωr 1 , 0 < λ < λ0 =: . (3.7)
N M1

If F0 = ∞, there is a component fi of f such that fi0 = ∞. Therefore, there exists a positive constant r2 < r1
such that

fi (t, u) ≥ ηu, u = (u1 , . . . , un ) ∈ Rn+ , u ≤ r2 , t ∈ [0, 1],

where η > 0 is chosen such that

λP η > 1.

If u = (u1 , . . . , un ) ∈ ∂Ωr 2 , then


n
fi (t, u(t)) ≥ η ui (t), t ∈ [0, 1],
1

which implies, according to Lemma 2.5, that

Tλ u ≥ λP ηu > u, u ∈ ∂Ωr 2 . (3.8)

www.mn-journal.com 
c 2011 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
1480 X. Li and Z. Zhang: Positive periodic solutions

If F∞ = ∞, there is a component fi of f such that fi∞ = ∞. Therefore, there exists H > r1 > 0 such that

fi (t, u) ≥ ηu, u = (u1 , . . . , un ) ∈ R+

n
, u ≥ H, t ∈ [0, 1],

where η satisfies

λP η > 1.
 
Set r3 = max 2r1 , Hσ . Then, if u = (u1 , . . . , un ) ∈ ∂Ωr 3 ,


n
min ui (t) ≥ σu ≥ H, u ∈ ∂Ωr 3 ,
0≤t≤1
i=1

which implies that


n
fi (t, u(t)) ≥ η ui (t), t ∈ [0, 1].
1

It follows, from Lemma 2.5, that

Tλ u ≥ λP ηu > u, u ∈ ∂Ωr 3 . (3.9)

Thus, in view of Lemma 2.3, Tλ has a fixed point on Ωr 1 \Ωr 2 or Ωr 3 \Ωr 1 for 0 < λ < λ0 , according to (3.7)
and (3.8), or (3.7) and (3.9) respectively.
If F0 = F∞ = +∞, it is easy to see from the above proof that Tλ has a fixed point u1 (t) ∈ Ωr 1 \Ωr 2 and a
fixed point u2 (t) ∈ Ωr 3 \Ωr 1 , such that

r2 < u1  < r1 < u2  < r3 .

Therefore problem (1.5) has two different positive solutions.

The proof of Part (c) Since F0 < ∞ and F∞ < ∞ then fi0 < ∞ and fi∞ < ∞, i = 1, . . . , n. It is easy to
see that there exists an ε such that

fi (t, u) ≤ εu, u ∈ Rn+ , t ∈ [0, 1], i = 1, . . . , n.

Assume that v(t)=(v1 (t), . . . , vn (t)) is one positive solution of problem (1.5). We will show this leads to a
contradiction for 0 < λ < λ0 := N1ε . In fact, for 0 < λ < λ0 , since Tλ v(t) = v(t), we find

n
v = Tλ v = max Tλi v(t)
0≤t≤1
i=1

n 
n
≤ λBi εv = λε Bi v
i=1 i=1
< v,
which is a contradiction.
The proof of Part (d) Since F0 > 0 and F∞ > 0, there exist two components fi and fj of f and positive
numbers c1 , c2 , r1 and r2 such that r1 < r2 and
fi (t, u) ≥ c1 u, u ∈ Rn+ , u ≤ r1 ,
fj (t, u) ≥ c2 u, u∈ Rn+ , u ≥ r2 .
Let
  
fj (t, u)
η = min c1 , c2 , min : u ∈ Rn+ , r1 σ ≤ u ≤ r2 > 0.
u

c 2011 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.mn-journal.com
Math. Nachr. 284, No. 11–12 (2011) / www.mn-journal.com 1481

Thus, we have

fi (t, u) ≥ ηu, u ∈ Rn+ , u ≤ r1 , (3.10)

and

fj (t, u) ≥ ηu, u ∈ Rn+ , u ≥ σr1 . (3.11)

Assume that v(t) = (v1 (t), . . . , vn (t)) is one positive solution of problem (1.5), we will show this leads to a
contradiction for λ > λ0 := P1η . In fact, if v ≤ r1 , according to (3.10), we have


n
fi (t, v(t)) ≥ η vi (t), t ∈ [0, 1].
i=1

On the other hand, if v > r1 , then


n
min vi (t) ≥ σv > σr1 ,
0≤t≤1
i=1

which, together with (3.11), implies that


n
fi (t, v(t)) ≥ η vi (t), t ∈ [0, 1].
i=1

Since Tλ v(t) = v(t) for t ∈ [0, 1], it follows from Lemma 2.5, for λ > λ0 , that

v = Tλ v ≥ λP ηv > v,

which is a contradiction. 
Similar to Corollary 3.1, we have

Corollary 3.2 Assume that (H1 ) and (H2 ) hold.
(a) If 0 < F∞ < ∞, then there exists λ0 > 0 such that for 0 < λ < λ0 problem (1.5) has one positive
solution.
(b) If F∞ = ∞, then there exists λ0 > 0 such that for 0 < λ < λ0 problem (1.5) has two positive solutions.

3.3 The proof of Theorem 1.3

Without loss of generality, we may assume that F∞ ≥ F0 . The other case F∞ < F0 can be treated similarly. If
λ satisfies
1 1
<λ< ,
σP F∞ N F0
then there exists an 0 < ε < F∞ such that
1 1
<λ< .
σP (F∞ − ε) N (F0 + ε)

Now, according to the definition of F0 , there exists r1 > 0 such that

fi (t, u) ≤ (F0 + ε)u, u ∈ Rn+ , u ≤ r1 , i = 1, . . . , n, t ∈ [0, 1].

Thus,

fi (t, u(t)) ≤ (F0 + ε)u, u ∈ ∂Ωr 1 , i = 1, . . . , n, t ∈ [0, 1].

www.mn-journal.com 
c 2011 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
1482 X. Li and Z. Zhang: Positive periodic solutions

Therefore we have, by Lemma 2.6, that

Tλ u ≤ λN (F0 + )u < u, u ∈ ∂Ωr 1 . (3.12)
On the other hand, there is some i such that fi∞ = F∞ > 0. Thus, there exists H > r1 > 0 such that
fi (t, u) ≥ (F∞ − ε)u, u = (u1 , . . . , un ) ∈ R+
n
, u ≥ H, t ∈ [0, 1].
 
Set r2 = 2r1 , Hσ . Then it follows that

n
min uj (t) ≥ σu = σr2 ≥ H, u = (u1 , . . . , un ) ∈ ∂Ωr 2 ,
0≤t≤1
j =1

which implies that


n
fi (t, u(t)) ≥ (F∞ − ε) uj (t), u ∈ ∂Ωr 2 , t ∈ [0, 1].
j =1

In view of Lemma 2.5, we have

Tλ u ≥ λσP (F∞ − ε)u > u, u ∈ ∂Ωr 2 . (3.13)
Hence, according to (3.12) and (3.13), by Lemma 2.3 Tλ has a fixed point on Ωr 2 \Ωr 1 . Consequently, problem
(1.5) has one positive solution. 

Acknowledgements The first author is partially supported by NSFC (11031002), the Fundamental Research Funds for the
Central Universities and SRF for ROCS, SEM.

References
[1] F. M. Atici, A. Cabada, and V. Otero-Espinar, Criteria for existence and nonexistence of positive solutions to a discrete
periodic boundary value problem, J. Difference Equ. Appl. 9, 765–775 (2003).
[2] S. Cheng and G. Zhang, Existence of positive periodic solutions for non-autonomous functional differenrial equations,
Electron. J. Differ. Equations 59, 1–8 (2001).
[3] C. De Coster and P. Habets, Upper and Lower Solutions in the Theory of ODE Boundary Value Problems: Classical and
Recent Results, in: Nonlinear Analysis and Boundary Value Problems for Ordinary Differential Equations, edited by
F. Zanolin, CISM-ICMS Courses and Lectures Vol. 371 (Springer, New York, 1996), pp. 1–78.
[4] L. H. Erbe and R. M. Mathsen, Positive solutions for singular nonlinear boundary value problems, Nonlinear Anal. 46,
979–986 (2001).
[5] L. H. Erbe and H. Wang, On the existence of positive solutions of ordinary differential equations, Proc. Am. Math. Soc.
120, 743–748 (1994).
[6] D. Jiang, J. Chu, and M. Zhang, Multiplicity of positive periodic solutions to superlinear repulsive singular equations,
J. Differ. Equations 211, 282–302 (2005).
[7] M. A. Krasnosel’ skiı̆, Positive Solutions of Operator Equations (Noordhoff, Groningen, 1964).
[8] X. Lin, D. Jiang, D. O’Regan, and R. Agarwal, Twin positive periodic solutions of second order singular differential
systems, Topol. Methods Nonlinear Anal. 25, 263–273 (2005).
[9] J. Mawhin, Topological Degree and Boundary Value Problems for Nonlinear Differential Equations, in: Topological
Methods for Ordinary Differential Equations, edited by M. Furi and P. Zecca, Lecture Notes in Mathematics Vol. 1537
(Springer, New York/Berlin, 1993), pp. 74–142.
[10] D. O’Regan and H. Wang, Positive periodic solutions of systems of second order ordinary didderential equations, Posi-
tivity 10, 285–298 (2006).
[11] I. Rachunková, M. Tvrdý, and I. Vrkoc̆, Existence of nonnegative and nonpositive solutions for second order periodic
boundary value problems, J. Differ. Equations 176, 445–469 (2001).
[12] P. J. Torres, Existence of one-signed periodic solutions of some second-order differential equations via a Krasnoselskii
fixed point theorem, J. Differ. Equations 190, 643–662 (2003).
[13] P. J. Torres and M. Zhang, A monotone iterative scheme for a nonlinear second order equation based on a generalized
anti-maximum principle, Math. Nachr. 251, 101–107 (2003).
[14] A. Wan and D. Jiang, Existence of positive periodic solutions for functional differential equations, Kyushu J. Math. 56,
193–202 (2002).
[15] H. Wang, Positive periodic solutions of functional differential equations, J. Differ. Equations 202, 354–366 (2004).


c 2011 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.mn-journal.com