Electronic Journal of Differential Equations, Vol. 2014 (2014), No. 61, pp. 1–14. ISSN: 1072-6691. URL: http://ejde.math.txstate.edu or http://ejde.math.unt.edu ftp ejde.math.txstate.edu SOLUTIONS TO THIRD-ORDER MULTI-POINT BOUNDARY-VALUE PROBLEMS AT RESONANCE WITH THREE DIMENSIONAL KERNELS SHUANG LI, JIAN YIN, ZENGJI DU Abstract. In this article, we consider the boundary-value problem x000 (t) = f (t, x(t), x0 (t), x00 (t)), x00 (0) = m X αi x00 (ξi ), x0 (0) = i=1 l X γk x0 (σk ), t ∈ (0, 1), x(1) = n X βj x(ηj ), j=1 k=1 where f : [0, 1] × R3 → R is a Carathéodory function, and the kernel to the linear operator has dimension three. Under some resonance conditions, by using the coincidence degree theorem, we show the existence of solutions. An example is given to illustrate our results. 1. Introduction This concerns the third-order nonlinear differential equation x000 (t) = f (t, x(t), x0 (t), x00 (t)), t ∈ (0, 1), (1.1) with the boundary conditions x00 (0) = m X i=1 αi x00 (ξi ), x0 (0) = l X γk x0 (σk ), x(1) = n X βj x(ηj ), (1.2) j=1 k=1 where f : [0, 1] × R3 → R is a Carathéodory function, 0 < ξ1 < · · · < ξm < 1, 0 < σ1 < · · · < σl < 1, 0 < η1 < · · · < ηn < 1, αi , γk , βj ∈ R (i = 1, . . . , m; k = 1, . . . , l; j = 1, . . . , n) and σ1 > {ξ1 , . . . , ξm }. The existence of solutions for multi-point boundary-value problems at resonance case has been extensively studied by many authors [1, 2, 3, 4, 5, 6, 7, 8, 10]. When the linear equation Lx = x000 = 0 with the boundary conditions (1.2) has a nontrivial solution, i.e. dim ker L ≥ 1. we say that boundary value problem (BVP for short) (1.1) and (1.2) is a resonance problem. The case of dim ker L = 1 has been discussed by many authors [1, 2, 4, 6]. For the case of dim ker L = 2, there are some results in [3, 7, 8, 10]., Lin, Du and Meng [7] 2000 Mathematics Subject Classification. 34B15. Key words and phrases. Multi-point boundary-value problem; coincidence degree theory; resonance. c 2014 Texas State University - San Marcos. Submitted October 13, 2013. Published March 5, 2014. 1 2 S. LI, J. YIN, Z. DU EJDE-2014/61 discussed the third-order multi-point boundary value-problem with dim ker L = 2, x000 (t) = f (t, x(t), x0 (t), x00 (t)), x00 (0) = m X αi x00 (ξi ), x0 (0) = 0, t ∈ (0, 1), x(1) = i=1 n X βj x(ηj ). (1.3) (1.4) j=1 Zhao, Liang and Ren [11] studied the nonlinear third-order boundary-value problems with dim ker L = 3, x000 (t) = f (t, x(t), x0 (t), x00 (t)) + e(t), x0 (0) = n X αj x0 (ξj ), x00 (0) = x00 (η), t ∈ (0, 1), x00 (1) = x00 (ζ). j=1 In [11], the results are obtained under m11 M = m21 m31 the assumption that m12 m13 m22 m23 6= 0. m32 m33 (1.5) This condition is difficult to verify and looks superfluous. Inspired by the above results, in this paper, we study (1.1)-(1.2) at resonance and show that the assumption Λ(p, q, r) (p + 1)(p + 2) Pm α ξ p i i Pi=1 = (q + 1)(q + 2) m α ξiq i i=1 Pm r (r + 1)(r + 2) i=1 αi ξi Pl (p + 2) k=1 γk σkp+1 Pl (q + 2) k=1 γk σkq+1 Pl (r + 2) k=1 γk σkr+1 Pn 2(1 − j=1 βj ηjp+2 ) Pn 2(1 − j=1 βj ηjq+2 ) 6= 0 Pn 2(1 − j=1 βj ηjr+2 ) can replace (1.5) from [5, 11], by using Lemma 3.1 below. If there exists σk (k = Pl 1, 2, . . . , l) satisfying x0 (0) = k=1 γk x0 (σk ) = 0, then BVP (1.3)-(1.4) is a special case of BVP (1.1)-(1.2). The remaining part of this article is organized as follows: In section 2, we will state some definitions and lemmas which would be useful in the proving of main results of this paper. In section 3, by applying Mawhin coincidence degree theory, we obtain some sufficient conditions which guarantee the existence of solution for BVP (1.1)-(1.2) at resonance case. In the last section, an example is given to illustrate our results. 2. Preliminaries Now, some notation and an abstract existence result [9] are introduced. Let Y, Z be real Banach spaces and let L : dom L ⊂ Y → Z be a linear operator which is a Fredholm map of index zero and P : Y → Y , Q : Z → Z be continuous projectors such that Im P = ker L, ker Q = Im L and Y = ker L ⊕ ker P , Z = Im L ⊕ Im Q. It follows that L|dom L∩ker P : dom L ∩ ker P → Im L is invertible, we denote the inverse of that map by KP . Let Ω be an open bounded subset of Y such that dom L ∩ Ω 6= ∅, the map N : Y → Z is said to be L-compact on Ω̄ if the map QN (Ω̄) is bounded and KP (I − Q)N : Ω̄ → Y is compact. Lemma 2.1 ([9, Theorem IV]). Let L be a Fredholm map of index zero and let N be L-compact on Ω̄. Assume that the following conditions are satisfied: (i) Lx 6= λN x for every (x, λ) ∈ ((dom L\ ker L) ∩ ∂Ω) × [0, 1]; EJDE-2014/61 SOLUTIONS TO THIRD-ORDER MULTI-POINT BVPS 3 (ii) N x ∈ / Im L for every x ∈ ker L ∩ ∂Ω; (iii) deg(QN |ker L , Ω ∩ ker L, 0) 6= 0, where Q : Z → Z is a continuous projector as above with Im L = ker Q. Then the abstract equation Lx = N x has at least one solution in dom L ∩ Ω̄. We use the classical spaces C[0, 1], C 1 [0, 1], C 2 [0, 1] and L1 [0, 1]. For x in C 2 [0, 1], we use the norms kxk∞ = maxt∈(0,1) {|x(t)|} and kxk = max{kxk∞ , kx0 k∞ , kx00 k∞ }. For L1 [0, 1] we denote the norm by k · k1 . We define the Sobolev space W 3,1 (0, 1) = x : [0, 1] → R : x, x0 , x00 are absolutely continuous on [0, 1] with x000 ∈ L1 [0, 1] . Let Y = C 2 [0, 1], Z = L1 [0, 1], and define the linear operator L : dom L ⊂ Y → Z as Lx = x000 , x ∈ dom L, where dom L = {x ∈ W 3,1 (0, 1) : x satisfies boundary conditions (1.2)}. We define N : Y → Z as N x = f (t, x(t), x0 (t), x00 (t)), t ∈ (0, 1), then (1.1)-(1.2) can be written as Lx = N x. The resonance conditions of (1.1)-(1.2) are as follows m X αi = 1, i=1 n X βj ηj = 1, j=1 l X γk = 1, j=1 βj ηj2 βj = 1, j=1 k=1 n X n X = 1, l X (2.1) γk σk = 0. k=1 Define the following symbols: Λ(p, q, r) (p + 1)(p + 2) Pm α ξ p (p + 2) Pl γ σ p+1 2(1 − Pn β η p+2 ) i=1 i i k=1 k k j=1 j j P Pl Pn q q+1 q+2 = (q + 1)(q + 2) m α ξ (q + 2) γ σ 2(1 − β η ) , i k j i j k Pi=1 Pk=1 Pj=1 (r + 1)(r + 2) m αi ξir (r + 2) l γk σ r+1 2(1 − n βj ηjr+2 ) i=1 k=1 j=1 k (q + 2) Pl γ σ q+1 2(1 − Pn β η q+2 ) k=1 k k j=1 j j P P M11 = , (r + 2) lk=1 γk σkr+1 2(1 − nj=1 βj ηjr+2 ) (q + 1)(q + 2) Pm α ξ q 2(1 − Pn β η q+2 ) j i i j Pi=1 Pnj=1 M12 = − r+2 , r (r + 1)(r + 2) m α ξ 2(1 − β η ) i j i j i=1 j=1 Pm Pl q q+1 (q + 1)(q + 2) i=1 αi ξi (q + 2) k=1 γk σk Pm P M13 = , (r + 1)(r + 2) i=1 αi ξir (r + 2) lk=1 γk σkr+1 (p + 2) Pl γ σ p+1 2(1 − Pn β η p+2 ) k=1 k k j=1 j j P P M21 = − , (r + 2) lk=1 γk σkr+1 2(1 − nj=1 βj ηjr+2 ) (p + 1)(p + 2) Pm α ξ p 2(1 − Pn β η p+2 ) j j Pi=1 i ir Pj=1 M22 = n r+2 , (r + 1)(r + 2) m α ξ 2(1 − β η ) i=1 i i j=1 j j 4 S. LI, J. YIN, Z. DU EJDE-2014/61 Pl Pm p+1 p (p + 1)(p + 2) i=1 αi ξi (p + 2) k=1 γk σk P Pl M23 = − , r (r + 1)(r + 2) m (r + 2) k=1 γk σkr+1 i=1 αi ξi (p + 2) Pl γ σ p+1 2(1 − Pn β η p+2 ) k=1 k k j=1 j j P P M31 = , (q + 2) lk=1 γk σkq+1 2(1 − nj=1 βj ηjq+2 ) (p + 1)(p + 2) Pm α ξ p 2(1 − Pn β η p+2 ) j i j i Pi=1 Pj=1 M32 = − n q q+2 , (q + 1)(q + 2) m 2(1 − j=1 βj ηj ) i=1 αi ξi Pm Pl p p+1 (p + 1)(p + 2) i=1 αi ξi (p + 2) k=1 γk σk Pl P M33 = q (q + 1)(q + 2) m (q + 2) k=1 γk σkq+1 i=1 αi ξi 3. Main results Lemma 3.1. Assume condition (2.1) holds, then there exist p ∈ {1, 2, . . . , n}, q ∈ Z + , q ≥ p + 1 and r ∈ Z + large enough number, such that Λ(p, q, r) 6= 0. Pm Proof. Clearly there exists p ∈ {1, 2, . . . , n} such that i=1 αi ξip+2 6= 0. Otherwise, Pm we have i=1 αi ξip+2 = 0 and p ∈ {1, 2, . . . , n}, then m X αi ξij (1 − ξi ) = 0, j = 1, 2, . . . , n − 1; i=1 i.e., ξ1 (1 − ξ1 ) ... .. .. . . ξ1n−1 (1 − ξ1 ) . . . Because ξ1 (1 − ξ1 ) ... .. .. . n−1 . ξ (1 − ξ ) . . . 1 1 ξm (1 − ξm ) α1 0 .. .. .. . = . . . n−1 αm 0 ξm (1 − ξm ) m Y Y ξi (1 − ξi ) (ξj − ξi ) 6= 0. = n−1 1≤i≤j≤m−1 (1 − ξm ) i=1 ξm Pm So we have αi = 0, i ∈ {1, 2, . . . , m}. Which is a contradiction to i=1 αi = 1 of condition (2.1). Pm Similarly, there exists q ∈ {1, 2, . . . , n + 1}, such that i=1 αi ξiq 6= 0. And for each s ∈ Z, s ≥ 0, there exists ks ∈ {sn + 1, . . . , (s + 1)n} such that ξm (1 − ξm ) .. . l X k=1 γk σkks +1 6= 0, m X αi ξiks 6= 0. i=1 Set Pm Pl l n X (q + 1) i=1 αi ξiq k=1 γk σkks +1 o S = ks ∈ Z : γk σkq+1 = , Pm (ks + 1) i=1 αi ξiks k=1 then S is a finite set. If else, there exists a monotone sequence {kst }, t = 1, 2, . . . , kst ≤ kst+1 , such that Pm Pl k +1 l X (q + 1) i=1 αi ξiq k=1 γk σk st q+1 γk σk = . Pm k (kst + 1) i=1 αi ξi st k=1 EJDE-2014/61 SOLUTIONS TO THIRD-ORDER MULTI-POINT BVPS 5 For σ1 > {ξ1 , . . . , ξm }, l X γk σkq+1 = lim kst →∞ k=1 Pl k +1 αi ξiq k=1 γk σk st = ∞, Pm k (kst + 1) i=1 αi ξi st (q + 1) Pm which is a contradiction. Then Pm p (p + 1)(p + 2) i=1 αi ξi Pm q (q + 1)(q + 2) i=1 αi ξi i=1 Pl (p + 2) k=1 γk σkp+1 Pl q+1 6= 0 (q + 2) k=1 γk σk Thus, we have lim Λ(p, q, r) (p + 1)(p + 2) Pm α ξ p (p + 2) Pl γ σ p+1 2(1 − Pn β η p+2 ) i k j i j k Pi=1 Plk=1 Pj=1 n q q+1 q+2 = (q + 1)(q + 2) m (q + 2) k=1 γk σk 2(1 − j=1 βj ηj ) i=1 αi ξi 0 0 2 Pm Pl p p+1 (p + 1)(p + 2) i=1 αi ξi (p + 2) k=1 γk σk Pm P = 2 6= 0. (q + 1)(q + 2) i=1 αi ξiq (q + 2) lk=1 γk σkq+1 r→∞ So there exist p ∈ {1, 2, . . . , n}, q ∈ Z + , q ≥ p + 1 and r ∈ Z + large enough number, such that Λ(p, q, r) 6= 0. Lemma 3.2. If condition (2.1) holds, then L : dom L ⊂ Y → Z is a Fredholm operator of index zero. Furthermore, the continuous projector operator Q : Z → Z can be defined by Q(y) = (T1 y(t))tp−1 + (T2 y(t))tq−1 + (T3 y(t))tr−1 , where p, q, r are given by Lemma 3.1 and p(p + 1)(p + 2) [M11 Q1 y + M12 Q2 y + M13 Q3 y], Λ(p, q, r) q(q + 1)(q + 2) [M21 Q1 y + M22 Q2 y + M23 Q3 y], T2 y = Λ(p, q, r) r(r + 1)(r + 2) T3 y = [M31 Q1 y + M32 Q2 y + M33 Q3 y], Λ(p, q, r) Z σk m Z ξi l X X Q1 y = y(s)ds, Q2 y = γk (σk − s)y(s)ds, T1 y = 0 i=1 Z Q3 y = 1 (1 − s)2 y(s)ds − 0 k=1 n X 0 Z βj j=1 ηj (ηj − s)2 y(s)ds. 0 The linear operator KP : Im L → dom L ∩ ker P can be written as Z 1 t KP y(t) = (t − s)2 y(s)ds, y ∈ Im L. 2 0 Furthermore kKP yk ≤ kyk1 , y ∈ Im L. Proof. It is clear that ker L = {x ∈ dom L : x = a + bt + ct2 , a, b, c ∈ R}. 6 S. LI, J. YIN, Z. DU EJDE-2014/61 Now we show that Im L = {y ∈ Z : Q1 y = Q2 y = Q3 y = 0}. (3.1) x000 = y (3.2) Since the problem has a solution x(t), such that boundary conditions (1.2) hold; i.e., x00 (0) = m X αi x00 (ξi ), x0 (0) = i=1 l X γk x0 (σk ), x(1) = n X βj x(ηj ), j=1 k=1 if and only if Q1 y = Q2 y = Q3 y = 0. (3.3) In fact, if (3.2) has a solution x(t) that satisfies the boundary conditions (1.2), then we have Z 1 t 1 (t − s)2 y(s)ds. x(t) = x(0) + x0 (0)t + x00 (0)t2 + 2 2 0 According to condition (2.1), we have Q1 y = Q2 y = Q3 y = 0. On the other hand, we let Z 1 t x(t) = a + bt + ct2 + (t − s)2 y(s)ds, 2 0 where a, b, c are arbitrary constants. If (3.3) holds, then x(t) is a solution of (3.2) and (1.2). Hence (3.1) holds. By Lemma 3.1, there exists p ∈ {1, 2, . . . , n}, q ∈ Z + , q ≥ p + 1 and r ∈ Z + is a large enough number, such that Λ(p, q, r) 6= 0. Set p(p + 1)(p + 2) [M11 Q1 y + M12 Q2 y + M13 Q3 y], Λ(p, q, r) q(q + 1)(q + 2) [M21 Q1 y + M22 Q2 y + M23 Q3 y], T2 y = Λ(p, q, r) r(r + 1)(r + 2) T3 y = [M31 Q1 y + M32 Q2 y + M33 Q3 y]. Λ(p, q, r) T1 y = And we define Q(y) = (T1 y(t))tp−1 + (T2 y(t))tq−1 + (T3 y(t))tr−1 , then dim Im Q = 3. So we have T1 ((T1 y)tp−1 ) p(p + 1)(p + 2) [M11 Q1 ((T1 y)tp−1 ) + M12 Q2 ((T1 y)tp−1 ) + M13 Q3 ((T1 y)tp−1 )] Λ(p, q, r) p(p + 1)(p + 2) = [M11 Q1 (tp−1 ) + M12 Q2 (tp−1 ) + M13 Q3 (tp−1 )](T1 y) Λ(p, q, r) = = m l X X 1 [M11 (p + 1)(p + 2) αi ξip + M12 (p + 2) γk σkp+1 Λ(p, q, r) i=1 k=1 + 2M13 (1 − n X j=1 = T1 y, βj ηjp+2 )](T1 y) EJDE-2014/61 SOLUTIONS TO THIRD-ORDER MULTI-POINT BVPS 7 T1 ((T2 y)tq−1 ) q(q + 1)(q + 2) [M11 Q1 ((T2 y)tq−1 ) + M12 Q2 ((T2 y)tq−1 ) + M13 Q3 ((T2 y)tq−1 )] Λ(p, q, r) q(q + 1)(q + 2) = [M11 Q1 (tq−1 ) + M12 Q2 (tq−1 ) + M13 Q3 (tq−1 )](T2 y) Λ(p, q, r) = = m l X X 1 [M11 (q + 1)(q + 2) αi ξiq + M12 (q + 2) γk σkq+1 Λ(p, q, r) i=1 k=1 + 2M13 (1 − n X βj ηjq+2 )](T2 y) = 0, j=1 T1 ((T3 y)tr−1 ) r(r + 1)(r + 2) [M11 Q1 ((T3 y)tr−1 ) + M12 Q2 ((T3 y)tr−1 ) + M13 Q3 ((T3 y)tr−1 )] Λ(p, q, r) r(r + 1)(r + 2) [M11 Q1 (tr−1 ) + M12 Q2 (tr−1 ) + M13 Q3 (tr−1 )](T3 y) = Λ(p, q, r) = = l m X X 1 γk σkr+1 αi ξir + M12 (r + 2) [M11 (r + 1)(r + 2) Λ(p, q, r) i=1 k=1 + 2M13 (1 − n X βj ηjr+2 )](T3 y) = 0. j=1 Similarly, we obtain T2 ((T1 y)tp−1 ) = 0, T2 ((T2 y)tq−1 ) = T2 y, T3 ((T1 y)tp−1 ) = 0, T3 ((T2 y)tq−1 ) = 0, T2 ((T3 y)tr−1 ) = 0, T3 ((T3 y)tr−1 ) = T3 y. So we have Q2 y = Q(T1 y(t))tp−1 + (T2 y(t))tq−1 + (T3 y(t))tr−1 ) = T1 ((T1 y(t))tp−1 + (T2 y(t))tq−1 + (T3 y(t))tr−1 )tp−1 + T2 ((T1 y(t))tp−1 + (T2 y(t))tq−1 + (T3 y(t))tr−1 )tq−1 + T3 ((T1 y(t))tp−1 + (T2 y(t))tq−1 + (T3 y(t))tr−1 )tr−1 = (T1 y(t))tp−1 + (T2 y(t))tq−1 + (T3 y(t))tr−1 = Qy. Thus, Q is a well defined operator. Now we need to show that ker Q = Im L. If y ∈ ker Q, then Qy = 0. By the definition of Qy, we have M11 Q1 y + M12 Q2 y + M13 Q3 y = 0, M21 Q1 y + M22 Q2 y + M23 Q3 y = 0, M31 Q1 y + M32 Q2 y + M33 Q3 y = 0. Because of M11 M21 M31 M12 M22 M32 M13 M23 = Λ2 (p, q, r) 6= 0, M33 8 S. LI, J. YIN, Z. DU EJDE-2014/61 we have Q1 y = Q2 y = Q3 y = 0, i.e. y ∈ ImL. If y ∈ Im L, we have Q1 y = Q2 y = Q3 y = 0. From the definition of Qy, it is obvious that Qy = 0, thus y ∈ KerQ. Hence, ker Q = Im L. For y ∈ Z, let y = (y − Qy) + Qy, since Q(y − Qy) = Qy − Q2 y = 0, we know y − Qy ∈ ker Q = Im L, and we have Qy ∈ Im Q. Thus Z = Im L + Im Q. And for any y ∈ Im L ∩ Im Q, from y ∈ Im Q, there exists constants a, b, c ∈ R, such that y(t) = atp−1 + btq−1 + ctr−1 . From y ∈ Im L, we obtain m m m bX cX aX αi ξip + αi ξiq + αi ξir = 0, p i=1 q i=1 r i=1 l l l k=1 k=1 k=1 X X X b c a γk σkp+1 + γk σkq+1 + γk σkr+1 = 0, p(p + 1) q(q + 1) r(r + 1) n n X X 1 2 1 2 1 1 βj ηjp+2 ) + b( − βj ηjq+2 ) + )(1 − + )(1 − a( − p p+1 p+2 q q + 1 q + 2 j=1 j=1 n X 1 2 1 + c( − + )(1 − βj ηjr+2 ) = 0. r r+1 r+2 j=1 (3.4) In view of Pm p 1 i=1 αi ξi p 1 Pl p(p+1) k=1 γk σkp+1 A31 Pm Pm q 1 r i=1 αi ξi i=1 αi ξi , r P P l l q+1 r+1 1 1 γ σ γ σ k k k=1 k=1 k k q(q+1) r(r+1) A32 A33 1 = Λ(p, q, r) 6= 0, p(p + 1)(p + 2)q(q + 1)(q + 2)r(r + 1)(r + 2) 1 q where the entries of the third row are n X 2 1 1 βj ηjp+2 ) + )(1 − A31 = ( − p p+1 p+2 j=1 n X 1 2 1 A32 = ( − + )(1 − βj ηjq+2 ) q q+1 q+2 j=1 A33 I split the above matrix, please check it n X 1 2 1 =( − + )(1 − βj ηjr+2 ) . r r+1 r+2 j=1 Therefore (3.4) has an unique solution a = b = c = 0, which implies Im Q ∩ Im L = {0} and Z = Im L ⊕ Im Q. Since dim ker L = dimImQ = codimImL = 3, thus L is a Fredholm map of index zero. Let P : Y → Y be defined by 1 P x(t) = x(0) + x0 (0)t + x00 (0)t2 , 2 t ∈ (0, 1). then, the generalized inverse operator KP : Im L → dom L ∩ ker P be defined by Z 1 t (t − s)2 y(s)ds, y ∈ Im L. KP (y(t)) = 2 0 EJDE-2014/61 SOLUTIONS TO THIRD-ORDER MULTI-POINT BVPS 9 If y ∈ Im L, we have (LKP )(y(t)) = (KP y)000 = y(t). If x(t) ∈ dom L ∩ ker P , we have (KP L)x(t) = (KP )(x000 (t)) Z 1 t (t − s)2 x000 (s)ds = 2 0 1 = x(t) − [x(0) + x0 (0)t + x00 (0)t2 ] 2 = x(t) − P x(t). Because x ∈ dom L ∩ ker P , we know that P x(t) = 0. Thus (KP L)(x(t)) = x(t). It is obviously that kKP yk ≤ kyk1 . For the next theorem we use the assumptions: (H1) There exists α(t), β(t), γ(t), θ(t) ∈ L1 [0, 1], such that for all (x1 , x2 , x3 ) in R3 , t ∈ (0, 1), |f (t, x1 , x2 , x3 )| ≤ α(t)|x1 | + β(t)|x2 | + γ(t)|x3 | + θ(t). (H2) There exists a constant A > 0 such that for x(t) ∈ dom L, if |x(t)| > A or |x0 (t)| > A or |x00 (t)| > A for all t ∈ (0, 1), then Q1 N (x(t)) 6= 0 or Q2 N (x(t)) 6= 0 or Q3 N (x(t)) 6= 0. (H3) There exists a constant B > 0 such that for a, b, c ∈ R, if |a| > B, |b| > B, |c| > B, then either Q1 N (a + bt + ct2 ) + Q2 N (a + bt + ct2 ) + Q3 N (a + bt + ct2 ) > 0, (3.5) or Q1 N (a + bt + ct2 ) + Q2 N (a + bt + ct2 ) + Q3 N (a + bt + ct2 ) < 0. (3.6) Theorem 3.3. Let the conditions (2.1), (H1), (H2), (H3) hold. Then BVP (1.1)(1.2) have at least one solution in C 2 [0, 1], provided that kαk1 + kβk1 + kγk1 < 1. Proof. We divide the proof into four steps. Step 1: Let Ω1 = {x ∈ dom L\ ker L : Lx = λN x, λ ∈ [0, 1]}. Then Ω1 is bounded. Suppose that x ∈ Ω1 , we have Lx = λN x. Thus λ 6= 0, N x ∈ Im L = ker Q, hence Q1 (N x(t)) = Q2 (N x(t)) = Q3 (N x(t)) = 0. From (H2), there exists t1 , t2 , t3 ∈ (0, 1), such that |x(t1 )| ≤ A, |x0 (t2 )| ≤ A, |x00 (t3 )| ≤ A. Since x, x0 , x00 are absolutely continuous for all t ∈ (0, 1), and Z t Z t 0 0 00 00 00 x (t) = x (t2 ) + x (s)ds, x (t) = x (t3 ) + x000 (s)ds. t2 t3 Thus kx00 k∞ ≤ A + kx000 k1 , kx0 k∞ ≤ 2A + kx000 k1 . And kxk∞ ≤ k(I − P )xk∞ + kP xk∞ = kP xk∞ + kKP L(I − P )xk∞ = kP xk∞ + kKP Lxk∞ ≤ kP xk∞ + kKP kLxk1 ≤ kP xk∞ + kLxk1 = kP xk∞ + kx000 k1 . 10 S. LI, J. YIN, Z. DU EJDE-2014/61 From (H1 ), we obtain kx000 k1 = kLxk1 ≤ kN xk1 ≤ kαk1 kxk∞ + kβk1 kx0 k∞ + kγk1 kx00 k∞ + kθk1 ≤ (kαk1 + kβk1 + kγk1 )kx000 k1 + (2kβk1 + kγk1 )A + kαk1 kP xk∞ + kθk1 . Then kx000 k1 ≤ 1 [(2kβk1 + kγk1 )A + kαk1 kP xk∞ + kθk1 ]. 1 − (kαk1 + kβk1 + kγk1 ) So there exists a constant M1 > 0 such that kxk ≤ M1 . Hence we show that Ω1 is bounded. Step 2: Let Ω2 = {x ∈ ker L : N x ∈ Im L}. Then Ω2 is bounded. Since x ∈ Ω2 , x ∈ ker L = {x ∈ dom L : x = a + bt + ct2 , a, b, c ∈ R}, and QN x = 0, thus, Q1 (N (a + bt + ct2 )) = Q2 (N (a + bt + ct2 )) = Q3 (N (a + bt + ct2 )) = 0. From (H3), kxk ≤ |a| + |b| + |c| ≤ 3B. So Ω2 is bounded. Step 3: Let Ω3 = {x ∈ ker L : λJx + (1 − λ)QN x = 0, λ ∈ [0, 1]}. Here J : ker L → Im Q is the linear isomorphism given by 1 (a1 tp−1 + b1 tq−1 + c1 tr−1 ), J(a + bt + ct2 ) = Λ(p, q, r) where a, b, c ∈ R a1 = p(p + 1)(p + 2)(M11 |a| + M12 |b| + M13 |c|), b1 = q(q + 1)(q + 2)(M21 |a| + M22 |b| + M23 |c|), c1 = r(r + 1)(r + 2)(M31 |a| + M32 |b| + M33 |c|). Then Ω3 is bounded. Set p(p + 1)(p + 2) B1 = , Λ(p, q, r) B2 = q(q + 1)(q + 2) , Λ(p, q, r) X1 = λ|a| + (1 − λ)Q1 N (a + bt + ct2 ), B3 = r(r + 1)(r + 2) , Λ(p, q, r) X2 = λ|b| + (1 − λ)Q2 N (a + bt + ct2 ), X3 = λ|c| + (1 − λ)Q3 N (a + bt + ct2 ). Since x(t) = a + bt + ct2 ∈ Ω3 , then we have λJx + (1 − λ)QN x = 0; i.e., B1 M11 X1 + B1 M12 X2 + B1 M13 X3 = 0, B2 M21 X1 + B2 M22 X2 + B2 M23 X3 = 0, B3 M31 X1 + B3 M32 X2 + B3 M33 X3 = 0. Because B1 M11 B2 M21 B3 M31 B1 M12 B2 M22 B3 M32 M11 B1 M13 B2 M23 = (B1 B2 B3 ) M21 M31 B3 M33 = M12 M22 M32 M13 M23 M33 p(p + 1)(p + 2)q(q + 1)(q + 2)r(r + 1)(r + 2) 6= 0. Λ(p, q, r) EJDE-2014/61 SOLUTIONS TO THIRD-ORDER MULTI-POINT BVPS 11 Then X1 = X2 = X3 = 0. If λ = 1, then |a| = |b| = |c| = 0. If λ 6= 1 and |a| > B or |b| > B or |c| > B and (3.5) hold, then λ(|a| + |b| + |c|) = −(1 − λ) Q1 (N (a + bt + ct2 )) + Q2 (N (a + bt + ct2 )) + Q3 (N (a + bt + ct2 )) < 0, which contradicts λ(|a| + |b| + |c|) > 0. Thus kxk ≤ |a| + |b| + |c| ≤ 3B. So Ω3 is bounded. If (3.6) holds, we let Ω3 = {x ∈ ker L : −λJx + (1 − λ)QN x = 0, λ ∈ [0, 1]}. Similarly, we can proof that Ω3 is bounded. Step 4: In the following, we shall prove that all the conditions of Lemma 2.1 are satisfied. Let Ω ⊂ Y be a bounded open set such that ∪3i=1 Ωi ⊂ Ω. By the AscoliArzela theorem, we can show that KP (I − Q)N : Ω → Y is compact, thus N is L−compact on Ω. Then by the above argument, we obtain (i) Lx 6= λN x for each (x, λ) ∈ ((dom L\ ker L) ∩ ∂Ω) × [0, 1]; (ii) N x ∈ / Im L for each x ∈ ker L ∩ ∂Ω. At last we prove that (iii) of the Lemma 2.1 is satisfied. Let H(x, λ) = ±λJx + (1 − λ)QN x. According to the above argument, we have H(x, λ) 6= 0, for x ∈ ker L ∩ ∂Ω. Thus, by the homotopy property of degree, we get deg(QN |ker L , Ω ∩ ker L, 0) = deg(H(·, 0), Ω ∩ ker L, 0) = deg(H(·, 1), Ω ∩ ker L, 0) = deg(±J, Ω ∩ ker L, 0) 6= 0. Then by Lemma 2.1, Lx = N x has at least one solution in dom L ∩ Ω, so that BVP (1.1)-(1.2) has at least one solution in C 2 [0, 1]. 4. Applications We consider the boundary value problem x000 (t) = x00 (t) + sin x(t)(1 − cos x0 (t)), t ∈ (0, 1), 1 1 1 1 x00 (0) = 2x00 ( ) − x00 ( ), x0 (0) = 3x0 ( ) − 2x0 ( ), 5 4 3 2 9 1 1 16 3 x(1) = x( ) − 4x( ) + x( ). 5 3 2 5 4 (4.1) (4.2) (4.3) So f (t, x(t), x0 (t), x00 (t)) = x00 (t) + sin x(t)(1 − cos x0 (t)), α1 = 2, α2 = −1, ξ1 = 51 , 1 ξ2 = 14 , γ1 = 3, γ2 = −2, σ1 = 13 , σ2 = 12 , β1 = 95 , β2 = −4, β3 = 16 5 , η1 = 3 , 1 3 η2 = 2 , η3 = 4 . Then we have α1 − α2 = 1, γ1 + γ2 = 1, β1 + β2 + β3 = 1, β1 η1 + β2 η2 + β3 η3 = 1, β1 η12 + β2 η22 + β3 η32 = 1, γ1 σ1 + γ2 σ2 = 0. So the condition (2.1) holds. By calculations, we obtain Z 1/5 Z 13 Z 1/2 1 1 Q1 y = 2 y(s)ds, Q2 y = 3 ( − s)y(s)ds − 2 ( − s)y(s)ds, 3 2 0 0 0 12 S. LI, J. YIN, Z. DU EJDE-2014/61 Z 1 9 3 1 (1 − s) y(s)ds − Q3 y = ( − s)2 y(s)ds 5 0 3 0 Z 1/2 Z 1 16 3/4 3 ( − s)2 y(s)ds − ( − s)2 y(s)ds, +4 2 5 0 4 0 1225231 Λ(1, 2, 3) = , 1834400 5731015 45787 3429 M11 = − , M12 = , M13 = − , 1508544 311040 38880 70758 3089 8649 M21 = , M22 = , M23 = , 279935 9600 2260 1269 79 477 , M32 = − , M33 = − . M31 = − 3888 3600 200 6 [M11 Q1 y + M12 Q2 y + M13 Q3 y], T1 y = Λ(p, q, r) 24 T2 y = [M21 Q1 y + M22 Q2 y + M23 Q3 y], Λ(p, q, r) 60 [M31 Q1 y + M32 Q2 y + M33 Q3 y]. T3 y = Λ(p, q, r) Z 1 2 Define Qy by Qy = T1 y + (T2 y)t + (T3 y)t2 , and we take KP y(t) as in Lemma 3.2, then Lemma 3.2 holds. On the other hand, we have |f (t, x(t), x0 (t), x00 (t))| ≤ |x00 (t)| + 2, t ∈ (0, 1). And let α(t) = 0, β(t) = 0, γ(t) = 1, θ(t) = 2, then the condition (H1) of Theorem 3.3 is satisfied. If x00 (t) > 8 = A, t ∈ (0, 1), then Z 1/5 Q1 y = 2 (x00 (t) + sin x(t)(1 − cos x0 (t)))dt 0 Z 1/4 (x00 (t) + sin x(t)(1 − cos x0 (t)))dt − Z 0 1/5 Z 1/4 7dt − >2 0 10dt > 0 9 , 10 If x00 (t) < −8 = −A, t ∈ (0, 1), then Z 1/5 Q1 y = 2 (x00 (t) + sin x(t)(1 − cos x0 (t)))dt 0 Z 1/4 − Z (x00 (t) + sin x(t)(1 − cos x0 (t)))dt 0 1/5 Z (−6)dt − <2 0 1/4 (−9)dt < − 0 3 , 20 So condition (H2) is satisfied. If |a| > 16 = B, |b| > 16 = B, |c| > 16 = B, then Q1 N (a + bt + ct2 ) + Q2 N (a + bt + ct2 ) + Q3 N (a + bt + ct2 ) Z 1/5 =2 (|2c| + sin(a + bt + ct2 )[1 − cos(b + 2ct)])dt 0 EJDE-2014/61 SOLUTIONS TO THIRD-ORDER MULTI-POINT BVPS 13 1/4 Z (|2c| + sin(a + bt + ct2 )[1 − cos(b + 2ct)])dt − 0 1 3 Z +3 0 1 ( − t)(|2c| + sin(a + bt + ct2 )[1 − cos(b + 2ct)])dt 3 1/2 Z −2 0 1 Z 1 ( − t)(|2c| + sin(a + bt + ct2 )[1 − cos(b + 2ct)])dt 2 (1 − t)2 (|2c| + sin(a + bt + ct2 )[1 − cos(b + 2ct)])dt + 0 − 9 5 1 3 Z Z 0 1/2 +4 0 16 − 5 1 ( − t)2 (|2c| + sin(a + bt + ct2 )[1 − cos(b + 2ct)])dt 3 Z 0 1 ( − t)2 (|2c| + sin(a + bt + ct2 )[1 − cos(b + 2ct)])dt 2 3/4 3 ( − t)2 (|2c| + sin(a + bt + ct2 )[1 − cos(b + 2ct)])dt > 0. 4 So condition (H3) is satisfied. Thus all the conditions of Theorem 3.3 are satisfied. Hence BVP (4.1)-(4.3) has at least one solution in C 2 [0, 1]. Acknowledgments. This research was sponsored by the Natural Science Foundation of China (11071205, 11101349) and by PAPD of Jiangsu Higher Education Institutions, the Humanistic and Social Scientific Research Planning Program in Ministry of Education of China (No. 12YJA630063), the Social Science Foundation of Jiangsu Province of China (No. 10GLB001). The authors express their sincere thanks to the anonymous reviewer for his or her valuable comments and suggestions for improving the quality of this article. References [1] Z. Du; Solvability of functional differential equations with multi-point boundary value problems at resonance, Comput. Math. Appl. 55 (2008) 2653-2661. [2] Z. Du, X. Lin, W. Ge; On a third-order multi-point boundary value problem at resonance, J. Math. Anal. Appl. 302 (2005) 217-229. [3] Z. Du, F. Meng; Solutions to a second-order multi-point boundary value problem at resonance, Acta Math. Sci. 30 (2010) 1567-1576. [4] W. Ge; The boundary value problems of nonlinear ordinary differential equations, Science Press, Beijing, (2007) (in Chinese). [5] J. Liang, L. Ren, Z. Zhao; Existence of Solutions for a Multi-point Boundary Value Problems with Three Dimension Kernal at Resonance, Chin. Quart. J. Math., 26(1) (2011) 138-143. [6] X. Lin, W. Liu, A nonlinear third-order multi-point boundary value problem in the resonance case, J. Appl. Math. Comput. 29 (2009) 35-51. [7] X. Lin, Z. Du, F. Meng; A note on a third-order multi-point boundary value problem at resonance, Math. Nachr. 284 (13) (2011) 1690-1700. [8] B. Liu, Z. Zhao; A note on multi-point boundary value problems, Nonl. Anal. 67 (2007) 2680-2689. [9] J. Mawhin; Topological degree methods in nonlinear boundary value problems, NSFCBMS Regional Conference Series in Mathematics, American Mathematical Society, Providence, RI, (1979). [10] C. Xue, Z. Du, W. Ge; Solutions for m-point boundary value problems of third-order ordinary differential equations at resonance, J. Appl. Math. Comput. 17 (2004) 229-244. [11] Z. Zhao, R. Liang, L. Ren; A Multi-point Boundary Value Problems With Three Dimension Kernel at Resonance, Adv. Math. (China) 38 (3) (2009) 345-358. 14 S. LI, J. YIN, Z. DU EJDE-2014/61 Shuang Li School of Management, China University of Mining and Technology, Xuzhou, Jiangsu 221116, China E-mail address: lishuangchina@163.com Jian Yin School of Mathematics and Statistics, Jiangsu Normal University, Xuzhou, Jiangsu 221116, China E-mail address: yinjian719@163.com Zengji Du (corresponding author) School of Mathematics and Statistics, Jiangsu Normal University, Xuzhou, Jiangsu 221116, China E-mail address: duzengji@163.com, Fax 0086-516-83403299