Electronic Journal of Differential Equations, Vol. 2006(2006), No. 68, pp. 1–11. ISSN: 1072-6691. URL: http://ejde.math.txstate.edu or http://ejde.math.unt.edu ftp ejde.math.txstate.edu (login: ftp) EXISTENCE OF POSITIVE SOLUTIONS FOR BOUNDARY-VALUE PROBLEMS FOR SINGULAR HIGHER-ORDER FUNCTIONAL DIFFERENTIAL EQUATIONS CHUANZHI BAI, QING YANG, JING GE Abstract. We study the existence of positive solutions for the boundaryvalue problem of the singular higher-order functional differential equation (Ly (n−2) )(t) + h(t)f (t, yt ) = 0, y (i) (0) = 0, for t ∈ [0, 1], 0 ≤ i ≤ n − 3, αy (n−2) (t) − βy (n−1) (t) = η(t), γy (n−2) (t) + δy (n−1) (t) = ξ(t), for t ∈ [−τ, 0], for t ∈ [1, 1 + a], where Ly := −(py 0 )0 + qy, p ∈ C([0, 1], (0, +∞)), and q ∈ C([0, 1], [0, +∞)). Our main tool is the fixed point theorem on a cone. 1. Introduction As pointed out in [5], boundary-value problems associated with functional differential equations arise from problems in physics, from variational problems in control theory, and from applied mathematics; see for example [6, 8]. Many authors have investigated the existence of solutions for boundary-value problems of functional differential equations; see [3, 9, 15, 18]. Recently an increasing interest in studying the existence of positive solutions for such problems has been observed. Among others publication, we refer to [1, 2, 10, 11, 13, 19]. In this paper, we investigate the existence of positive solutions for singular boundary-value problems (BVP) of an n-th order (n ≥ 3) functional differential equation (FDE) of the form (Ly (n−2) )(t) + h(t)f (t, yt ) = 0, (i) y (0) = 0, 0 ≤ i ≤ n − 3, αy (n−2) (t) − βy (n−1) (t) = η(t), γy (n−2) (t) + δy (n−1) for t ∈ [0, 1], (t) = ξ(t), (1.2) for t ∈ [−τ, 0], (1.3) for t ∈ [1, 1 + a], (1.4) 2000 Mathematics Subject Classification. 34K10, 34B16. Key words and phrases. Boundary value problem; higher-order; positive solution; functional differential equation; fixed point. c 2006 Texas State University - San Marcos. Submitted April 21, 2006. Published July 6, 2006. Supported by the Natural Science Foundation of Jiangsu Education Office and by Jiangsu Planned Projects for Postdoctoral Research Funds. 1 (1.1) 2 C. BAI, Q. YANG, J. GE EJDE-2006/68 where Ly := −(py 0 )0 + qy, p ∈ C([0, 1], (0, +∞)), and q ∈ C([0, 1], [0, +∞)); α, β, γ, δ ≥ 0, and αδ + αγ + βγ > 0; η ∈ C([−τ, 0], R), ξ ∈ C([1, b], R) (b = 1 + a), and η(0) = ξ(1) = 0; h ∈ C((0, 1), R) (h(t) is allowed to have singularity at t = 0 or 1); f ∈ C([0, 1] × D, R), D = C([−τ, a], R), for every t ∈ [0, 1], yt ∈ D is defined by yt (θ) = y(t + θ), θ ∈ [−τ, a]. The study of higher-order functional differential equation has received also some attention; see for example [3, 10, 17]. Recently, Hong et al. [12] imposed conditions on f (t, y t ) to yield at least one positive solution to (1.1)-(1.4) for the special case h(t) ≡ 1, p(t) ≡ 1, and q(t) ≡ 0. They applied the Krasnosel’skii fixed-point theorem. The purpose of this paper is to establish the existence of positive solutions of the singular higher-order functional differential equation (1.1) with boundary conditions (1.2)-(1.4) under suitable conditions on f . 2. Preliminaries To abbreviate our discussion, we assume the following hypotheses: (H1) G(t, s) is the Green’s function of the differential equation (Ly (n−2) )(t) = 0, 0<t<1 subject to the boundary condition (1.2)-(1.4) with τ = a = 0. (H2) g(t, s) is the Green’s function of the differential equation Ly(t) = 0 t ∈ (0, 1) subject to the boundary conditions αy(0) − βy 0 (0) = 0, γy(1) + δy 0 (1) = 0, where α, β, γ and δ are as in (1.3) and (1.4). (H3) h ∈ C((0, 1), [0, +∞)) and satisfies Z 1 0< g(s, s)h(s)ds < +∞. 0 + (H4) f ∈ C([0, 1] × D , [0, ∞)), where D+ = C([−τ, a], [0, +∞)). (H5) η ∈ C([−τ, 0], [0, +∞)), ξ ∈ C([1, 1 + a], [0, +∞)), and η(0) = ξ(1) = 0. It is easy to see that ∂ n−2 G(t, s) = g(t, s), t, s ∈ [0, 1]. ∂tn−2 It is also well known that the Green’s function g(t, s) is ( 1 φ(s)ψ(t), if 0 ≤ s ≤ t ≤ 1, g(t, s) = c φ(t)ψ(s), if 0 ≤ t ≤ s ≤ 1, where φ and ψ are solutions, respectively, of Lφ = 0, Lψ = 0, φ(0) = β, ψ(1) = δ, φ0 (0) = α, 0 ψ (1) = −γ. (2.1) (2.2) One can show that c = −p(t)(φ(t)ψ 0 (t) − φ0 (t)ψ(t)) > 0 and φ0 (t) > 0 on (0, 1] and ψ 0 (t) < 0 on [0, 1). Clearly g(t, s) ≤ g(s, s), 0 ≤ t, s ≤ 1. (2.3) EJDE-2006/68 EXISTENCE OF POSITIVE SOLUTIONS 3 By (H3), there exists t0 ∈ (0, 1) such that h(t0 ) > 0. We may choose ε ∈ (0, 1/2) such that t0 ∈ (ε, 1 − ε). Then for ε ≤ t ≤ 1 − ε we have φ(ε) ≤ φ(t) ≤ φ(1 − ε) and ψ(1 − ε) ≤ ψ(t) ≤ ψ(ε). Also for (t, s) ∈ [ε, 1 − ε] × (0, 1) ψ(1 − ε) φ(ε) ψ(1 − ε) φ(ε) g(t, s) ≥ min , ≥ min , := σ. (2.4) g(s, s) ψ(s) φ(s) ψ(0) φ(1) Let E = C (n−2) ([−τ, b]; R) with a norm kuk[−τ,b] = sup−τ ≤t≤b |u(n−2) (t)| for u ∈ E. Obviously, E is a Banach space. And let C = C (n−2) ([−τ, a], R) be a space with norm kψk[−τ,a] = sup−τ ≤t≤a |ψ (n−2) (x)| for ψ ∈ C. Let C + = {ψ ∈ C : ψ(x) ≥ 0, x ∈ [−τ, a]}. It is easy to see that C + is a subspace of C. Define a cone K ⊂ E as follows: K = {y ∈ E : y(t) ≥ 0, min y (n−2) (t) ≥ σkyk[−τ,b] }, (2.5) t∈[ε,1−ε] where σ = 1b min{ε, σ}, σ is as in (2.4). For each ρ > 0, we define Kρ = {y ∈ K : kyk[−τ,b] < ρ}. Furthermore, we define a set Ωρ as follows: Ωρ = y ∈ K : min y (n−2) (t) < σρ . ε≤t≤1−ε Similar to the [14, Lemma 2.5], we have Lemma 2.1. Ωρ defined above has the following properties: (a) Ωρ is open relative to K. (b) Kσρ ⊂ Ωρ ⊂ Kρ . (c) y ∈ ∂Ωρ if and only if minε≤t≤1−ε y (n−2) (t) = σρ. (d) If y ∈ ∂Ωρ , then σρ ≤ y (n−2) (t) ≤ ρ for t ∈ [ε, 1 − ε]. To obtain the positive solutions of (1.1)-(1.4), the following fixed point theorem in cones will be fundamental. Lemma 2.2. Let K be a cone in a Banach space E. Let D be an open bounded subset of E with DK = D ∩ K 6= ∅ and DK 6= K. Assume that A : DK → K is a compact map such that x 6= Ax for x ∈ ∂DK . Then the following results hold. (1) kAxk ≤ kxk, x ∈ ∂DK , then iK (A, DK ) = 1. (2) If there exists e ∈ K\{0} such that x 6= Ax + λe for all x ∈ ∂DK and all λ > 0, then iK (A, DK ) = 0. (3) Let U be an open set in E such that U ⊂ DK . If iK (A, DK ) = 1 and iK (A, UK ) = 0, then A has a fixed point in DK \U K . The same results holds if iK (A, DK ) = 0 and iK (A, UK ) = 1. Suppose that y(t) is a solution of (1.1)-(1.4), then it can be written as ; t), −τ ≤ t ≤ 0, y(−τ R1 y(t) = G(t, s)h(s)f (s, y )ds, 0 ≤ t ≤ 1, s 0 y(b; t), 1 ≤ t ≤ b, where y(−τ ; t) and y(b; t) satisfy ( α R0 α −βs 1 (n−2) βt e e η(s)ds + y (0) , t ∈ [−τ, 0], β 6= 0, β t y (n−2) (−τ ; t) = 1 t ∈ [−τ, 0], β = 0, α η(t), 4 C. BAI, Q. YANG, J. GE EJDE-2006/68 and y (n−2) (b; t) = ( γ Rt γ γ e− δ t 1δ 1 e δ s ξ(s)ds + e δ y (n−2) (1) , t ∈ [1, b], δ 6= 0, 1 γ ξ(t), t ∈ [1, b], δ = 0. Throughout this paper, we assume that u0 (t) is the solution of (1.1)-(1.4) with (n−2) f ≡ 0, and ku0 k[−τ,b] =: M0 . Clearly, u0 (t) can be expressed as follows: (n−2) (−τ ; t), −τ ≤ t ≤ 0, u0 (n−2) u0 (t) = 0, 0 ≤ t ≤ 1, (n−2) u0 (b; t), 1 ≤ t ≤ b. where ( (n−2) u0 (−τ ; t) = R 0 −αs 1 α βt e β η(s)ds, βe t 1 α η(t), t ∈ [−τ, 0], β 6= 0, t ∈ [−τ, 0], β = 0, R 1 − γδ t t δe 1 1 γ ξ(t), t ∈ [1, b], δ 6= 0, t ∈ [1, b], δ = 0. and ( (n−2) u0 (b; t) = γ e δ s ξ(s)ds, Let y(t) be a solution of BVP (1.1)-(1.4) and u(t) = y(t) − u0 (t). Noting that u(t) ≡ y(t) for 0 ≤ t ≤ 1, we have (n−2) (−τ ; t), −τ ≤ t ≤ 0, u R1 (n−2) u (t) = g(t, s)h(s)f (s, u + (u ) )ds, 0 ≤ t ≤ 1, s 0 s 0 (n−2) u (b; t), 1 ≤ t ≤ b, where u (n−2) ( α R1 e β t 0 g(0, s)h(s)f (s, us + (u0 )s )ds, (−τ ; t) = 0, t ∈ [−τ, 0], β 6= 0, t ∈ [−τ, 0], β = 0, and ( γ R1 e− δ (t−1) 0 g(1, s)h(s)f (s, us + (u0 )s )ds, (n−2) u (b; t) = 0, t ∈ [1, b], δ 6= 0, t ∈ [1, b], δ = 0. It is easy to see that y(t) is a solution of BVP (1.1)-(1.4) if and only if u(t) = y(t) − u0 (t) is a solution of the operator equation u(t) = Au(t) for t ∈ [−τ, b]. (2.6) Here, operator A : E → E is defined by u(t), −τ ≤ t ≤ 0, B R 11 Au(t) := G(t, s)h(s)f (s, us + (u0 )s )ds, 0 ≤ t ≤ 1, 0 B2 u(t), 1 ≤ t ≤ b, where B1 u(t) := R1 β n−2 α e β t 0 g(0, s)h(s)f (s, us + (u0 )s )ds, α R1 tn−2 g(0, s)h(s)f (s, us + (u0 )s )ds, (n−2)! 0 β= 6 0, α 6= 0, β= 6 0, α = 0, 0, β=0 EJDE-2006/68 EXISTENCE OF POSITIVE SOLUTIONS 5 for each t ∈ [−τ, 0], and R δ n−2 − γ (t−1) 1 g(1, s)h(s)f (s, us + (u0 )s )ds, δ = 6 0, γ 6= 0, −γ R e δ 0 1 tn−2 B2 u(t) := (n−2)! g(1, s)h(s)f (s, u + (u ) )ds, δ 6 0, γ = 0, = s 0 s 0 0, δ=0 for any t ∈ [1, b]. Obviously, (n−2) (t), −τ ≤ t ≤ 0, 1 u) (B R 1 (n−2) (Au) (t) := g(t, s)h(s)f (s, us + (u0 )s )ds, 0 ≤ t ≤ 1, 0 (B2 u)(n−2) (t), 1 ≤ t ≤ b, where (n−2) (B1 u) ( α R1 e β t 0 g(0, s)h(s)f (s, us + (u0 )s )ds, (t) := 0, t ∈ [−τ, 0], β 6= 0, t ∈ [−τ, 0], β = 0, and (n−2) (B2 u) ( γ R1 e− δ (t−1) 0 g(1, s)h(s)f (s, us + (u0 )s )ds, (t) := 0, t ∈ [1, b], δ 6= 0, t ∈ [1, b], δ = 0. Lemma 2.3. With the above notation, A(K) ⊂ K. Proof. By the assumptions of (H1)-(H5), it is easy to know that Au ∈ E and Au ≥ 0 for any u ∈ K. Moreover, it follows from 0 ≤ (Au)(n−2) (t) ≤ (Au)(n−2) (0) (n−2) 0 ≤ (Au) (n−2) (t) ≤ (Au) (1) for − τ ≤ t ≤ 0 for 1 ≤ t ≤ b that kAuk[−τ,b] = kAuk[0,1] . By (2.3) we have, for any u ∈ K and t ∈ [0, 1] that Z 1 kAuk[−τ,b] = kAuk[0,1] ≤ g(s, s)h(s)f (s, us + (u0 )s )ds. (2.7) 0 From (2.4), we get min (Au)(n−2) (t) = ε≤t≤1−ε Z min ε≤t≤1−ε Z 1 g(t, s)h(s)f (s, us + (u0 )s )ds 0 1 ≥σ g(s, s)h(s)f (s, us + (u0 )s )ds (2.8) 0 Z ≥σ 1 g(s, s)h(s)f (s, us + (u0 )s )ds. 0 In view of (2.7) and (2.8), we obtain min (Au)(n−2) (t) ≥ σkAuk[−τ,b] , ε≤t≤1−ε u ∈ K, which implies A(K) ⊂ K. Let + C[k,r] = {ϕ ∈ C + : k ≤ kϕk[−τ,a] ≤ r}, + C[k,∞) = {ϕ ∈ C + : k ≤ kϕk[−τ,a] < ∞}, where 0 ≤ k < r. 6 C. BAI, Q. YANG, J. GE EJDE-2006/68 Lemma 2.4. A : K → K is completely continuous. Proof. We apply a truncation technique (cf. [16]). We define the function hm for m ≥ 2, by 1 1 0<t≤ m , min h(t), h( m ) , 1 1 hm (t) = h(t), < t < 1 −m , m m−1 m−1 min h(t), h( m ) , m ≤ t < 1. It is clear that hm (t) is nonnegative and continuous on [0, 1]. We define the operator Am by u(t), −τ ≤ t ≤ 0, B R 11m Am u(t) := G(t, s)hm (s)f (s, us + (u0 )s )ds, 0 ≤ t ≤ 1, 0 B2m u(t), 1 ≤ t ≤ b, where n−2 α R1 β e β t 0 g(0, s)hm (s)f (s, us + (u0 )s )ds, α R1 tn−2 B1m u(t) := (n−2)! 0 g(0, s)hm (s)f (s, us + (u0 )s )ds, 0, β 6= 0, α 6= 0, β 6= 0, α = 0, β=0 for each t ∈ [−τ, 0], and R δ n−2 − γ (t−1) 1 g(1, s)hm (s)f (s, us + (u0 )s )ds, −γ R e δ 0 1 tn−2 B2m u(t) := (n−2)! g(1, s)h (s)f (s, us + (u0 )s )ds, m 0 0, δ 6= 0, γ 6= 0, δ 6= 0, γ = 0, δ=0 for any t ∈ [1, b]. By Lemma 2.3, it is easy to check that Am : K → K. And, Am is continuous, the proof is similar to that of [11, Theorem 2.1]. Next let B ⊂ K be a bounded subset of K, and M1 > 0 be a constant such that kuk[−τ,b] ≤ M1 for u ∈ B. Noting that if xt ∈ C = C n−2 ([−τ, a], R), then (n−2) (n−2) ∈ C([−τ, a], R), and xt (θ) = x(n−2) (t + θ), θ ∈ [−θ, a]. Thus xt kus + (u0 )s k[−τ,a] = sup |(us + us0 )(n−2) (θ)| −τ ≤θ≤a ≤ ≤ (n−2) sup |(u(n−2) (s + θ)| + −τ ≤θ≤a sup |u0 (n−2) sup |u(n−2) (t)| + sup |u0 −τ ≤t≤b (s + θ)| −τ ≤θ≤a (t)| = kuk[−τ,b] + ku0 k[−τ,b] −τ ≤t≤b ≤ M1 + M0 := M2 (2.9) for u ∈ B and s ∈ [0, 1]. Hence, there exists a constant M3 > 0 such that |f (s, us + (u0 )s )| ≤ M3 , + on [0, 1] × C[0,M , 3] since f is continuous on [0, 1] × C + . For u ∈ B we have u)(n−2) (t), −τ ≤ t ≤ 0, (B R 1 1m (n−2) (Am u) (t) := g(t, s)h (s)f (s, u + (u ) )ds, 0 ≤ t ≤ 1, m s 0 s 0 (n−2) (B2m u) (t), 1 ≤ t ≤ b, (2.10) EJDE-2006/68 EXISTENCE OF POSITIVE SOLUTIONS 7 where ( α R1 e β t 0 g(0, s)hm (s)f (s, us + (u0 )s )ds, t ∈ [−τ, 0], β 6= 0, (B1m u)(n−2) (t) := 0, t ∈ [−τ, 0], β = 0, and (n−2) (B2m u) ( γ R1 e− δ (t−1) 0 g(1, s)hm (s)f (s, us + (u0 )s )ds, t ∈ [1, b], δ 6= 0, (t) := 0, t ∈ [1, b], δ = 0. These and (2.10) imply (Am u)(n−2) (t) is continuous and uniformly bounded for u ∈ B. So (Am u)0 (t) is continuous and uniformly bounded for u ∈ B also. The Ascoli-Arzela Theorem implies that Am is a completely continuous operator on K for any m ≥ 2. Moreover, Am converges uniformly to A as m → ∞ on any bounded subset of K. To see this, note that if u ∈ K with kuk[−τ,b] ≤ M , then from (H3) and 0 ≤ hn (s) ≤ h(s), (n−2) |(Am u) (n−2) (t) − (Au) Z m1 g(t, s)[h(s) − hm (s)]f (s, us + (u0 )s )ds (t)| = 0 Z 1 g(t, s)[h(s) − hm (s)]f (s, us + (u0 )s )ds + m−1 m 1 m Z ≤ g(s, s)|h(s) − hm (s)|f (s, us + (u0 )s )ds 0 Z 1 + m−1 m ≤ 2M hZ g(s, s)|h(s) − hm (s)|f (s, us + (u0 )s )ds 1 m Z 0 →0 where M := maxt∈[0,1],ϕ∈C + 1 g(s, s)h(s)ds + g(s, s)h(s)ds i m−1 m as m → ∞, f (t, ϕ). Thus, we have [0,M +M0 ] kAm u − Auk[−τ,b] = kAm u − Auk[0,1] → 0, n → ∞, for each u ∈ K with kuk[−τ,b] ≤ M . Hence, Am converges uniformly to A as m → ∞ and therefore A is completely continuous also. This completes the proof of Lemma 2.4. 8 C. BAI, Q. YANG, J. GE EJDE-2006/68 3. Main results For convenience, we introduce the following notation. Let Z 1−ε Z 1 −1 −1 ω= g(s, s)h(s)ds ; N= min g(t, s)h(s)ds ; ε≤t≤1−ε 0 ε f (t, ϕ) ρ + fσρ = inf min : ϕ ∈ C[σρ,ρ+M ; ] 0 ρ t∈[ε,1−ε] f (t, ϕ) + f0ρ = sup max : ϕ ∈ C[0,ρ+M ; ] 0 ρ t∈[0,1] f (t, ϕ) ; fµ = lim sup max kϕk[−τ,a] →µ t∈[0,1] kϕk[−τ,a] fµ = lim kϕk[−τ,a] →µ inf f (t, ϕ) , t∈[ε,1−ε] kϕk[−τ,a] min (µ := ∞ or 0+ ). Now, we impose conditions on f which we assure that iK (A, Kρ ) = 1. Lemma 3.1. Assume that f0ρ ≤ ω u 6= Au and for u ∈ ∂Kρ . (3.1) Then iK (A, Kρ ) = 1. Proof. For u ∈ ∂Kρ , we have kus + (u0 )s k[−τ,a] ≤ ρ + M0 , for all s ∈ [0, 1], i.e., us + (u0 )s ∈ C[0,ρ+M0 ] for any s ∈ [0, 1]. It follows from (3.1) that for t ∈ [0, 1], Z 1 (n−2) (Au) (t) = g(t, s)h(s)f (s, us + (u0 )s )ds 0 Z 1 ≤ g(s, s)h(s)f (s, us + (u0 )s )ds 0 Z 1 < ρω g(s, s)h(s)ds 0 = ρ = kuk[−τ,b] . This implies that kAuk[−τ,b] < kuk[−τ,b] for u ∈ ∂Kρ . By Lemma 2.2 (1), we have iK (A, Kρ ) = 1. Let u ∈ ∂Ωρ , then for any s ∈ [ε, 1 − ε], we have by Lemma 2.1(c) that kus + (u0 )s k[−τ,a] = sup (u(n−2) (s + θ) + (u0 )(n−2) (s + θ)) θ∈[−τ,a] ≥ sup u(n−2) (s + θ) (since (u0 )(n−2) (t) ≥ 0 for t ∈ [−τ, b]) θ∈[−τ,a] ≥ u(n−2) (s) ≥ min u(n−2) (t) = σρ. t∈[ε,1−ε] By Lemma 2.1(b), we have Ωρ ⊂ K ρ , that is kuk[−τ,b] ≤ ρ. Thus, from (2.9) we get kus + (u0 )s k[−τ,a] ≤ kuk[−τ,b] + ku0 k[−τ,b] ≤ ρ + M0 , ∀s ∈ [0, 1]. Hence + us + (u0 )s ∈ C[σρ, ρ+M0 ] , for u ∈ ∂Ωρ , s ∈ [ε, 1 − ε]. Next, we impose conditions on f which assure that iK (A, Ωρ ) = 0. (3.2) EJDE-2006/68 EXISTENCE OF POSITIVE SOLUTIONS 9 Lemma 3.2. If f satisfies the condition ρ fσρ ≥ Nσ u 6= Au and for u ∈ ∂Ωρ . (3.3) Then iK (A, Ωρ ) = 0. Proof. Let tn−1 )(n−1)! , n is odd, −τ ≤ t ≤ 0, − (1+τ tn n is even , −τ ≤ t ≤ 0, e(t) = (1+τ )2 n! , tn−1 , 0 ≤ t ≤ b. b(n−1)! It is easy to verify that e ∈ C (n−2) ([−τ, b], R), e(t) ≥ 0 for t ∈ [−τ, b], and t , n is odd, −τ ≤ t ≤ 0, − 1+τ (n−2) t2 e (t) = 2(1+τ )2 , n is even, −τ ≤ t ≤ 0, t 0 ≤ t ≤ b, b, which implies that e ∈ K and kek[−τ,b] = 1, that is e ∈ ∂K1 . We claim that u 6= Au + λe, u ∈ ∂Ωρ , λ > 0. In fact, if not, there exist u ∈ ∂Ωρ and λ0 > 0 such that u = Au + λ0 e. Then by (3.2) for t ∈ [ε, 1 − ε], we get u(n−2) (t) = (Au)(n−2) (t) + λ0 e(n−2) (t) Z 1 = g(t, s)h(s)f (s, us + (u0 )s )ds + λ0 e(n−2) (t) 0 1−ε Z ≥ g(t, s)h(s)f (s, us + (u0 )s )ds + λ0 ε 1−ε Z ≥ g(t, s)h(s)f (s, us + (u0 )s )ds + λ0 σ min ε≤t≤1−ε ε Z ≥ ρN σ ε b 1−ε min ε≤t≤1−ε g(t, s)h(s)ds + λ0 σ ε ≥ σρ + λ0 σ. This implies that σρ ≥ σρ + λ0 σ, a contradiction. Moreover, it is easy to check that u 6= Au for u ∈ ∂Ωρ from (3.3). Hence, by Lemma 2.2 (2), it follows that iK (A, Ωρ ) = 0. Theorem 3.3. If one of the following conditions holds: (H7) There exist ρ1 , ρ2 , ρ3 ∈ (0, ∞) with ρ1 < σρ2 and ρ2 < ρ3 such that f0ρ1 ≤ ω, ρ2 fσρ ≥ N σ, 2 u 6= Au for u ∈ ∂Ωρ2 and f0ρ3 ≤ ω. (H8) There exist ρ1 , ρ2 , ρ3 ∈ (0, ∞) with ρ1 < ρ2 < σρ3 such that ρ1 fσρ ≥ N σ, 1 f0ρ2 ≤ ω, u 6= Au for u ∈ ∂Kρ2 and ρ3 fσρ ≥ N σ. 3 Then BVP (1.1)-(1.4) has two positive solutions. Moreover, if in (H7), f0ρ1 ≤ ω is replaced by f0ρ1 < ω, then (1.1)-(1.4) has a third positive solution u3 ∈ Kρ1 . 10 C. BAI, Q. YANG, J. GE EJDE-2006/68 Proof. Suppose that (H7) holds. We show that either A has a fixed point u1 in ∂Kρ1 or in Ωρ2 \ Kρ1 . If u 6= Au for u ∈ ∂Kρ1 ∪ ∂Kρ3 , by Lemmas 3.1-3.2, we have iK (A, Kρ1 ) = 1, iK (A, Ωρ2 ) = 0 and iK (A, Kρ3 ) = 1. Since ρ1 < σρ2 , we have Kρ1 ⊂ Kσρ2 ⊂ Ωρ2 by Lemma 2.1 (b). It follows from Lemma 2.2 that A has a fixed point u1 ∈ Ωρ2 \ Kρ1 . Similarly, A has a fixed point u2 ∈ Kρ3 \ Ωρ2 . The proof is similar when (H8) holds. As a special case of Theorem 3.3 we obtain the following result. Corollary 3.4. Let ξ(t) ≡ 0, η(t) ≡ 0. If there exists ρ > 0 such that one of the following conditions holds: ρ ≥ N σ, u 6= Au for u ∈ ∂Ωρ and 0 ≤ f ∞ < ω. (H9) 0 ≤ f 0 < ω, fσρ ρ (H10) N < f0 ≤ ∞, f0 ≤ ω, u 6= Au for u ∈ ∂Kρ and N < f∞ ≤ ∞. Then BVP (1.1)-(1.4) has two positive solutions. Proof. From ξ(t) ≡ 0, η(t) ≡ 0, it is clear that u0 (t) ≡ 0 for t ∈ [−τ, b], thus M0 = 0. We now show that (H9 ) implies (H7 ). It is easy to verify that 0 ≤ f 0 < ω implies that there exists ρ1 ∈ (0, σρ) such that f0ρ1 < ω. Let k ∈ (f ∞ , ω). Then there exists + r > ρ such that maxt∈[0,1] f (t, ϕ) ≤ kkϕk[−τ,a] for ϕ ∈ C[r,∞) since 0 ≤ f ∞ < ω. Let l + l = max{ max f (t, ϕ) : ϕ ∈ C[0,r] }, and ρ3 > max , ρ . ω−k t∈[0,1] Then we have max f (t, ϕ) ≤ kkϕk[−τ,a] + l ≤ kρ3 + l < ωρ3 t∈[0,1] + for ϕ ∈ C[0,ρ . 3] This implies that f0ρ3 < ω and (H7) holds. Similarly, (H10) implies (H8). By a similar argument to that of Theorem 3.3, we obtain the following results on existence of at least one positive solution of (1.1)-(1.4). Theorem 3.5. If one of the following conditions holds: (H11) There exist ρ1 , ρ2 ∈ (0, ∞) with ρ1 < σρ2 such that f0ρ1 ≤ ω and ρ2 fσρ ≥ N σ. 2 (H12) There exist ρ1 , ρ2 ∈ (0, ∞) with ρ1 < ρ2 such that ρ1 fσρ ≥ Nσ 1 and f0ρ2 ≤ ω. Then BVP (1.1)-(1.4) has a positive solution. As a special case of Theorem 3.5, we obtain the following result. Corollary 3.6. Let ξ(t) ≡ 0, η(t) ≡ 0. If one of the following conditions holds: (H13) 0 ≤ f 0 < ω and N < f∞ ≤ ∞. (H14) 0 ≤ f ∞ < ω and N < f0 ≤ ∞. Then BVP (1.1)-(1.4) has a positive solution. EJDE-2006/68 EXISTENCE OF POSITIVE SOLUTIONS 11 References [1] C. Bai, J. Ma; Eigenvalue criteria for existence of multiple positive solutions to boundaryvalue problems of second-order delay differential equations, J. Math. Anal. Appl. 301 (2005) 457-476. [2] C. Bai, J. Fang; Existence of multiple positive solutions for functional differential equations, Comput. Math. Appl. 45(2003) 1797-1806. [3] J.M. Davis, K.R. Prasad, W.K.C. Yin; Nonlinear eigenvalue problems involving two classa of functional differential equations, Houston J. Math. 26(2000) 597-608. [4] K. Deimling; Nonlinear Functional Analysis, Springer-Verlag, New York, 1985. [5] L.H. Erbe, Q.K. Kong; Boundary value problems for singular second order functional differential equations, J. Comput. Appl. Math. 53 (1994) 640-648. [6] L.J. Grimm, K. Schmitt; Boundary value problem problems for differential equations with deviating arguments, Aequationes Math. 4 (1970) 176-190. [7] D. Guo, V. Lakshmikantham; Nonlinear Problems in Abstract Cones, Academic Press, New York, 1988. [8] G.B. Gustafson, K. Schmitt; Nonzero solutions of boundary-value problems for second order ordinary and delay-differential equations, J. Differential Equations 12 (1972) 129-147. [9] J. Henderson, Boundary Value Problems for Functional Differential Equations, World Scientific, 1995. [10] J. Henderson, W.K.C. Yin; Positive solutions and nonlinear eigenvalue problems for functional differential equations, Appl. Math. Lett. 12 (1999) 63-68. [11] C.H. Hong, C.C. Yeh, C.F. Lee, F. Hsiang, F.H. Wong; Existence of positive solutions for functional differential equations, Comput. Math. Appl. 40 (2000) 783-792. [12] C.H. Hong, F.H. Wong, C.C. Yeh; Existence of positive solutions for higher-order functional differential equations, J. Math Anal. Appl. 297(2004) 14-23. [13] G. L. Karakostas, K. G. Mavridis, P. Ch. Tsamotos; Multiple positive solutions for a functional second-order boundary-value problem, J. Math. Anal. Appl. 282 (2003) 567-577. [14] K.Q. Lan; Multiple positive solutions of semilinear differential equations with singularities, J. London. Math. Soc. 63(2001) 690-704. [15] S.K. Ntouyas, Y. Sficas, P.Ch. Tsamatos; An existence principle for boundary-value problems for second order functional differential equations, Nonlinear Anal. 20 (1993) 215-222. [16] Y. Shi, S. Chen; Multiple positive solutions of singular boundary value problems, Indian J. Pure Appl. Math. 30 (1999) 847-855. [17] D. Taunton, W.K.C. Yin; Existence of some functional differential equations, Comm. Appl. Nonlinear Anal. 4 (1997) 31-43. [18] P.Ch. Tsamatos; On a boundary-value problem for a system for functional differential equations with nonlinear boundary conditions, Funkcial. Ekvac. 42 (1999) 105-114. [19] P. Weng, D. Jiang; Existence of some positive solutions for boundary-value problem of secondorder FDE, J. Comput. Math. Appl. 37(1999) 1-9. Chuanzhi Bai Department of Mathematics, Huaiyin Teachers College, Huaian, Jiangsi 223001, China; and Department of Mathematics, Nanjing University, Nanjing 210093, China E-mail address: czbai8@sohu.com Qing Yang Department of Mathematics, Huaiyin Teachers College, Huaian, Jiangsi 223001, China E-mail address: yangqing3511115@163.com Jing Ge Department of Mathematics, Huaiyin Teachers College, Huaian, Jiangsi 223001, China E-mail address: gejing0512@163.com