Electronic Journal of Differential Equations, Vol. 2008(2008), No. 147, pp. 1–29. ISSN: 1072-6691. URL: http://ejde.math.txstate.edu or http://ejde.math.unt.edu ftp ejde.math.txstate.edu (login: ftp) MULTIPLE POSITIVE SOLUTIONS FOR SINGULAR m-POINT BOUNDARY-VALUE PROBLEMS WITH NONLINEARITIES DEPENDING ON THE DERIVATIVE YA MA, BAOQIANG YAN Abstract. Using the fixed point theorem in cones, this paper shows the existence of multiple positive solutions for the singular m-point boundary-value problem x00 (t) + a(t)f (t, x(t), x0 (t)) = 0, x0 (0) = 0, x(1) = m−2 X 0 < t < 1, ai x(ξi ), i=1 where 0 < ξ1 < ξ2 < · · · < ξm−2 < 1, ai ∈ [0, 1), i = 1, 2, . . . , m − 2, with P 0 0 < m−2 i=1 ai < 1 and f maybe singular at x = 0 and x = 0. 1. Introduction The study of multi-point boundary-value problems (BVP) for linear second-order ordinary differential equations was initiated by Il’in and Moiseev [5, 6]. Since then, many authors have studied general nonlinear multi-point BVP; see for examples [4, 17], and references therein. Gupta, Ntouyas and Tsamatos [4] considered the existence of a solution in C 1 [0, 1] for the m-point boundary-value problem x00 (t) = f (t, x(t), x0 (t)) + e(t), x0 (0) = 0, x(1) = m−2 X 0 < t < 1, ai x(ξi ), (1.1) i=1 where ξi ∈ (0, 1), i = 1, 2, . . . , m − 2, 0 < ξ1 < ξ2 < · · · < ξm−2 < 1, ai ∈ R, Pm−2 i = 1, 2, . . . , m−2, have the same sign, i=1 ai 6= 1, e ∈ L1 [0, 1], f : [0, 1]×R2 → R is a function satisfying Carathèodory conditions and a growth condition of the form |f (t, u, v)| ≤ p1 (t)|u| + q1 (t)|v| + r1 (t), where p1 , q1 , r1 ∈ L1 [0, 1]. Recently, using Leray-Schauder continuation theorem, Ma and O’Regan proved the existence of positive solutions of C 1 [0, 1) solutions for the above BVP, where f : [0, 1] × R2 → R satisfies the Carathèodory conditions (see [17]). Khan and Webb [10] obtained 2000 Mathematics Subject Classification. 34B10, 34B15. Key words and phrases. m-point boundary-value problem; singularity; positive solutions; fixed point theorem. c 2008 Texas State University - San Marcos. Submitted August 13, 2008. Published October 24, 2008. Supported by grants 10871120 from the National Natural Science, and Y2005A07 from the Natural Science of Shandong Province, China. 1 2 Y. MA, B. YAN EJDE-2008/147 a interesting result which presents the multiplicity of existence of at least three solutions of a second-order three-point boundary-value problem. However, up to now, there are a fewer results on the existence of multiple solutions to (1.1). In view of the importance of the research on the multiplicity of positive solutions for differential equations [1, 2, 5, 9, 10, 11, 12, 14, 15], the goal of this paper is to fill this gap in the literature. There are main four sections in this paper. In section 2, we give a special cone and its properties. In section 3, using the theory of fixed point index on a cone, we present the existence of multiple positive solutions to (1.1) with f may be singular at x0 = 0 but not at x = 0. In section 4, under the condition f is singular at x0 = 0 and x = 0, we present the existence of multiple positive solutions to (1.1). In section 5, under the condition f is singular at x = 0 but not at x0 = 0, we present the existence of multiple positive solutions to (1.1). 2. Preliminaries Let C [0, 1] = {x : [0, 1] → R such that x(t) be continuous on [0, 1] and x0 (t) continuous on [0, 1] } with norm kxk = max{γkxk1 , γδkxk2 }, where 1 kxk2 = max |x0 (t)|, kxk1 = max |x(t)|, t∈[0,1] t∈[0,1] Pm−2 ai (1 − ξi ) , Pm−2 1 − i=1 ai ξi i=1 γ= δ= m−2 X ai (1 − ξi ). i=1 Let P = {x ∈ C 1 [0, 1] : x(t) ≥ γkxk1 , ∀t ∈ [0, 1], x(0) ≥ δkxk2 }. Obviously, C 1 [0, 1] is a Banach space and P is a cone in C 1 [0, 1]. Lemma 2.1. Let Ω be a bounded open set in real Banach space E, θ ∈ Ω, P be a cone in E and A : Ω̄ ∩ P → P be continuous and completely continuous. Suppose λAx 6= x, ∀x ∈ ∂Ω ∩ P, λ ∈ (0, 1]. (2.1) Then i(A, Ω ∩ P, P ) = 1. Lemma 2.2. Let Ω be a bounded open set in real Banach space E, θ ∈ Ω, P be a cone in E and A : Ω̄ ∩ P → P be continuous and completely continuous. Suppose Ax 6≤ x, ∀x ∈ ∂Ω ∩ P. (2.2) Then i(A, Ω ∩ P, P ) = 0. Let R+ = (0, +∞), R− = (−∞, 0), R = (−∞, +∞). The following conditions will be used in this article. a(t) ∈ C(0, 1) ∩ L1 [0, 1], a(t) > 0, t ∈ (0, 1); f ∈ C([0, 1] × R+ × R− , [0, +∞)); (2.3) (2.4) There exists g ∈ C([0, +∞) × (−∞, 0], [0, +∞)) such that f (t, x, y) ≤ g(x, y), ∀(t, x, y) ∈ [0, 1] × R+ × R− . (2.5) EJDE-2008/147 MULTIPLE POSITIVE SOLUTIONS 3 For x ∈ P and t ∈ [0, 1], define operator Z (Ax)(t) = − t (t − s)a(s)f (s, x(s)x0 (s))ds 0 + − 1− m−2 X Z 1 Pm−2 ai i=1 ξi Z ai 1 (1 − s)a(s)f (s, x(s), x0 (s))ds 0 (2.6) (ξi − s)a(s)f (s, x(s), x0 (s)) ds . 0 i=1 Lemma 2.3 ([17]). Assume (2.1). Then for y ∈ C[0, 1] the problem x00 + y(t) = 0, t ∈ (0, 1) 0 x (0) = m−2 X 0 bi x (ξi ), x(1) = i=1 m−2 X ai x(ξi ) (2.7) i=1 has a unique solution Z x(t) = − t (t − s)y(s)ds + M t + N, (2.8) 0 where, Rξ bi 0 i a(s)y(s)ds , M= Pm−2 i=1 bi − 1 m−2 Z 1 X Z (1 − s)a(s)y(s)ds − ai Pm−2 i=1 N= 1 Pm−2 1 − i=1 ai 0 i=1 Pm−2 R ξi m−2 X bi 0 a(s)y(s)ds (1 − ai ξi ) . − i=1Pm−2 i=1 bi − 1 i=1 ξi (ξi − s)a(s)y(s)ds 0 Further, if y ≥ 0, for all t ∈ [0, 1], x satisfies inf x(t) ≥ γkxk1 , (2.9) t∈[0,1] where γ = Pm−2 i=1 Pm−2 ai (1 − ξi ) / 1 − i=1 ai ξi . Lemma 2.4. Suppose (2.3)–(2.5)hold. Then A : P → P is a completely continuous operator. 4 Y. MA, B. YAN EJDE-2008/147 Proof. For x ∈ P , from (2.6), one has Z 1 (Ax)(t) ≥ − (1 − s)a(s)f (s, x(s), x0 (s))ds 0 + − 1− m−2 X ai i=1 ξi Z ai − 0 i=1 ai P m−2 i=1 m−2 X ai (2.10) 1 Z (1 − s)a(s)f (s, x(s), x0 (s))ds 0 1 Z (ξi − s)a(s)f (s, x(s), x0 (s))ds ai 0 i=1 Pm−2 ai (1 − ξi ) Pm−2 1 − i=1 ai i=1 ≥ 0 i=1 1− (1 − s)a(s)f (s, x(s), x0 (s))ds (ξi − s)a(s)f (s, x(s), x0 (s))ds Pm−2 ≥ 1 Z 1 Pm−2 1 Z a(s)f (s, x(s), x0 (s))ds ≥ 0, t ∈ [0, 1], 0 1 Z (t − s)a(s)f (s, x(s)x0 (s))ds |(Ax)(t)| = − 0 + − 1− m−2 X ≤ ≤ ai i=1 ξi Z ai 1 (1 − s)a(s)f (s, x(s), x0 (s))ds 0 (ξi − s)a(s)f (s, x(s), x0 (s))ds 0 i=1 ≤ Z 1 Pm−2 1− 1 Pm−2 1− 1 Pm−2 1− 1 Pm−2 i=1 i=1 i=1 1 Z ai (1 − s)a(s)f (s, x(s), x0 (s))ds 0 1 Z ai a(s)f (s, x(s), x0 (s))ds 0 1 Z a(s)ds ai 0 max 0≤c≤kxk,−kxk≤c0 ≤0 g(c, c0 ) < +∞, t ∈ [0, 1], (2.11) Z t |(Ax)0 (t)| = − a(s)f (s, x(s)x0 (s))ds 0 Z t = a(s)f (s, x(s)x0 (s))ds 0 Z (2.12) 1 0 ≤ a(s)f (s, x(s)x (s))ds 0 Z ≤ 1 a(s)ds 0 max 0≤c≤kxk,−kxk≤c0 ≤0 g(c, c0 ) < +∞, which implies that A is well defined. Z 1 1 0 (Ax)(0) = Pm−2 ( (1 − s)a(s)f (s, x(s), x (s))ds 1 − i=1 ai 0 EJDE-2008/147 MULTIPLE POSITIVE SOLUTIONS − m−2 X Z 1− − 1 Pm−2 i=1 m−2 X Z = m−2 X ai ( Z 1 ai (1 − s)a(s)f (s, x(s), x0 (s))ds 0 i=1 1 (ξi − s)a(s)f (s, x(s), x0 (s))ds) ai 0 i=1 m−2 X (ξi − s)a(s)f (s, x(s), x0 (s))ds) 0 i=1 ≥ ξi ai 5 Z 1 a(s)f (s, x(s), x0 (s))ds. ai (1 − ξi ) 0 i=1 On the other hand, kAxk2 = max |(Ax)0 (t)| t∈[0,1] Z = max | − t∈[0,1] Z = t a(s)f (s, x(s)x0 (s))ds| 0 1 a(s)f (s, x(s)x0 (s))ds. 0 Then (Ax)(0) ≥ δkAxk2 . (2.13) By Lemma 2.3, we have (Ax)(t) ≥ γkAxk1 . As a result Ax ∈ P , which implies AP ⊆ P . By a standard argument, we know that A : P → P is continuous and completely continuous. 3. Singularities at x0 = 0 but not at x = 0 In this section the nonlinearity f may be singular at x0 = 0 but not at x = 0. We will assume that the following conditions hold. (H1) a(t) ∈ C(0, 1) ∩ L1 [0, 1], a(t) > 0, t ∈ (0, 1) (H2) f (t, u, z) ≤ h(u)[g(z) + r(z)], where f ∈ C([0, 1] × R+ × R− , R+ ), g(z) > 0 continuous and nondecreasing on R− , h(u) ≥ 0 continuous and nondecreasing on R+ , r(z) > 0 continuous and non-increasing on (−∞, 0]; (H3) c > 1, sup Pm−2 R a ξ +1 i i −1 (h(c) 1 a(s)ds) c∈R+ − i=1 Pm−2 I 0 1− a i=1 R0 i du where I(z) = z g(u)+r(u) , z ∈ R− ; (H4) There exists a function g1 ∈ C([0, +∞) × (−∞, 0], [0, +∞)), such that f (t, u, z) ≥ g1 (u, z), ∀(t, u, z) ∈ [0, 1] × R+ × R− , and limu→+∞ g1 (u,z) = u +∞, uniformly for z ∈ R− . (H5) There exists a function ΨH ∈ C([0, 1], R+ ) and a constant 0 ≤ δ < 1 such that f (t, u, z) ≥ ΨH (t)uδ , for all (t, u, z) ∈ [0, 1] × [0, H] × R− . 6 Y. MA, B. YAN EJDE-2008/147 For n ∈ {1, 2, . . . } and x ∈ P , define operator Z t 1 (An x)(t) = − (t − s)a(s)f (s, x(s), −|x0 (s)| − )ds n 0 Z 1 1 1 (1 − s)a(s)f (s, x(s), −|x0 (s)| − )ds + Pm−2 (3.1) n 1 − i=1 ai 0 m−2 X Z ξi 1 − ai (ξi − s)a(s)f (s, x(s), −|x0 (s)| − )ds , t ∈ [0, 1]. n 0 i=1 Theorem 3.1. Suppose (H1)–(H5) hold. Then (1.1) has at least two positive solutions x0,1 , x0,2 ∈ C 1 [0, 1] ∩ C 2 (0, 1) with x0,1 (t), x0,2 (t) > 0, t ∈ [0, 1]. Proof. Choose R1 > 0 such that R1 Pm−2 R1 a ξ +1 i i i=1 Pm−2 I −1 (h(R1 ) 0 − 1− a i i=1 > 1. (3.2) a(s)ds) From the continuity of I −1 and h, we can choose ε > 0 and ε < R1 with R1 Pm−2 R1 a ξ +1 i i i=1 Pm−2 I −1 (h(R1 ) 0 − 1− i=1 ai > 1, (3.3) a(s)ds + I(−ε)) n0 ∈ {1, 2, . . . } with n10 < min{ε, δ/2} and let N0 = {n0 , n0 + 1, . . . }. Lemma 2.4 guarantees that for n ∈ N0 , An : P → P is a completely continuous operator. Let Ω1 = {x ∈ C 1 [0, 1] : kxk < R1 }. We show that x 6= µAn x, ∀x ∈ P ∩ ∂Ω1 , µ ∈ (0, 1], n ∈ N0 . (3.4) In fact, if there exists an x0 ∈ P ∩ ∂Ω1 and µ0 ∈ (0, 1] such that x0 = µ0 An x0 , Z t 1 x0 (t) = −µ0 (t − s)a(s)f (s, x0 (s), −|x00 (s)| − )ds n 0 Z 1 µ0 1 + (1 − s)a(s)f (s, x0 (s), −|x00 (s)| − )ds Pm−2 n 1 − i=1 ai 0 Z m−2 ξi X 1 − ai (ξi − s)a(s)f (s, x0 (s), −|x00 (s)| − )ds , t ∈ [0, 1]. n 0 i=1 Then Z t 1 )ds, ∀t ∈ [0, 1]. (3.5) n 0 Obviously, x00 (t) ≤ 0, t ∈ (0, 1), and since x0 (1) > 0, x0 (t) > 0, t ∈ [0, 1]. Differentiating (3.5), we have x00 (t) = −µ0 a(s)f (s, x0 (s), −|x00 (s)| − 1 ) = 0, 0 < t < 1, n m−2 X x0 (1) = ai x0 (ξi ). x000 (t) + µ0 a(t)f (t, x0 (t), x00 (t) − x00 (0) = 0, i=1 (3.6) EJDE-2008/147 MULTIPLE POSITIVE SOLUTIONS 7 Then 1 ) n −x000 (t) = µ0 a(t)f (t, x0 (t), x00 (t) − ≤ a(t)h(x0 (t))[g(x00 (t) − g(x00 (t) − −x000 (t) 1 0 n ) + r(x0 (t) − n1 ) 1 1 ) + r(x00 (t) − )], n n ≤ a(t)h(x0 (t)), ∀t ∈ (0, 1), ∀t ∈ (0, 1). Integrating from 0 to t, we have I(x00 (t) 1 1 − ) − I(− ) ≤ n n t Z Z 0 1 − ) ≤ h(R1 ) n Z Z t Then x00 (t) ≥I −1 a(s)ds, 0 and I(x00 (t) t a(s)h(x0 (s))ds ≤ h(R1 ) t a(s)ds + I(−ε). 0 (h(R1 ) a(s)ds + I(−ε)); 0 that is, −x00 (t) ≤ −I −1 Z (h(R1 ) t a(s)ds + I(−ε)), t ∈ (0, 1). (3.7) 0 Then Rξ ai 0 i −x00 (s)ds Pm−2 1 − i=1 ai 0 1 − i=1 ai Z 1 Z s 1 ≤ −I −1 h(R1 ) a(τ )dτ + I(−ε) ds Pm−2 1 − i=1 ai 0 0 Pm−2 Z ξi Z s −1 i=1 ai −I (h(R ) + a(τ )dτ + I(−ε))ds P 1 m−2 1 − i=1 ai 0 0 Z 1 Z 1 1 −I −1 h(R1 ) ≤ a(τ )dτ + I(−ε) ds Pm−2 1 − i=1 ai 0 0 Pm−2 Z ξi Z 1 −1 i=1 ai −I (h(R1 ) a(τ )dτ + I(−ε))ds + Pm−2 1 − i=1 ai 0 0 Pm−2 Z 1 1 + i=1 ai ξi −1 −I h(R1 ) a(s)ds + I(−ε) . = Pm−2 1 − i=1 ai 0 x0 (0) = µ0 Pm−2 Z 1 −x00 (s)ds − µ0 Pm−2 i=1 Since x0 (0) ≥ x0 (t) ≥ γkx0 k1 ≥ γx0 (0) ≥ γδkx0 k2 , x0 (0) ≥ kx0 k = R1 , R1 Pm−2 R1 1+ i=1 ai ξi −1 − 1−Pm−2 a I (h(R1 ) 0 i i=1 ≤ 1, (3.8) a(s)ds + I(−ε)) which is a contradiction to (3.3). Then (3.4) holds. From Lemma 2.1, for n ∈ N0 , i(An , Ω1 ∩ P, P ) = 1. (3.9) Now we show that there exists a set Ω2 such that An x 6≤ x, ∀x ∈ ∂Ω2 ∩ P. (3.10) 8 Y. MA, B. YAN EJDE-2008/147 Choose a∗ with 0 < a∗ < 1. Let −1 1 N∗ = + 1. Pm−2 R1 i=1 ai (1−ξi ) Pm−2 γa∗ 1− a(s)ds 0 a i i=1 From (H4), there exists R2 > R1 such that g1 (x, y) ≥ N ∗ x, 1 Let Ω2 = {x ∈ C [0, 1] : kxk < R2 a∗ }. ∀x ≥ R2 , y ∈ R− . (3.11) Then Ax 6≤ x, ∀x ∈ ∂Ω2 ∩ P. In fact, if there exists x0 ∈ ∂Ω2 ∩ P with x0 ≥ An x0 . By the definition of the cone and Lemma 2.3, one has R2 x0 (t) ≥ γkx0 k1 ≥ γx(0) ≥ γδkx0 k2 , x0 (t) ≥ ∗ > R2 , ∀t ∈ [0, 1], a from (3.11), γx0 (t) ≥ γAn x0 (t) Z t 1 =γ − (t − s)a(s)f (s, x0 (s), −|x00 (s)| − )ds n 0 Z 1 1 1 + (1 − s)a(s)f (s, x0 (s), −|x00 (s)| − )ds Pm−2 n 1 − i=1 ai 0 Z m−2 ξi X 1 ai (ξi − s)a(s)f (s, x0 (s), −|x00 (s)| − )ds − n 0 i=1 Z t 1 ≥γ − (t − s)a(s)f (s, x0 (s), −|x00 (s)| − )ds n 0 Z 1 1 1 + (1 − s)a(s)f (s, x0 (s), −|x00 (s)| − )ds Pm−2 n 1 − i=1 ai 0 m−2 X Z ξi 1 − ai (ξi − s)a(s)f (s, x0 (s), −|x00 (s)| − )ds n 0 i=1 Pm−2 a Z t 1 i i=1 ≥γ (t − s)a(s)f (s, x0 (s), −|x00 (s)| − )ds P m−2 n 1 − i=1 ai 0 Pm−2 R 1 ai 0 (ξi − s)a(s)f (s, x0 (s), −|x00 (s)| − n1 )ds − i=1 Pm−2 1 − i=1 ai Pm−2 Z γ i=1 ai (1 − ξi ) 1 1 = a(s)f (s, x0 (s), −|x00 (s)| − )ds Pm−2 n 1 − i=1 ai 0 Pm−2 Z 1 γ i=1 ai (1 − ξi ) 1 ≥ a(s)g1 (x0 (s), −|x00 (s)| − )ds Pm−2 n 1 − i=1 ai 0 Pm−2 Z γ i=1 ai (1 − ξi ) 1 ≥ a(s)dsN ∗ x0 (s) Pm−2 1 − i=1 ai 0 Pm−2 Z γ i=1 ai (1 − ξi ) 1 R2 R2 ≥ a∗ a(s)dsN ∗ ∗ > ∗ . Pm−2 a a 1 − i=1 ai 0 EJDE-2008/147 MULTIPLE POSITIVE SOLUTIONS 9 2 Then kx0 k ≥ γkx0 k1 > R a∗ , which is a contradiction to x0 ∈ ∂Ω2 ∩ P . Then (3.10) holds. From Lemma 2.2, i(An , Ω2 ∩ P, P ) = 0. (3.12) which with (3.9) guarantee that i(An , (Ω2 − Ω̄1 ) ∩ P, P ) = −1. (3.13) From this equality and (3.9), An has two fixed points with xn,1 ∈ Ω1 ∩ P, xn,2 ∈ (Ω2 − Ω̄1 ) ∩ P . For each n ∈ N0 , there exists xn,1 ∈ Ω1 ∩ P such that xn,1 = An xn,1 ; that is, Z t 1 xn,1 (t) = − (t − s)a(s)f (s, xn,1 (s), −|x0n,1 (s)| − )ds n 0 Z 1 1 1 + (1 − s)a(s)f (s, xn,1 (s), −|x0n,1 (s)| − )ds (3.14) Pm−2 n 1 − i=1 ai 0 m−2 X Z ξi 1 − ai (ξi − s)a(s)f (s, xn,1 (s), −|x0n,1 (s)| − )ds . n 0 i=1 As in the proof of (3.5), we have x0n,1 (t) ≤ 0, t ∈ (0, 1) and Z t 1 x0n,1 (t) = − a(s)f (s, xn,1 (s), x0n,1 (s) − )ds, n ∈ N0 , t ∈ (0, 1). n 0 Now we consider {xn,1 (t)}n∈N0 and {x0n,1 (t)}n∈N0 . Since kxn,1 k ≤ R1 , it follows that {xn,1 (t)} is uniformly bounded on [0, 1], (3.15) {x0n,1 (t)} (3.16) is uniformly bounded on [0, 1]. Then {xn,1 (t)} is equicontinuous on [0, 1]. As in the proof as (3.6), 1 x00n,1 (t) + a(t)f (t, xn,1 (t), x0n,1 (t) − ) = 0, 0 < t < 1, n m−2 X x0n,1 (0) = 0, xn,1 (1) = ai xn,1 (ξi ). (3.17) (3.18) i=1 Now we show that for all t1 , t2 ∈ [0, 1], |I(x0n,1 (t2 ) − 1 1 ) − I(x0n,1 (t1 ) − )| ≤ h(R1 )| n n Z t2 a(t)dt|. (3.19) t1 From (3.18), 1 ) n 1 1 ≤ a(t)h(xn,1 (t)) g x0n,1 (t) − + r(x0n,1 (t) − ) , n n −x00n,1 (t) = a(t)f (t, xn,1 (t), x0n,1 (t) − ∀t ∈ (0, 1), and 1 n 1 1 0 ≥ −a(t)h(xn,1 (t)) g xn,1 (t) − + r(x0n,1 (t) − ) , n n x00n,1 (t) = −a(t)f t, xn,1 (t), x0n,1 (t) − ∀t ∈ (0, 1), 10 Y. MA, B. YAN so −x00n,1 (t) g(x0n,1 (t) − n1 ) + r(x0n,1 (t) − n1 ) EJDE-2008/147 ≤ a(t)h(xn,1 (t)), ∀t ∈ (0, 1), (3.20) and x00n,1 (t) g(x0n,1 (t) − n1 ) + r(x0n,1 (t) − n1 ) ≥ −a(t)h(xn,1 (t)), ∀t ∈ (0, 1). (3.21) Then, for all t1 , t2 in [0, 1] and t1 < t2 , Z t2 Z t2 1 1 0 − a(t)dt 1 1 d(xn,1 (s) − n ) ≤ h(R1 ) 0 0 t1 g(xn,1 (s) − n ) + r(xn,1 (s) − n ) t1 Z t2 = h(R1 )| a(t)dt|, t1 Inequality (3.19) holds. Since I −1 is uniformly continuous on [0, I(−R1 − )], for all ¯ > 0, there exists 0 > 0 such that |I −1 (s1 ) − I −1 (s2 )| < ¯, ∀|s1 − s2 | < 0 , s1 , s2 ∈ [0, I(−R1 − )]. (3.22) And (3.19) guarantees that for 0 > 0, there exists δ 0 > 0 such that 1 1 ) − I(x0n,1 (t1 ) − )| < 0 , n n From this inequality and (3.22), |I(x0n,1 (t2 ) − ∀ |t1 − t2 | < δ 0 , t1 , t2 ∈ [0, 1]. 1 1 − (x0n,1 (t1 ) − )| n n 1 1 = |I −1 (I(x0n,1 (t2 ) − )) − I −1 (I(x0n,1 (t1 ) − ))| n n < ¯, ∀|t1 − t2 | < δ 0 , t1 , t2 ∈ [0, 1]; (3.23) |x0n,1 (t2 ) − x0n,1 (t1 )| = |x0n,1 (t2 ) − (3.24) that is, {x0n,1 (t)} is equicontinuous on [0, 1]. (3.25) {x0n,1 (t)} From (3.15)–(3.17), (3.25) and the Arzela-Ascoli Theorem, {xn,1 (t)} and are relatively compact on C[0, 1], which implies there exists a subsequence {xnj ,1 } of {xn,1 } and function x0,1 (t) ∈ C[0, 1] such that lim max |xnj ,1 (t) − x0,1 (t)| = 0, j→+∞ t∈[0,1] Since x0nj ,1 (0) = 0, xnj ,1 (1) = (0, 1), j ∈ {1, 2, . . . }, x00,1 (0) = 0, x0,1 (1) = m−2 X Pm−2 i=1 lim max |x0nj ,1 (t) − x00,1 (t)| = 0. j→+∞ t∈[0,1] ai xnj ,1 (ξi ), x0nj ,1 (t) < 0, xnj ,1 (t) > 0, t ∈ ai x0,1 (ξi ), x00,1 (t) ≤ 0, x0,1 (t) ≥ 0, t ∈ (0, 1). i=1 (3.26) For (t, xnj ,1 (t), x0nj ,1 (t) − n1 ) ∈ [0, 1] × [0, R1 + ] × (−∞, 0), from (H5) there exists a function ΨR1 ∈ C([0, 1], R+ ) such that f (t, xnj ,1 (t), x0nj ,1 (t) − 1 )ds ≥ ΨR1 (t)(xnj ,1 (t))δ , nj 0 ≤ δ < 1. EJDE-2008/147 MULTIPLE POSITIVE SOLUTIONS 11 Then, for n ∈ N0 , t Z xnj ,1 (t) = − 0 + − (t − s)a(s)f (s, xnj ,1 (s), x0nj ,1 (s) − 1− m−2 X 1 Pm−2 i=1 Z 0 i=1 1 Z ≥− 0 + − ξi ai Z ai 1 (1 − s)a(s)f (s, xnj ,1 (s), x0nj ,1 (s) − 0 (ξi − s)a(s)f (s, xnj ,1 (s), x0nj ,1 (s) − (1 − s)a(s)f (s, xnj ,1 (s), x0nj ,1 (s) − 1− m−2 X 1 Pm−2 i=1 Z ξi ai 0 i=1 1 )ds nj Z ai 0 1 )ds nj 1 )ds nj 1 )ds nj 1 (1 − s)a(s)f (s, xnj ,1 (s), x0nj ,1 (s) − (ξi − s)a(s)f (s, xnj ,1 (s), x0nj ,1 (s) − 1 )ds nj 1 )ds nj Pm−2 Z 1 1 i=1 ai ≥ (1 − s)a(s)f (s, xnj ,1 (s), x0nj ,1 (s) − )ds Pm−2 n j 1 − i=1 ai 0 Pm−2 Z 1 1 i=1 ai − (ξi − s)a(s)f (s, xnj ,1 (s), x0nj ,1 (s) − )ds P m−2 n j 1 − i=1 ai 0 Pm−2 Z 1 ai (1 − ξi ) 1 = i=1 Pm−2 a(s)f (s, xnj ,1 (s), x0nj ,1 (s) − )ds n j 1 − i=1 ai 0 Pm−2 Z ai (1 − ξi ) 1 a(s)ΨR1 (s)(xnj ,1 (s))δ ds Pm−2 1 − i=1 ai 0 Pm−2 Z 1 ai (1 − ξi ) ≥ i=1 Pm−2 a(s)ΨR1 (s)γ δ dskxnj ,1 kδ1 , 1 − i=1 ai 0 ≥ i=1 which implies 1 1−δ Pm−2 a (1 − ξ ) Z 1 i i i=1 a(s)ΨR1 (s)γ δ ds , kxnj ,1 k1 ≥ Pm−2 1 − i=1 ai 0 and xnj ,1 (t) ≥ 1 Pm−2 a (1 − ξ ) Z 1 1−δ i i δ i=1 a(s)Ψ (s)γ ds = a0 > 0. Pm−2 R1 1 − i=1 ai 0 Thus x0nj ,1 (t) = − Z t a(s)f (s, xnj ,1 (s), x0nj ,1 (s) − 0 Z 1 )ds nj t a(s)ΨR1 (s)(xnj ,1 (s))δ ds ≤− 0 Z ≤− 0 t a(s)ΨR1 (s)dsaδ0 , t ∈ [0, 1], n ∈ N0 . 12 Y. MA, B. YAN EJDE-2008/147 Consequently, inf min |x0nj ,1 (s)| > 0, j≥1 s∈[ 1 ,t] 2 inf min |x0nj ,1 (s)| > 0, j≥1 s∈[t, 1 ] 2 1 t ∈ [ , 1), 2 1 t ∈ (0, ]. 2 Since 1 x0nj ,1 (t) − x0nj ,1 ( ) = − 2 Z t 1/2 a(s)f (s, xnj ,1 (s), x0nj ,1 (s) − 1 )ds, nj t ∈ (0, 1), and f (t, xnj ,1 (t), x0nj ,1 (t) − 1 1 1 ) ≤ h(xnj ,1 (t))[g(x0nj ,1 (t) − ) + r(x0nj ,1 (t) − )] nj nj nj Z t R1 R1 ≤ h( )[g(− a(s)ΨR1 (s)dsaδ0 ) + r(− − )], γ γδ 0 letting j → +∞, the Lebesgue Dominated Convergence Theorem guarantees that Z t 1 0 0 a(s)f (s, x0,1 (s), x00,1 (s))ds, t ∈ (0, 1). (3.27) x0,1 (t) − x0,1 ( ) = − 2 1/2 Differentiating, we have x000,1 (t) + a(t)f (t, x0,1 (t), x00,1 (t)) = 0, 0 < t < 1, and from (3.26) x0,1 (t) is a positive solution of (1.1) with x0,1 ∈ C 1 [0, 1] ∩ C 2 (0, 1). For the set {xn,2 }n∈N0 ⊆ (Ω2 − Ω1 ) ∩ P , the proof is as that for the set {xn,1 }n∈N0 . We can obtain a convergent subsequence {xni ,2 }n∈N0 of {xn,2 }n∈N0 with limi→+∞ xni ,2 = x0,2 ∈ C 1 [0, 1]∩C 2 (0, 1). Moreover, x0,2 is a positive solution to (1.1). Example 3.1 In (1.1), let f (t, u, z) = µ[1 + (−z)−a ][1 + ub + ud ] and a(t) ≡ 1 with 0 < a < 1, b > 1, 0 < d < 1 and µ > 0. If µ < sup P c(1− m−2 ai ) ) i ξi i=1 . 1 + cb + cd i=1 I(− 1+Pm−2 a c∈R+ (3.28) Then (1.1) has at least two positive solutions x0,1 , x0,2 ∈ C 1 [0, 1] ∩ C 2 (0, 1). We apply Theorem 3.1 with g(z) = (−z)−a , r(z) = 1, h(u) = µ(1 + ub + d u ), Ψ(t) = µ, g1 (u, z) = µub . (H1), (H2), (H4), (H5) hold. Also sup c∈R+ c Pm−2 R1 i=1 ai ξi +1 −1 Pm−2 − 1− I (h(c) 0 i=1 ai = sup c∈R+ a(s)ds) c Pm−2 i=1 ai ξi +1 −1 Pm−2 − 1− I (µ(1 i=1 ai and (3.28) guarantees that (H3) holds. , + cb + cd )) EJDE-2008/147 MULTIPLE POSITIVE SOLUTIONS 13 4. Singularities at x0 = 0 and x = 0 In this section the nonlinearity f may be singular at x0 = 0 and x = 0. We assume that the following conditions hold. (P1) a(t) ∈ C[0, 1], a(t) > 0, t ∈ (0, 1); (P2) f (t, u, z) ≤ [h(u) + ω(u)][g(z) + r(z)], where f ∈ C([0, 1] × R+ × R− , R+ ), g(z) > 0 continuous and non-increasing on (−∞, 0], ω(u) > 0 continuous and non-increasing on [0, +∞), h(u) ≥ 0 continuous and nondecreasing on R+ , r(z) > 0 continuous and nondecreasing on R− ; (P3) c > 1, sup Pm−2 Rc a ξ +1 i i −1 [− max c∈R+ − i=1 Pm−2 (I a(t)(ch(c) + ω(s)ds)]) t∈[0,1] 0 1− a i=1 i R0 Ra udu where I(z) = z g(u)+r(u) , z ∈ R− , 0 ω(s)ds < +∞; (P4) There exists a function g1 ∈ C([0, +∞) × (−∞, 0], [0, +∞)), such that f (t, u, z) ≥ g1 (u, z), ∀(t, u, z) ∈ [0, 1] × R+ × R− , and limu→+∞ g1 (u,z) = u +∞, uniformly for z ∈ R− . (P5) There exists a function ΨH ∈ C([0, 1], R+ ) with f (t, u, z) ≥ ΨH (t), for all (t, u, z) ∈ [0, 1] × [0, H] × [−H, 0). For n ∈ {1, 2, . . . }, x ∈ P , t ∈ [0, 1], define operator Z t 1 1 (An x)(t) = − (t − s)a(s)f (s, x(s) + , −|x0 (s)| − )ds n n 0 Z 1 1 1 1 + (1 − s)a(s)f (s, x(s) + , −|x0 (s)| − )ds (4.1) Pm−2 n n 1 − i=1 ai 0 m−2 X Z ξi 1 1 ai − (ξi − s)a(s)f (s, x(s) + , −|x0 (s)| − )ds . n n 0 i=1 Theorem 4.1. Suppose (P1)–(P5) hold. Then (1.1) has at least two positive solutions x0,1 , x0,2 ∈ C 1 [0, 1] ∩ C 2 (0, 1) and x0,1 (t), x0,2 (t) > 0, t ∈ [0, 1]. Proof. Choose R1 > 0 such that R1 Pm−2 i=1 ai ξi +1 Pm−2 − 1− (I −1 [− maxt∈[0,1] i=1 ai a(t)(R1 h(R1 ) + R R1 0 > 1. (4.2) ω(s)ds)]) From the continuity of I −1 and h, we can choose > 0 and < R1 such that R1 Pm−2 i=1 ai ξi +1 Pm−2 − 1− (I −1 [I(−) i=1 ai − maxt∈[0,1] a(t)((R1 + )h(R1 + ) + R R1 + 0 ω(s)ds)]) (4.3) is greater than 1, n0 ∈ {1, 2, . . . } with n10 < min{, δ/2} and let N0 = {n0 , n0 + 1, . . . }. Then Lemma 2.4 guarantees that for n ∈ N0 , An : P → P is a completely continuous operator. Let Ω1 = {x ∈ C 1 [0, 1] : kxk < R1 }. We show that x 6= µAn x, ∀x ∈ P ∩ ∂Ω1 , µ ∈ (0, 1], n ∈ N0 . (4.4) 14 Y. MA, B. YAN EJDE-2008/147 In fact, if there exists an x0 ∈ P ∩ ∂Ω1 and µ0 ∈ (0, 1] with x0 = µ0 An x0 , Z t 1 1 x0 (t) = −µ0 (t − s)a(s)f (s, x0 (s) + , −|x00 (s)| − )ds n n 0 Z 1 µ0 1 1 0 + Pm−2 ( (1 − s)a(s)f (s, x0 (s) + , −|x0 (s)| − )ds n n 1 − i=1 ai 0 Z m−2 ξi X 1 1 − ai (ξi − s)a(s)f (s, x0 (s) + , −|x00 (s)| − )ds), t ∈ [0, 1]. n n 0 i=1 Then x00 (t) = −µ0 Z t a(s)f (s, x0 (s) + 0 1 1 , −|x00 (s)| − )ds, n n ∀t ∈ [0, 1]. (4.5) Obviously, x00 (t) ≤ 0, t ∈ (0, 1), and since x0 (1) > 0, x0 (t) > 0, t ∈ [0, 1]. Differentiating (4.5), we have 1 0 1 , x (t) − ) = 0, 0 < t < 1, n 0 n m−2 X x0 (1) = ai x0 (ξi ). x000 (t) + µ0 a(t)f (t, x0 (t) + x00 (0) = 0, (4.6) i=1 Then, for t ∈ (0, 1), 1 0 1 , x (t) − ) n 0 n 1 1 1 1 ≤ a(t) h x0 (t) + + ω(x0 (t) + ) g x00 (t) − + r(x00 (t) − ) , n n n n −x000 (t) = µ0 a(t)f (t, x0 (t) + and −x000 (t) 1 1 1 1 ≤ a(t)[h(x0 (t) + n ) + ω(x0 (t) + n )], 0 0 g(x0 (t) − n ) + r(x0 (t) − n ) and ∀t ∈ (0, 1), −x000 (t)(x00 (t) − n1 ) g(x00 (t) − n1 ) + r(x00 (t) − n1 ) 1 1 1 ≥ a(t)[h(x0 (t) + ) + ω(x0 (t) + )](x00 (t) − ) n n n (4.7) 1 1 0 ≥ a(t)[h(R1 + ) + ω(x0 (t) + (1 − t))](x0 (t) − ) n n 1 1 1 0 = a(t)[h(R1 + )(x0 (t) − ) + ω(x0 (t) + (1 − t))(x00 (t) − )] n n n Integrating from 0 to t, we have 1 1 I(x00 (t) − ) − I(− ) n n Z t 1 1 1 ≥ a(s) h(R1 + )(x00 (s) − ) + ω(x0 (s) + (1 − s))(x00 (s) − ) ds n n n 0 Z t Z t Z t 1 1 ≥ max a(t)h(R1 + ) x00 (s)ds − ds + ω(x0 (s) + (1 − s))d(x0 (s) n n t∈[0,1] 0 0 0 1 + (1 − s)) n EJDE-2008/147 MULTIPLE POSITIVE SOLUTIONS 15 R1 + Z ≥ − max a(t)(h(R1 + )(R1 + ) + ω(s)ds), t∈[0,1] 0 I(x00 (t) − 1 ) ≥ I(−) − max a(t)(h(R1 + )(R1 + ) + n t∈[0,1] Z R1 + ω(s)ds). 0 Then R1 + Z x00 (t) ≥ I −1 I(−) − max {a(t)} h(R1 + )(R1 + ) + t∈[0,1] ω(s)ds ; 0 that is, −x00 (t) ≤ −I −1 R1 + Z I(−)− max a(t)(h(R1 +)(R1 +)+ t∈[0,1] ω(s)ds) , t ∈ (0, 1). 0 (4.8) Since x0 (0) = ≤ 1− µ0 Pm−2 i=1 Z ai −x00 (s)ds − µ0 Rξ ai 0 i −x00 (s)ds Pm−2 1 − i=1 ai Pm−2 i=1 0 Z 1 Pm−2 1 1 −I −1 I(−) − max {a(t)}(h(R1 + )(R1 + ) 1 − i=1 ai 0 Z R1 + + ω(s)ds) ds + 0 t∈[0,1] Pm−2 1− Z i=1 ai P m−2 i=1 ξi ai 0 Z R1 + − max a(t)(h(R1 + )(R1 + ) + t∈[0,1] = 1+ −I −1 I(−) ω(s)ds) ds 0 Pm−2 ai ξi (−I −1 I(−) − max a(t)(h(R1 + )(R1 + ) Pi=1 m−2 t∈[0,1] 1 − i=1 ai Z R1 + + ω(s)ds) . 0 Since x0 (0) ≥ x0 (t) ≥ γkx0 k1 ≥ γx0 (0) ≥ γδkx0 k2 , x0 (0) ≥ kx0 k = R1 . So R1 Pm−2 i=1 ai ξi +1 Pm−2 − 1− (I −1 [I(−) i=1 ai − max{a(t)}((R1 + )h(R1 + ) + R R1 + 0 ≤ 1, ω(s)ds)]) (4.9) which is a contradiction to (4.3). Then (4.4) holds. From Lemma 2.1, for n ∈ N0 , i(An , Ω1 ∩ P, P ) = 1. (4.10) Now we show that there exists a set Ω2 such that An x 6≤ x, ∀x ∈ ∂Ω2 ∩ P. Choose a∗ , N ∗ as in section 3. Let Ω2 = {x ∈ C 1 [0, 1] : kxk < R2 }. a∗ Then An x 6≤ x, ∀x ∈ ∂Ω2 ∩ P. (4.11) 16 Y. MA, B. YAN EJDE-2008/147 In fact, if there exists x0 ∈ ∂Ω2 ∩ P with x0 ≥ An x0 . By the definition of the cone and Lemma 2.3, one has x0 (t) ≥ γkx0 k1 ≥ γx(0) ≥ γδkx0 k2 , and so x0 (t) ≥ R2 a∗ > R2 , for all t ∈ [0, 1], x0 (t) + 1 n > R2 , from (3.11), γx0 (t) ≥ γAn x0 (t) Z t 1 1 (t − s)a(s)f (s, x0 (s) + , −|x00 (s)| − )ds =γ − n n 0 Z 1 1 1 1 + (1 − s)a(s)f (s, x0 (s) + , −|x00 (s)| − )ds Pm−2 n n 1 − i=1 ai 0 Z m−2 ξ i X 1 1 − ai (ξi − s)a(s)f (s, x0 (s) + , −|x00 (s)| − )ds n n 0 i=1 Z t 1 1 ≥γ − (t − s)a(s)f (s, x0 (s) + , −|x00 (s)| − )ds n n 0 Z 1 1 1 1 + (1 − s)a(s)f (s, x0 (s) + , −|x00 (s)| − )ds Pm−2 n n 1 − i=1 ai 0 m−2 X Z ξi 1 1 − ai (ξi − s)a(s)f (s, x0 (s) + , −|x00 (s)| − )ds n n 0 i=1 P Z t m−2 1 1 i=1 ai ≥γ (t − s)a(s)f (s, x0 (s) + , −|x00 (s)| − )ds P m−2 n n 1 − i=1 ai 0 Pm−2 R 1 1 1 ai 0 (ξi − s)a(s)f s, x0 (s) + n , −|x00 (s)| − n ds − i=1 Pm−2 1 − i=1 ai Pm−2 Z γ i=1 ai (1 − ξi ) 1 1 1 = a(s)f (s, x0 (s) + , −|x00 (s)| − )ds Pm−2 n n 1 − i=1 ai 0 Pm−2 Z γ i=1 ai (1 − ξi ) 1 1 1 ≥ a(s)g1 (x0 (s) + , −|x00 (s)| − )ds Pm−2 n n 1 − i=1 ai 0 Pm−2 Z 1 γ i=1 ai (1 − ξi ) ≥ a(s)dsN ∗ x0 (s) Pm−2 1 − i=1 ai 0 Pm−2 Z 1 R2 R2 ∗γ i=1 ai (1 − ξi ) a(s)dsN ∗ ∗ > ∗ . ≥a Pm−2 a a 1 − i=1 ai 0 Then kx0 k ≥ γkx0 k1 > holds. From Lemma 2.2, R2 a∗ , which is a contradiction to x0 ∈ ∂Ω2 ∩ P . Then (4.11) i(An , Ω2 ∩ P, P ) = 0. (4.12) This equality and (4.10) guarantee, i(An , (Ω2 − Ω̄1 ) ∩ P, P ) = −1. (4.13) From this equality and (4.10), An has two fixed points with xn,1 ∈ Ω1 ∩ P, xn,2 ∈ (Ω2 − Ω1 ) ∩ P . For each n ∈ N0 , there exists xn,1 ∈ Ω1 ∩ P with xn,1 = An xn,1 ; EJDE-2008/147 MULTIPLE POSITIVE SOLUTIONS 17 that is, t Z (t − s)a(s)f (s, xn,1 (s) + xn,1 (t) = − 0 + − 1 Pm−2 1− i=1 m−2 X Z ai 1 1 , −|x0n,1 (s)| − )ds n n 1 (1 − s)a(s)f (s, xn,1 (s) + 0 ξi Z (ξi − s)a(s)f (s, xn,1 (s) + ai 0 i=1 1 1 , −|x0n,1 (s)| − )ds n n 1 1 , −|x0n,1 (s)| − )ds . n n (4.14) As in the same proof of (3.5), x0n (t) ≤ 0, t ∈ (0, 1) and Z t 1 1 a(s)f (s, xn,1 (s) + , x0n,1 (s) − )ds, x0n,1 (t) = − n n 0 n ∈ N0 , t ∈ (0, 1). Now we consider {xn,1 (t)}n∈N0 and {x0n,1 (t)}n∈N0 , since kxn,1 k ≤ R1 , it follows that {xn,1 (t)} is uniformly bounded on [0, 1], (4.15) {x0n,1 (t)} (4.16) is uniformly bounded on [0, 1]. Then {xn,1 (t)} is equicontinuous on [0, 1]. (4.17) As in the same proof of (3.6), 1 0 1 , x (t) − ) = 0, 0 < t < 1, n n,1 n m−2 X xn,1 (1) = ai xn,1 (ξi ). x00n,1 (t) + a(t)f (t, xn,1 (t) + x0n,1 (0) = 0, (4.18) i=1 Now we show for all t1 , t2 in [0, 1], 1 1 |I(x0n,1 (t2 ) − ) − I(x0n,1 (t1 ) − )| n n ≤ max a(t) h(R1 + )(|xn,1 (t2 ) − xn,1 (t1 )| + |t2 − t1 |) t∈[0,1] Z 1 xn,1 (t2 )+ n (1−t2 ) +| (4.19) ω(s)dt| . 1 xn,1 (t1 )+ n (1−t1 ) From (4.18), it follows that for t ∈ (0, 1), 1 0 1 , x (t) − ) n n,1 n 1 1 1 1 ≤ a(t)[h(xn,1 (t) + ) + ω(xn,1 (t) + )][g(x0n,1 (t) − ) + r(x0n,1 (t) − )], n n n n −x00n,1 (t) = a(t)f (t, xn,1 (t) + 1 0 1 , x (t) − ) n n,1 n 1 1 1 1 ≥ −a(t)[h(xn,1 (t) + ) + ω(xn,1 (t) + )][g(x0n,1 (t) − ) + r(x0n,1 (t) − )], n n n n x00n,1 (t) = −a(t)f (t, xn,1 (t) + 18 Y. MA, B. YAN EJDE-2008/147 and so for t ∈ (0, 1), −x00n,1 (t)(x0n,1 (t) − n1 ) g(x0n,1 (t) − n1 ) + r(x0n,1 (t) − n1 ) 1 1 1 ≥ a(t)[h(xn,1 (t) + ) + ω(xn,1 (t) + )](x0n,1 (t) − ) n n n 1 1 1 ≥ a(t)[h(R1 + )(x0n,1 (t) − ) + ω(xn,1 (t) + (1 − t))(xn,1 (t) + (1 − t))0 ], n n n (4.20) x00n,1 (t)(x0n,1 (t) − n1 ) g(x0n,1 (t) − n1 ) + r(x0n,1 (t) − n1 ) 1 1 1 ≤ −a(t)[h(xn,1 (t) + ) + ω(xn,1 (t) + )](x0n,1 (t) − ) n n n 1 1 1 ≤ −a(t)[h(R1 + )(x0n,1 (t) − ) + ω(xn,1 (t) + (1 − t))(xn,1 (t) + (1 − t))0 ]. n n n (4.21) Then, for all t1 , t2 ∈ [0, 1] and t1 < t2 , 1 1 I(x0n,1 (t2 ) − ) − I(x0n,1 (t1 ) − ) n n Z t2 Z xn,1 (t2 )+ n1 (1−t2 ) 1 0 ≥ max a(t)[h(R1 + )( xn,1 (s)dt − (t2 − t1 ) + ω(s)dt] 1 n t∈[0,1] t1 (1−t1 ) xn,1 (t1 )+ n ≥ − max a(t) h(R1 + )(|xn,1 (t2 ) − xn,1 (t1 )| + |t2 − t1 |) t∈[0,1] Z 1 (1−t2 ) xn,1 (t2 )+ n +| ω(s)dt| , 1 xn,1 (t1 )+ n (1−t1 ) 1 1 ) − I(x0n,1 (t2 ) − ) n n ≤ max a(t) h(R1 + )(|xn,1 (t2 ) − xn,1 (t1 )| + |t2 − t1 |) I(x0n,1 (t1 ) − t∈[0,1] Z 1 xn,1 (t2 )+ n (1−t2 ) +| ω(s)dt| . 1 xn,1 (t1 )+ n (1−t1 ) Therefore, (4.19) holds. Since I −1 is uniformly continuous on [0, I(−R1 − )], for all ¯ > 0, there exists 0 > 0 such that |I −1 (s1 ) − I −1 (s2 )| < ¯, ∀ |s1 − s2 | < 0 , s1 , s2 ∈ [0, I(−R1 − )]. Then (4.19) guarantees that for 0 > 0, there exists δ 0 > 0 such that 1 1 |I(x0n,1 (t2 ) − ) − I(x0n,1 (t1 ) − )| < 0 , ∀|t1 − t2 | < δ 0 , t1 , t2 ∈ [0, 1]. n n From this inequality and (4.22), 1 1 |x0n,1 (t2 ) − x0n,1 (t1 )| = |x0n,1 (t2 ) − − (x0n,1 (t1 ) − )| n n 1 1 = |I −1 (I(x0n,1 (t2 ) − )) − I −1 (I(x0n,1 (t1 ) − ))| n n < ¯, ∀|t1 − t2 | < δ 0 , t1 , t2 ∈ [0, 1]; (4.22) (4.23) (4.24) EJDE-2008/147 MULTIPLE POSITIVE SOLUTIONS 19 that is, {x0n,1 (t)} is equicontinuous on [0, 1]. (4.25) 0 From (4.15)–(4.17), (4.25) and the Arzela-Ascoli Theorem, {xn,1 (t)} and {xn,1 (t)} are relatively compact on C[0, 1], which implies, there exists a subsequence {xnj ,1 } of {xn,1 } and function x0,1 (t) ∈ C[0, 1] such that lim max |xnj ,1 (t) − x0,1 (t)| = 0, j→+∞ t∈[0,1] Since x0nj ,1 (0) = 0, xnj ,1 (1) = (0, 1), j ∈ {1, 2, . . . }, x00,1 (0) = 0, x0,1 (1) = m−2 X Pm−2 i=1 lim max |x0nj ,1 (t) − x00,1 (t)| = 0. j→+∞ t∈[0,1] ai xnj ,1 (ξi ), x0nj ,1 (t) < 0, xnj ,1 (t) > 0, t ∈ ai x0,1 (ξi ), x00,1 (t) ≤ 0, x0,1 (t) ≥ 0, t ∈ (0, 1). (4.26) i=1 For (t, xnj ,1 (t) + n1j , x0nj ,1 (t) − n1j ) ∈ [0, 1] × [0, R1 + ] × (−∞, 0), from (P5) there exists a function ΨR1 ∈ C([0, 1], R+ ) such that 1 1 f (t, xnj ,1 (t) + , x0nj ,1 (t) − )ds ≥ ΨR1 (t). nj nj Then, for n ∈ N0 , Z t 1 1 xnj ,1 (t) = − (t − s)a(s)f (s, xnj ,1 (s) + , x0nj ,1 (s) − )ds nj nj 0 Z 1 1 1 1 + (1 − s)a(s)f (s, xnj ,1 (s) + , x0nj ,1 (s) − )ds Pm−2 nj nj 1 − i=1 ai 0 m−2 X Z ξi 1 1 − ai (ξi − s)a(s)f (s, xnj ,1 (s) + , x0nj ,1 (s) − )ds nj nj 0 i=1 Z 1 1 1 ≥− (1 − s)a(s)f (s, xnj ,1 (s) + , x0nj ,1 (s) − )ds n n j j 0 Z 1 1 1 1 0 + Pm−2 ( (1 − s)a(s)f (s, xnj ,1 (s) + , xnj ,1 (s) − )ds nj nj 1 − i=1 ai 0 Z m−2 ξi X 1 1 − ai (ξi − s)a(s)f (s, xnj ,1 (s) + , x0nj ,1 (s) − )ds) n n j j 0 i=1 Pm−2 Z 1 1 1 i=1 ai ≥ (1 − s)a(s)f (s, xnj ,1 (s) + , x0nj ,1 (s) − )ds P m−2 n n j j 1 − i=1 ai 0 Pm−2 Z 1 1 1 i=1 ai − (ξi − s)a(s)f (s, xnj ,1 (s) + , x0nj ,1 (s) − )ds P m−2 n n j j 1 − i=1 ai 0 Pm−2 Z 1 ai (1 − ξi ) 1 1 a(s)f (s, xnj ,1 (s) + , x0nj ,1 (s) − )ds = i=1 Pm−2 n n j j 1 − i=1 ai 0 Pm−2 Z 1 ai (1 − ξi ) a(s)ΨR1 (s)ds = a0 , ≥ i=1 Pm−2 1 − i=1 ai 0 and x0nj ,1 (t) Z =− t a(s)f (s, xnj ,1 (s) + 0 1 0 1 ,x (s) − )ds nj nj ,1 nj 20 Y. MA, B. YAN Z EJDE-2008/147 t ≤− a(s)ΨR1 + (s)ds, t ∈ [0, 1], n ∈ N0 . 0 Thus, inf min{ min xnj ,1 (s), min |x0nj ,1 (s)|} > 0, 1 1 j≥1 s∈[ 2 ,t] s∈[ 2 ,t] inf min{ min1 xnj ,1 (s), min1 |x0nj ,1 (s)|} > 0, j≥1 s∈[t, 2 ] s∈[t, 2 ] 1 t ∈ [ , 1), 2 1 t ∈ (0, ]. 2 From (P2), we have 1 0 1 ,x (t) − nj nj ,1 nj 1 1 1 1 ≤ [h(xnj ,1 (t) + ) + ω(xnj ,1 (t) + )][g(x0nj ,1 (t) − ) + r(x0nj ,1 (t) − )] nj nj nj nj Z t R1 R1 + ) + ω(a0 )][r(− − )]. a(s)ΨR1 + (s)ds) + g(− ≤ [h( γ γδ 0 f t, xnj ,1 (t) + and since x0nj ,1 (t) − 1 x0nj ,1 ( ) 2 Z t =− a(s)f (s, xnj ,1 (s) + 1/2 1 0 1 ,x (s) − )ds, nj nj ,1 nj t ∈ (0, 1), letting j → +∞, the Lebesgue Dominated Convergence Theorem guarantees that Z t 1 a(s)f (s, x0,1 (s), x00,1 (s))ds, t ∈ (0, 1). (4.27) x00,1 (t) − x00,1 ( ) = − 2 1/2 Differentiating, we have x000,1 (t) + a(t)f (t, x0,1 (t), x00,1 (t)) = 0, 0 < t < 1, and from (4.26), x0,1 (t) is a positive solution of (1.1) with x0,1 ∈ C 1 [0, 1] ∩ C 2 (0, 1). For the set {xn,2 }n∈N0 ⊆ (Ω2 − Ω1 ) ∩ P , as in proof for the set {xn,1 }n∈N0 , we obtain a convergent subsequence {xni ,2 }n∈N0 of {xn,2 }n∈N0 with limi→+∞ xni ,2 = x0,2 ∈ C 1 [0, 1] ∩ C 2 (0, 1). Moreover, x0,2 is a positive solution to (1.1). Example 4.1 In (1.1), let f (t, u, z) = µ[1 + (−z)−a ][1 + ub + u−d ] and a(t) ≡ 1 with 0 < a < 1, b > 1, 0 < d < 1 and µ > 0. If µ < sup c∈R+ P c(1− m−2 ai ) ) i i +1 i=1 I(− Pm−2 i=1 a ξ maxt∈[0,1] a(t)(c + c1−d + c1+b 1+b ) . (4.28) Then equation (1.1) has at least two positive solutions x0,1 , x0,2 ∈ C 1 [0, 1]∩C 2 (0, 1). We apply Theorem 4.1 with g(z) = 1, r(z) = (−z)−a , h(u) = µu−d , ω(u) = µ(1 + ub ), Ψ(t) = µ, g1 (u, z) = µub . Note that (P1), (P2), (P4) and (P5) hold, and that c sup Pm−2 Rc a ξ +1 i i c∈R+ − i=1 P I −1 [maxt∈[0,1] a(t)(ch(c) + 0 ω(s)ds)] 1− m−2 a i=1 = sup c∈R+ i c Pm−2 i=1 ai ξi +1 −1 Pm−2 − 1− I [µ maxt∈[0,1] i=1 ai Then (4.28) guarantees that (P3) holds. a(t)(c + c1−d + c1+b 1+b )] EJDE-2008/147 MULTIPLE POSITIVE SOLUTIONS 21 5. Singularities at x = 0 but not x0 = 0 In this section the nonlinearity f may be singular at x = 0, but not at x0 = 0. We assume that the following conditions hold. (S1) a(t) ∈ C[0, 1], a(t) > 0, t ∈ (0, 1); (S2) f (t, u, z) ≤ [h(u) + ω(u)]r(z), where f ∈ C([0, 1] × R+ × R− , R+ ), ω(u) > 0 continuous and non-increasing on [0, +∞), h(u) ≥ 0 continuous and nondecreasing on R+ , r(z) > 0 continuous and nondecreasing on R− ; (S3) sup c∈R+ c Pm−2 i=1 ai ξi +1 Pm−2 − 1− (I −1 [− maxt∈[0,1] i=1 ai a(t)(ch(c) + Rc 0 > 1, ω(s)ds)]) Ra R 0 udu where I(z) = z r(u) , z ∈ R− , 0 ω(s)ds < +∞, a ∈ R+ ; (S4) There exists a function g1 ∈ C([0, +∞) × (−∞, 0], [0, +∞)), such that f (t, u, z) ≥ g1 (u, z), ∀(t, u, z) ∈ [0, 1] × R+ × R− , and limu→+∞ g1 (u,z) = u +∞, uniformly for z ∈ R− . (S5) There exists a function ΨH ∈ C([0, 1], R+ ) and a constant 0 ≤ δ such that f (t, u, z) ≥ ΨH (t)(−z)δ , for all (t, u, z) ∈ [0, 1] × [0, H] × [−H, 0). For n ∈ {1, 2, . . . }, x ∈ P , define operator Z t 1 (An x)(t) = − (t − s)a(s)f (s, x(s) + , −|x0 (s)|)ds n 0 Z 1 1 1 + (1 − s)a(s)f (s, x(s) + , −|x0 (s)|)ds Pm−2 (5.1) n 1 − i=1 ai 0 m−2 X Z ξi 1 − ai (ξi − s)a(s)f (s, x(s) + , −|x0 (s)|)ds , t ∈ [0, 1]. n 0 i=1 Theorem 5.1. Suppose (S1)–(S5) hold. Then (1.1) has at least two positive solutions x0,1 , x0,2 in C 1 [0, 1] ∩ C 2 (0, 1) with x0,1 (t), x0,2 (t) > 0, t ∈ [0, 1]. Proof. Choose R1 > 0 such that R1 Pm−2 i=1 ai ξi +1 Pm−2 − 1− (I −1 [− maxt∈[0,1] i=1 ai a(t)(R1 h(R1 ) + R R1 0 > 1. (5.2) ω(s)ds)]) From the continuity of I −1 and h, we can choose > 0 and < R1 such that R1 Pm−2 i=1 ai ξi +1 Pm−2 − 1− (I −1 [− maxt∈[0,1] i=1 ai a(t)(R1 h(R1 + ) + R R1 + 0 > 1. (5.3) ω(s)ds)]) Let n0 ∈ {1, 2, . . . } with n10 < min{, δ/2} and let N0 = {n0 , n0 + 1, . . . }. Then Lemma 2.4 guarantees that for n ∈ N0 , An : P → P is a completely continuous operator. Let Ω1 = {x ∈ C 1 [0, 1] : kxk < R1 }. We show that x 6= µAn x, ∀x ∈ P ∩ ∂Ω1 , µ ∈ (0, 1], n ∈ N0 . (5.4) 22 Y. MA, B. YAN EJDE-2008/147 In fact, if there exists an x0 ∈ P ∩ ∂Ω1 and µ0 ∈ (0, 1] with x0 = µ0 An x0 , Z t 1 x0 (t) = −µ0 (t − s)a(s)f (s, x0 (s) + , −|x00 (s)|)ds n 0 Z 1 1 µ0 (1 − s)a(s)f (s, x0 (s) + , −|x00 (s)|)ds + Pm−2 n 1 − i=1 ai 0 m−2 X Z ξi 1 − ai (ξi − s)a(s)f (s, x0 (s) + , −|x00 (s)|)ds . n 0 i=1 Then Z t 1 , −|x00 (s)|)ds, ∀t ∈ [0, 1]. (5.5) n 0 Obviously, x00 (t) ≤ 0, t ∈ (0, 1), and since x0 (1) > 0, x0 (t) > 0, t ∈ [0, 1]. Differentiating (5.5), we have x00 (t) = −µ0 a(s)f (s, x0 (s) + x000 (t) + µ0 a(t)f (t, x0 (t) + x00 (0) = 0, 1 , n x00 (t)) = 0, 0 < t < 1, m−2 X x0 (1) = (5.6) ai x0 (ξi ). i=1 and −x000 (t) = µ0 a(t)f (t, x0 (t) + ≤ a(t)[h(x0 (t) + Then 1 0 , x (t)) n 0 1 1 ) + ω(x0 (t) + )]r(x00 (t)), n n −x000 (t) 1 1 ≤ a(t)[h(x0 (t) + ) + ω(x0 (t) + )], 0 r(x0 (t)) n n ∀t ∈ (0, 1). ∀t ∈ (0, 1), and 1 −x000 (t)x00 (t) 1 ≥ a(t)[h(x0 (t) + ) + ω(x0 (t) + )]x00 (t) 0 r(x0 (t)) n n 1 0 ≥ a(t)[h(R1 + ) + ω(x0 (t) + )]x0 (t). n Integrating from 0 to t, we have Z t 1 0 I(x0 (t)) ≥ a(s)[h(R1 + )x00 (s)) + ω(x0 (s) + )x00 (s)]ds n 0 Z t 1 ≥ max a(t) [h(R1 + )x00 (s)ω(x0 (s) + )x00 (s)]ds n t∈[0,1] 0 Z t 1 1 ≥ − max a(t)(h(R1 + )R1 + ω(x0 (s) + )d(x0 (s) + )) n n t∈[0,1] 0 Z x0 (t)+ n1 = − max a(t)(h(R1 + )R1 + ω(s)ds) t∈[0,1] 1 x0 (0)+ n Z ≥ − max a(t)(h(R1 + )R1 + t∈[0,1] R1 + ω(s)ds) 0 EJDE-2008/147 MULTIPLE POSITIVE SOLUTIONS and R1 + Z I(x00 (t)) ≥ − max a(t)(h(R1 + )R1 + 23 ω(s)ds). t∈[0,1] 0 Then x00 (t) ≥ I −1 (− max a(t)(h(R1 + )R1 + R1 + Z ω(s)ds)); t∈[0,1] 0 that is, −x00 (t) ≤ −I −1 (− max a(t)(h(R1 + )R1 + t∈[0,1] Z R1 + ω(s)ds)), t ∈ (0, 1). (5.7) 0 Since x0 (0) Rξ ai 0 i −x00 (s)ds = − Pm−2 1 − i=1 ai 0 1 − i=1 ai Z 1 Z R1 + 1 −1 ≤ −I − max a(t)(h(R + )R + ω(s)ds) ds Pm−2 1 1 t∈[0,1] 1 − i=1 ai 0 0 Pm−2 Z ξi Z R1 + −1 i=1 ai + −I I(− max a(t)(h(R + )(R ) + ω(s)ds)) ds Pm−2 1 1 t∈[0,1] 1 − i=1 ai 0 0 Pm−2 Z R1 + 1 + i=1 ai ξi (−I −1 − max a(t)(h(R1 + )R1 + ω(s)ds) , = Pm−2 t∈[0,1] 1 − i=1 ai 0 (5.8) x0 (0) ≥ x0 (t) ≥ γkx0 k1 ≥ γx0 (0) ≥ γδkx0 k2 , x0 (0) ≥ kx0 k = R1 . So µ0 Pm−2 Z 1 −x00 (s)ds µ0 Pm−2 i=1 R1 Pm−2 i=1 ai ξi +1 Pm−2 (I −1 [− maxt∈[0,1] − 1− i=1 ai a(t)(R1 h(R1 + ) + R R1 + 0 ≤ 1, (5.9) ω(s)ds)]) which is a contradiction to (5.3). Then (5.4) holds. From Lemma 2.1, for n ∈ N0 , i(An , Ω1 ∩ P, P ) = 1. (5.10) Now we show that there exists a set Ω2 such that An x 6≤ x, ∀ x ∈ ∂Ω2 ∩ P. (5.11) Choose a∗ , N ∗ as in section 3. Let Ω2 = {x ∈ C 1 [0, 1] : kxk < R2 }. a∗ Then An x 6≤ x, ∀x ∈ ∂Ω2 ∩ P. In fact, if there exists x0 ∈ ∂Ω2 ∩ P with x0 ≥ An x0 , by the definition of the cone and Lemma 2.3, one has x0 (t) ≥ γkx0 k1 ≥ γx(0) ≥ γδkx0 k2 , x0 (t) ≥ R2 a∗ > R2 for all t ∈ [0, 1]. Then x0 (t) + 1 n > R2 . From (3.11), γx0 (t) ≥ γAn x0 (t) Z t 1 =γ − (t − s)a(s)f (s, x0 (s) + , x00 (s))ds n 0 24 Y. MA, B. YAN + − 1− m−2 X 1 Pm−2 i=1 ai 1 (1 − s)a(s)f (s, x0 (s) + 0 ξi Z (ξi − s)a(s)f (s, x0 (s) + ai 0 i=1 Z t Z ≥γ − (t − s)a(s)f (s, x0 (s) + 0 + − 1− m−2 X i=1 EJDE-2008/147 1 Pm−2 i=1 Z ai 1 (1 − s)a(s)f (s, x0 (s) + 0 (ξi − s)a(s)f (s, x0 (s) + ai 1 0 , x0 (s))ds n 1 0 , x (s))ds n 0 ξi Z 1 0 , x (s))ds n 0 0 1 0 , x (s))ds n 0 1 0 , x0 (s))ds n Pm−2 a Z t 1 i i=1 ≥γ (t − s)a(s)f (s, x0 (s) + , x00 (s))ds P m−2 n 1 − i=1 ai 0 Pm−2 R 1 ai 0 (ξi − s)a(s)f (s, x0 (s) + n1 , x00 (s) − i=1 ) Pm−2 1 − i=1 ai Pm−2 Z γ i=1 ai (1 − ξi ) 1 1 = a(s)f (s, x0 (s) + , x00 (s))ds Pm−2 n 1 − i=1 ai 0 Pm−2 Z 1 γ i=1 ai (1 − ξi ) 1 ≥ a(s)g1 (x0 (s) + , x00 (s))ds Pm−2 n 1 − i=1 ai 0 Pm−2 Z 1 γ i=1 ai (1 − ξi ) ≥ a(s)dsN ∗ x0 (s) Pm−2 1 − i=1 ai 0 Pm−2 Z 1 R2 R2 ∗γ i=1 ai (1 − ξi ) ≥a a(s)dsN ∗ ∗ > ∗ ; Pm−2 a a 1 − i=1 ai 0 2 that is, kx0 k ≥ γkx0 k1 > R a∗ , which is a contradiction to x0 ∈ ∂Ω2 ∩ P . Then (5.11) holds. From Lemma 2.2, i(An , Ω2 ∩ P, P ) = 0. (5.12) This equality and (5.10) guarantee i(An , (Ω2 − Ω̄1 ∩ P, P ) = −1. (5.13) From this equality and (5.10), An has two fixed points with xn,1 ∈ Ω1 ∩ P, xn,2 ∈ (Ω2 − Ω̄1 ) ∩ P . For each n ∈ N0 , there exists xn,1 ∈ Ω1 ∩ P such that xn,1 = An xn,1 ; that is, for t ∈ [0, 1], Z t 1 xn,1 (t) = − (t − s)a(s)f (s, xn,1 (s) + , −|x0n,1 (s)|)ds n 0 Z 1 1 1 + (1 − s)a(s)f (s, xn,1 (s) + , −|x0n,1 (s)|)ds (5.14) Pm−2 n 1 − i=1 ai 0 m−2 X Z ξi 1 − ai (ξi − s)a(s)f (s, xn,1 (s) + , −|x0n,1 (s)|)ds . n 0 i=1 EJDE-2008/147 MULTIPLE POSITIVE SOLUTIONS As in the proof of (3.5), x0n (t) ≤ 0, t ∈ (0, 1) and Z t 1 x0n,1 (t) = − a(s)f (s, xn,1 (s) + , x0n,1 (s))ds, n 0 25 n ∈ N0 , t ∈ (0, 1). Now we consider {xn,1 (t)}n∈N0 and {x0n,1 (t)}n∈N0 , since kxn,1 k ≤ R1 , {xn,1 (t)} is uniformly bounded on [0, 1], (5.15) {x0n,1 (t)} (5.16) is uniformly bounded on [0, 1]. Then {xn,1 (t)} is equicontinuous on [0, 1]. (5.17) As in the proof of (3.6), 1 0 , x (t)) = 0, 0 < t < 1, n n,1 m−2 X x0n,1 (0) = 0, xn,1 (1) = ai xn,1 (ξi ). x00n,1 (t) + a(t)f (t, xn,1 (t) + (5.18) i=1 Now we show that for all t1 , t2 ∈ [0, 1], |I(x0n,1 (t2 )) − I(x0n,1 (t1 ))| ≤ max a(t) h(R1 + )(|xn,1 (t2 ) − xn,1 (t1 )| + |t2 − t1 |) t∈[0,1] Z (5.19) 1 (1−t2 ) xn,1 (t2 )+ n +| ω(s)dt| . 1 xn,1 (t1 )+ n (1−t1 ) From (5.18), 1 0 , x (t)) n n,1 1 1 ≤ a(t)[h(xn,1 (t) + ) + ω(xn,1 (t) + )]r(x0n,1 (t)), n n −x00n,1 (t) = a(t)f (t, xn,1 (t) + 1 0 , x (t)) n n,1 1 1 ≥ −a(t)[h(xn,1 (t) + ) + ω(xn,1 (t) + )]r(x0n,1 (t)), n n ∀t ∈ (0, 1), x00n,1 (t) = −a(t)f (t, xn,1 (t) + ∀t ∈ (0, 1), −x00n,1 (t)x0n,1 (t) r(x0n,1 (t)) 1 1 ) + ω(xn,1 (t) + )]x0n,1 (t) n n 1 ≥ a(t)[h(R1 + )x0n,1 (t) + ω(xn,1 (t) + (1 − t))x0n,1 (t)], n (5.20) ≥ a(t)[h(xn,1 (t) + ∀t ∈ (0, 1), x00n,1 (t)x0n,1 (t) r(x0n,1 (t)) 1 1 ) + ω(xn,1 (t) + )]x0n,1 (t) n n 1 ≤ −a(t)[h(R1 + )x0n,1 (t) + ω(xn,1 (t) + (1 − t))x0n,1 (t)], n (5.21) ≤ −a(t)[h(xn,1 (t) + ∀t ∈ (0, 1). 26 Y. MA, B. YAN EJDE-2008/147 Since the right-hand sides of (5.20) and (5.21) are positive, for all t1 , t2 ∈ [0, 1] and t1 < t 2 , I(x0n,2 (t2 )) − I(x0n,1 (t1 )) Z ≥ max a(t)[h(R1 + ) t∈[0,1] t2 x0n,1 (s)dt + t1 Z t2 ω(xn,1 (s) + t1 Z 1 0 )xn,1 (s)ds ] n 1 (1−t2 ) xn,1 (t2 )+ n ≥ − max a(t)[h(R1 + )(|xn,1 (t2 ) − xn,1 (t1 )|) + | ω(s)dt|], t∈[0,1] 1 xn,1 (t1 )+ n (1−t1 ) I(x0n,1 (t1 )) − I(x0n,1 (t2 )) Z 1 xn,1 (t2 )+ n (1−t2 ) ≤ max a(t)[h(R1 + )(|xn,1 (t2 ) − xn,1 (t1 )|) + | t∈[0,1] ω(s)dt|] , 1 xn,1 (t1 )+ n (1−t1 ) (5.19) holds. Since I −1 is uniformly continuous on [0, I(−R1 − )], for all ¯ > 0, there exists 0 > 0 such that |I −1 (s1 ) − I −1 (s2 )| < ¯, ∀|s1 − s2 | < 0 , s1 , s2 ∈ [0, I(−R1 − )]. (5.22) Also (5.19) guarantees that for 0 > 0, there exists δ 0 > 0 such that |I(x0n,1 (t2 )) − I(x0n,1 (t1 ))| < 0 , ∀ |t1 − t2 | < δ 0 , t1 , t2 ∈ [0, 1]. (5.23) From this inequality and (5.22), |x0n,1 (t2 ) − x0n,1 (t1 )| = |I −1 (I(x0n,1 (t2 ))) − I −1 (I(x0n,1 (t1 )))| < ¯, ∀|t1 − t2 | < δ 0 , t1 , t2 ∈ [0, 1]; (5.24) that is, {x0n,1 (t)} is equi-continuous on [0, 1]. (5.25) 0 From (5.15)–(5.17), (5.25) and the Arzela-Ascoli Theorem, {xn,1 (t)} and {xn,1 (t)} are relatively compact on C 1 [0, 1]. This implies, there exists a subsequence {xnj ,1 } of {xn,1 } and function x0,1 (t) ∈ C 1 [0, 1] such that lim max |xnj ,1 (t) − x0,1 (t)| = 0, j→+∞ t∈[0,1] Since x0nj ,1 (0) = 0, xnj ,1 (1) = (0, 1), j ∈ {1, 2, . . . }, x00,1 (0) = 0, x0,1 (1) = m−2 X Pm−2 i=1 lim max |x0nj ,1 (t) − x00,1 (t)| = 0. j→+∞ t∈[0,1] ai xnj ,1 (ξi ), x0nj ,1 (t) < 0, xnj ,1 (t) > 0, t ∈ ai x0,1 (ξi ), x00,1 (t) ≤ 0, x0,1 (t) ≥ 0, t ∈ (0, 1). (5.26) i=1 For (t, xnj ,1 (t) + n1j , x0nj ,1 (t)) ∈ [0, 1] × [0, R1 + ] × (−∞, 0), from (S5) there exists a function ΨR1 ∈ C([0, 1], R+ ) such that 1 f (t, xnj ,1 (t) + , x0nj ,1 (t))ds ≥ ΨR1 (t)(−x0nj ,1 (t))δ , 0 ≤ δ < 1. nj Then, for n ∈ N0 , x0nj ,1 (t) = − Z t a(s)f (s, xnj ,1 (s) + 0 Z ≤− 0 1 0 ,x (s))ds nj nj ,1 t a(s)ΨR1 + (s)(−x0nj ,1 (s))δ ds, t ∈ [0, 1], n ∈ N0 , EJDE-2008/147 MULTIPLE POSITIVE SOLUTIONS 27 which implies x0nj ,1 (t) Z t 1 ≤ −(1 − δ)( a(s)ΨR1 + (s)ds) 1−δ , 0 and Z xnj ,1 (t) = − t (t − s)a(s)f (s, xnj ,1 (s) + 0 + − 1− m−2 X 1 Pm−2 i=1 Z Z 1 1 ( (1 − s)a(s)f (s, xnj ,1 (s) + , x0nj ,1 (s))ds n j ai 0 ξi (ξi − s)a(s)f (s, xnj ,1 (s) + ai 0 i=1 1 Z ≥− (1 − s)a(s)f (s, xnj ,1 (s) + 0 + − 1− m−2 X 1 0 ,x (s))ds nj nj ,1 1 Pm−2 i=1 Z i=1 1 0 ,x (s))ds nj nj ,1 Z 1 1 ( (1 − s)a(s)f (s, xnj ,1 (s) + , x0nj ,1 (s))ds nj ai 0 ξi (ξi − s)a(s)f (s, xnj ,1 (s) + ai 1 0 ,x (s))ds) nj nj ,1 0 1 0 ,x (s))ds) nj nj ,1 Pm−2 Z 1 1 i=1 ai ≥ (1 − s)a(s)f (s, xnj ,1 (s) + , x0nj ,1 (s))ds Pm−2 nj 1 − i=1 ai 0 Pm−2 Z 1 1 i=1 ai − (ξi − s)a(s)f (s, xnj ,1 (s) + , x0nj ,1 (s))ds P m−2 n j 1 − i=1 ai 0 Pm−2 Z 1 ai (1 − ξi ) 1 = i=1 Pm−2 a(s)f (s, xnj ,1 (s) + , x0nj ,1 (s))ds n j 1 − i=1 ai 0 Pm−2 Z 1 ai (1 − ξi ) ≥ i=1 Pm−2 a(s)ΨR1 + (s)(−x0nj ,1 (s))δ ds 1 − i=1 ai 0 Pm−2 Z 1 Z s 1 i=1 ai (1 − ξi ) ≥ a(s)ΨR1 + (s)((1 − δ)( a(τ )ΨR1 + (τ )dτ ) 1−δ )δ ds Pm−2 1 − i=1 ai 0 0 =: F, t ∈ [0, 1]. Since 1 0 ,x (t)) nj nj ,1 1 1 ≤ [h(xnj ,1 (t) + ) + ω(xnj ,1 (t) + )]r(x0nj ,1 (t)) nj nj Z t 1 1−δ R1 ≤ [h( + ) + ω(F )]r((δ − 1) a(s)ΨR1 + (s)ds )), γ 0 f (t, xnj ,1 (t) + and 1 x0nj ,1 (t) − x0nj ,1 ( ) = − 2 Z t a(s)f (s, xnj ,1 (s) + 1/2 1 0 ,x (s))ds, nj nj ,1 t ∈ (0, 1), 28 Y. MA, B. YAN EJDE-2008/147 letting j → +∞, the Lebesgue Dominated Convergence Theorem guarantees that Z t 1 x00,1 (t) − x00,1 ( ) = − a(s)f (s, x0,1 (s), x00,1 (s))ds, t ∈ (0, 1). (5.27) 2 1/2 Differentiating, we have x000,1 (t) + a(t)f (t, x0,1 (t), x00,1 (t)) = 0, 0 < t < 1. From this equality and from (5.26), x0,1 (t) is a positive solution of (1.1) with x0,1 ∈ C 1 [0, 1] ∩ C 2 (0, 1). For the set {xn,2 }n∈N0 ⊆ (Ω2 −Ω1 )∩P , as in the proof for the set {xn,1 }n∈N0 , we obtain a convergent subsequence {xni ,2 }n∈N0 of {xn,2 }n∈N0 with limi→+∞ xni ,2 = x0,2 ∈ C 1 [0, 1] ∩ C 2 (0, 1). Moreover, x0,2 is a positive solution to (1.1). Example 5.1 In (1.1), let f (t, u, z) = µ[1 + (−z)a ][1 + ub + u−d ] and a(t) ≡ 1 with 0 ≤ a < 1, b > 1, 0 < d < 1 and µ > 0. If µ < sup c∈R+ P c(1− m−2 ai ) ) i i +1 i=1 I( Pm−2 i=1 a ξ maxt∈[0,1] a(t)(c + c1−d + c1+b 1+b ) , (5.28) Then (1.1) has at least two positive solutions x0,1 , x0,2 ∈ C 1 [0, 1] ∩ C 2 (0, 1). References [1] R. P. Agarwal, D. O’Regan; Existence theory for single and multiple solutions to singular positone boundary-value problem, Journal of Differential Equations, 175 (2001), pp. 393-414. [2] R. P. Agarwal, D. O’Gegan; Twin solutions to singular Dirichlet problems, Journal of Mathematical Analysis and Applications, 240 (1999), pp. 433-455. [3] K. Deimling; Nonlinear Functional Analysis, Spring-Verlag, Berlin , 1985. [4] C. P. Gupta, S. K. Ntouyas, P. Ch. Tsamatos; Solvability of an m-point boundary-value problem for second order ordinary differential equations, J. Math. Anal. Appl. 189 (1995) pp. 575-584. [5] J. Henderson, H. B. Thompson; Multiple symmetric positive solutions for a second order boundary-value problem, Proc. Amer. Math. soc., 128 (2000), pp. 2372-2379. [6] V. A. Il’in, E. I. Moiseev; Nonlocal boundary-value problem of the second kind for a SturmLiouvile operator, Differential Equation 23 (8) (1987), pp. 979-987. [7] V. A. Il’in, E. I. Moiseev; Nonlocal value problem of the first kind for a Sturm-Liouvile operator in its differential and finite difference aspects, Differential Equations 23 (7) (1987), pp. 803-810. [8] Dajing jiang; Multiple positive solutions to singular boundary-value problems for superlinesr second order ODEs, Acta Math. Sci. Ser., 22 (2002), pp. 199-206. [9] Daqing jiang, D. O’Gegan, R. P. Agarwal; Existence theory for single and multiple solutions to singular boundary-value problems for the one-dimension p-Laplacian, Adv. Math. Sci. Appl., 13 (2003), pp. 179-199. [10] Rahmat Ali Khan, J. R. L. Webb; Existence of at least three solutions of nonlinear three point boundary-value problems with super-quadratic growth, Journal of Mathematical Analysis and Applications, 328 (2007), pp. 690-698. [11] Eun Kyoung Lee, Yong-Hoon Lee; Multiple positive solutions of singular two point boundaryvalue problems for second order impulsive differential equations, Appl. Math. Comput., 158 (2004), pp. 745-759. [12] Yansheng Liu; Multiple positive solutions of singular boundary-value problem for the onedimension p-Laplacian, Indian J. Pure Appl. Math., 33 (2002), pp. 1541-1555. [13] B. Liu; Positive solutions of a nonlinear three-point boundary-value problem, Applied Mathematics and Computation 132 (2002), pp. 11-28. [14] Fuyi Li, Yaijing Zhang; Multiple symmetric nonnegative solutions of second-order ordinary differential equations, Applied Mathematics Letters, 17 (2004), pp. 261-267. EJDE-2008/147 MULTIPLE POSITIVE SOLUTIONS 29 [15] M. Meehan, D. O’Gegan; Multiple nonnegative solutions of nonlinear integral equations on compact and semi-infinite intervals, Applicable Analysis, 74 (2000), pp. 413-427. [16] R. Ma; Positive solutions for a nonlinear three-point boundary-value problem, Electron. J.Differential Equations 1999 (1999), no. 34, pp. 1-8. [17] R. Ma, Donal O’Regan; Solvability of singular second order m-point boundary-value problems Journal of Mathematical Analysis and Applications, 301 (2005), pp. 124-134. Department of Mathematics, Shandong Normal University, Jinan, 250014, China E-mail address, Ya Ma: maya-0907@163.com E-mail address, Baoqiang Yan: yanbqcn@yahoo.com.cn