Electronic Journal of Differential Equations, Vol. 2013 (2013), No. 97, pp. 1–13. ISSN: 1072-6691. URL: http://ejde.math.txstate.edu or http://ejde.math.unt.edu ftp ejde.math.txstate.edu EXISTENCE OF INFINITELY MANY ANTI-PERIODIC SOLUTIONS FOR SECOND-ORDER IMPULSIVE DIFFERENTIAL INCLUSIONS SHAPOUR HEIDARKHANI, GHASEM A. AFROUZI, ARMIN HADJIAN, JOHNNY HENDERSON Abstract. In this article, we establish the existence of infinitely many antiperiodic solutions for a second-order impulsive differential inclusion with a perturbed nonlinearity and two parameters. The technical approach is mainly based on a critical point theorem for non-smooth functionals. 1. Introduction The aim of this article is to show the existence of infinitely many solutions for the following two parameter second-order impulsive differential inclusion subject to anti-periodic boundary conditions −(φp (u0 (x)))0 + M φp (u(x)) ∈ λF (u(x)) + µG(x, u(x)) 0 −∆φp (u (xk )) = Ik (u(xk )), u(0) = −u(T ), in [0, T ] \ Q, k = 1, 2, . . . , m, (1.1) 0 u (0) = −u0 (T ), where Q = {x1 , x2 , . . . , xm }, p > 1, T > 0, M ≥ 0, φp (x) := |x|p−2 x, 0 = x0 < 0 − 0 + x1 < · · · < xm < xm+1 = T , ∆φp (u0 (xk )) := φp (u0 (x+ k )) − φp (u (xk )), with u (xk ) − and u0 (xk ) denoting the right and left limits, respectively, of u0 (x) at x = xk , Ik ∈ C(R, R), k = 1, 2, . . . , m, λ is a positive parameter, µ is a nonnegative parameter, and F is a multifunction defined on R, satisfying (F1) F : R → 2R is upper semicontinuous with compact convex values; (F2) min F, max F : R → R are Borel measurable; (F3) |ξ| ≤ a(1 + |s|r−1 ) for all s ∈ R, ξ ∈ F (s), r > 1 (a > 0). Also, G is a multifunction defined on [0, T ] × R, satisfying (G1) G(x, ·) : R → 2R is upper semicontinuous with compact convex values for a.e. x ∈ [0, T ] \ Q; (G2) min G, max G : ([0, T ] \ Q) × R → R are Borel measurable; (G3) |ξ| ≤ a(1 + |s|r−1 ) for a.e. x ∈ [0, T ], s ∈ R, ξ ∈ G(x, s), r > 1 (a > 0). 2000 Mathematics Subject Classification. 58E05, 49J52, 34A60. Key words and phrases. Differential inclusions; impulsive; anti-periodic solution; non-smooth critical point theory. c 2013 Texas State University - San Marcos. Submitted January 29, 2013. Published April 16, 2013. 1 2 S. HEIDARKHANI, G. A. AFROUZI, A. HADJIAN, J. HENDERSON EJDE-2013/97 Impulsive differential equations are used to describe various models of real-world processes that are subject to a sudden change. These models are studied in physics, population dynamics, ecology, industrial robotics, biotechnology, economics, optimal control, and so forth. Associated with this development, a theory of impulsive differential equations has been given extensive attention. Differential inclusions arise in models for control systems, mechanical systems, economical systems, game theory, and biological systems to name a few. It is very important to study antiperiodic boundary value problems because they can be applied to interpolation problems [5], antiperiodic wavelets [3], the Hill differential operator [6], and so on. It is natural from both a physical standpoint as well as a theoretical view to give considerable attention to a synthesis involving problems for impulsive differential inclusion with anti-periodic boundary conditions. Recently, multiplicity of solutions for differential inclusions via non-smooth variational methods and critical point theory has been considered and here we cite the papers [9, 10, 11, 12, 16]. For instance, in [11], the author, employing a non-smooth Ricceri-type variational principle [15], developed by Marano and Motreanu [13], has established the existence of infinitely many, radially symmetric solutions for a differential inclusion problem in RN . Also, in [12], the authors extended a recent result of Ricceri concerning the existence of three critical points of certain non-smooth functionals. Two applications have been given, both in the theory of differential inclusions; the first one concerns a non-homogeneous Neumann boundary value problem, the second one treats a quasilinear elliptic inclusion problem in the whole RN . In [9], the author, under convenient assumptions, has investigated the existence of at least three positive solutions for a differential inclusion involving the p-Laplacian operator on a bounded domain, with homogeneous Dirichlet boundary conditions and a perturbed nonlinearity depending on two positive parameters; his result also ensured an estimate on the norms of the solutions independent of both the perturbation and the parameters. Very recently, Tian and Henderson in [16], based on a non-smooth version of critical point theory of Ricceri due to Iannizzotto [9], have established the existence of at least three solutions for the problem (1.1) whenever λ is large enough and µ is small enough. In the present paper, motivated by [16], employing an abstract critical point result (see Theorem 2.6 below), we are interested in ensuring the existence of infinitely many anti-periodic solutions for the problem (1.1); see Theorem 3.1 below. We refer to [2], in which related variational methods are used for non-homogeneous problems. To the best of our knowledge, no investigation has been devoted to establishing the existence of infinitely many solutions to such a problem as (1.1). For a couple of references on impulsive differential inclusions, we refer to [7] and [8]. A special case of our main result is the following theorem. Theorem 1.1. Assume that (F1)–(F3) hold, and Ii (0) = 0, Ii (s)s < 0, s ∈ R, i = 1, 2, . . . , m. Furthermore, suppose that Rt sup|t|≤ξ min 0 F (s)ds = 0, lim inf ξ→+∞ ξp RT R ξ( T −x) min 0 2 F (s) ds dx 0 P lim sup = +∞. m R ξ( T2 −xi ) 1 2M T p T + p+1 − ξ→+∞ Ii (s)ds i=1 0 pξ p+1 ( 2 ) EJDE-2013/97 EXISTENCE OF INFINITELY MANY ANTI-PERIODIC SOLUTIONS 3 Then, the problem (1.1), for λ = 1 and µ = 0, admits a sequence of pairwise distinct solutions. 2. Basic definitions and preliminary results Let (X, k · kX ) be a real Banach space. We denote by X ∗ the dual space of X, while h·, ·i stands for the duality pairing between X ∗ and X. A function ϕ : X → R is called locally Lipschitz if, for all u ∈ X, there exist a neighborhood U of u and a real number L > 0 such that |ϕ(v) − ϕ(w)| ≤ Lkv − wkX for all v, w ∈ U. If ϕ is locally Lipschitz and u ∈ X, the generalized directional derivative of ϕ at u along the direction v ∈ X is ϕ◦ (u; v) := lim sup w→u, τ →0+ ϕ(w + τ v) − ϕ(w) . τ The generalized gradient of ϕ at u is the set ∂ϕ(u) := {u∗ ∈ X ∗ : hu∗ , vi ≤ ϕ◦ (u; v) for all v ∈ X}. ∗ So ∂ϕ : X → 2X is a multifunction. We say that ϕ has compact gradient if ∂ϕ maps bounded subsets of X into relatively compact subsets of X ∗ . Lemma 2.1 ([14, Proposition 1.1]). Let ϕ ∈ C 1 (X) be a functional. Then ϕ is locally Lipschitz and ϕ◦ (u; v) = hϕ0 (u), vi 0 ∂ϕ(u) = {ϕ (u)} for all u, v ∈ X; for all u ∈ X. Lemma 2.2 ([14, Proposition 1.3]). Let ϕ : X → R be a locally Lipschitz functional. Then ϕ◦ (u; ·) is subadditive and positively homogeneous for all u ∈ X, and ϕ◦ (u; v) ≤ Lkvk for all u, v ∈ X, with L > 0 being a Lipschitz constant for ϕ around u. Lemma 2.3 ([4]). Let ϕ : X → R be a locally Lipschitz functional. Then ϕ◦ : X × X → R is upper semicontinuous and for all λ ≥ 0, u, v ∈ X, (λϕ)◦ (u; v) = λϕ◦ (u; v). Moreover, if ϕ, ψ : X → R are locally Lipschitz functionals, then (ϕ + ψ)◦ (u; v) ≤ ϕ◦ (u; v) + ψ ◦ (u; v) for all u, v ∈ X. Lemma 2.4 ([14, Proposition 1.6]). Let ϕ, ψ : X → R be locally Lipschitz functionals. Then ∂(λϕ)(u) = λ∂ϕ(u) for all u ∈ X, λ ∈ R, ∂(ϕ + ψ)(u) ⊆ ∂ϕ(u) + ∂ψ(u) for all u ∈ X. Lemma 2.5 ([9, Proposition 1.6]). Let ϕ : X → R be a locally Lipschitz functional with a compact gradient. Then ϕ is sequentially weakly continuous. 4 S. HEIDARKHANI, G. A. AFROUZI, A. HADJIAN, J. HENDERSON EJDE-2013/97 We say that u ∈ X is a (generalized) critical point of a locally Lipschitz functional ϕ if 0 ∈ ∂ϕ(u); i.e., ϕ◦ (u; v) ≥ 0 for all v ∈ X. When a non-smooth functional, g : X → (−∞, +∞), is expressed as a sum of a locally Lipschitz function, ϕ : X → R, and a convex, proper, and lower semicontinuous function, j : X → (−∞, +∞); that is, g := ϕ + j, a (generalized) critical point of g is every u ∈ X such that ϕ◦ (u; v − u) + j(v) − j(u) ≥ 0 for all v ∈ X (see [14, Chapter 3]). Hereafter, we assume that X is a reflexive real Banach space, N : X → R is a sequentially weakly lower semicontinuous functional, Υ : X → R is a sequentially weakly upper semicontinuous functional, λ is a positive parameter, j : X → (−∞, +∞) is a convex, proper, and lower semicontinuous functional, and D(j) is the effective domain of j. Write M := Υ − j, Iλ := N − λM = (N − λΥ) + λj. We also assume that N is coercive and D(j) ∩ N −1 ((−∞, r)) 6= ∅ (2.1) for all r > inf X N . Moreover, owing to (2.1) and provided r > inf X N , we can define supv∈N −1 ((−∞,r)) M(v) − M(u) , ϕ(r) := inf r − N (u) u∈N −1 ((−∞,r)) γ := lim inf ϕ(r), δ := lim inf + ϕ(r). r→+∞ r→(inf X N ) If N and Υ are locally Lipschitz functionals, in [1, Theorem 2.1] the following result is proved; it is a more precise version of [13, Theorem 1.1] (see also [15]). Theorem 2.6. Under the above assumption on X, N and M, one has (a) For every r > inf X N and every λ ∈ (0, 1/ϕ(r)), the restriction of the functional Iλ = N −λM to N −1 ((−∞, r)) admits a global minimum, which is a critical point (local minimum) of Iλ in X. (b) If γ < +∞, then for each λ ∈ (0, 1/γ), the following alternative holds: either (b1) Iλ possesses a global minimum, or (b2) there is a sequence {un } of critical points (local minima) of Iλ such that limn→+∞ N (un ) = +∞. (c) If δ < +∞, then for each λ ∈ (0, 1/δ), the following alternative holds: either (c1) there is a global minimum of N which is a local minimum of Iλ , or (c2) there is a sequence {un } of pairwise distinct critical points (local minima) of Iλ , with limn→+∞ N (un ) = inf X N , which converges weakly to a global minimum of N . Now we recall some basic definitions and notation. On the reflexive Banach space X := {u ∈ W 1,p ([0, T ]) : u(0) = −u(T )} we consider the norm Z T 1/p kukX := |u0 (x)|p + M |u(x)|p dx 0 EJDE-2013/97 EXISTENCE OF INFINITELY MANY ANTI-PERIODIC SOLUTIONS 5 for all u ∈ X, which is equivalent to the usual norm (note that M ≥ 0). We recall that X is compactly embedded into the space C 0 ([0, T ]) endowed with the maximum norm k · kC 0 . Lemma 2.7 ([16, Lemma 3.3]). Let u ∈ X. Then 1 kukC 0 ≤ T 1/q kukX , 2 where 1/p + 1/q = 1. (2.2) Obviously, X is compactly embedded into Lγ ([0, T ]) endowed with the usual norm k · kLγ , for all γ ≥ 1. Definition 2.8. A function u ∈ X is a weak solution of the problem (1.1) if there exists u∗ ∈ Lγ ([0, T ]) (for some γ > 1) such that Z Th m i X φp (u0 (x))v 0 (x) + M φp (u(x))v(x) − u∗ (x)v(x) dx − Ii (u(xi ))v(xi ) = 0 0 i=1 for all v ∈ X and u∗ ∈ λF (u(x)) + µG(x, u(x)) for a.e. x ∈ [0, T ]. Definition 2.9. By a solution of the impulsive differential inclusion (1.1) we will understand a function u : [0, T ] \ Q → R is of class C 1 with φp (u0 ) absolutely continuous, satisfying −(φp (u0 (x)))0 + M φp (u(x)) = u∗ 0 −∆φp (u (xk )) = Ik (u(xk )), u(0) = −u(T ), in [0, T ] \ Q, k = 1, 2, . . . , m, 0 u (0) = −u0 (T ), where u∗ ∈ λF (u(x)) + µG(x, u(x)) and u∗ ∈ Lγ ([0, T ]) (for some γ > 1). Lemma 2.10 ([16, Lemma 3.5]). If a function u ∈ X is a weak solution of (1.1), then u is a classical solution of (1.1). We introduce for a.e. x ∈ [0, T ] and all s ∈ R, the Aumann-type set-valued integral Z s nZ s o F (t)dt = f (t)dt : f : R → R is a measurable selection of F 0 0 RT Ru and set F(u) = 0 min 0 F (s) ds dx for all u ∈ Lp ([0, T ]); the Aumann-type setvalued integral Z s nZ s o G(x, t)dt = g(x, t)dt : g : [0, T ] × R → R is a measurable selection of G 0 0 and set G(u) = RT 0 min Ru 0 G(x, s) ds dx for all u ∈ Lp ([0, T ]). Lemma 2.11 ([10, Lemma 3.1]). The functionals F, G : Lp ([0, T ]) → R are well defined and Lipschitz on any bounded subset of Lp ([0, T ]). Moreover, for all u ∈ Lp ([0, T ]) and all u∗ ∈ ∂(F(u) + G(u)), u∗ (x) ∈ F (u(x)) + G(x, u(x)) for a.e. x ∈ [0, T ]. We define an energy functional for the problem (1.1) by setting m Z u(xi ) X 1 p Ii (s)ds Iλ (u) = kukX − λF(u) − µG(u) − p i=1 0 for all u ∈ X. 6 S. HEIDARKHANI, G. A. AFROUZI, A. HADJIAN, J. HENDERSON EJDE-2013/97 Lemma 2.12 ([16, Lemma 4.4]). The functional Iλ : X → R is locally Lipschitz. Moreover, for each critical point u ∈ X of Iλ , u is a weak solution of (1.1). 3. Main results We formulate our main result using the following assumptions: (F4) Rt sup|t|≤ξ min 0 F (s)ds lim inf ξ→+∞ ξp RT R ξ( T −x) F (s) ds dx min 0 2 1 2 p 0 ; < ( ) lim sup Pm R ξ( T2 −xi ) 2M T 1 p T p p+1 − ξ→+∞ Ii (s)ds i=1 0 p ξ T + p+1 ( 2 ) (I1) Ii (0) = 0, Ii (s)s < 0, s ∈ R, i = 1, 2, . . . , m. Theorem 3.1. Assume that (F1)–(F4), (I1) hold. Let RT R ξ( T −x) . min 0 2 F (s) ds dx 0 P λ1 := 1 lim sup , m R ξ( T2 −xi ) ξ→+∞ 1 ξ p T + 2M ( T )p+1 − Ii (s)ds i=1 0 p p+1 2 Rt . sup|t|≤ξ min 0 F (s)ds p . λ2 := 1 lim inf ξ→+∞ 1 p 2ξ T Then, for every λ ∈ (λ1 , λ2 ), and every multifunction G satisfying RT Rt (G4) 0 min 0 G(x, s) ds dx ≥ 0 for all t ∈ R, and (G5) G∞ := limξ→+∞ if we put R sup|t|≤ξ min 0t G(x,s)ds ξp < +∞, Rt sup|t|≤ξ min 0 F (s)ds 1 2p pT p µG,λ := 1 − λ p lim inf , pG∞ T p 2 ξ→+∞ ξp where µG,λ = +∞ when G∞ = 0, problem (1.1) admits an unbounded sequence of solutions for every µ ∈ [0, µG,λ ) in X. Proof. Our aim is to apply Theorem 2.6(b) to (1.1). To this end, we fix λ ∈ (λ1 , λ2 ) and let G be a multifunction satisfying (G1)–(G5). Since λ < λ2 , we have Rt sup|t|≤ξ min 0 F (s)ds 1 2p pT p µG,λ = 1 − λ p lim inf > 0. pG∞ T p 2 ξ→+∞ ξp Now fix µ ∈ (0, µg,λ ), put ν1 := λ1 , and ν2 := 1+ λ2 pT p µ 2p λ λ2 G∞ . If G∞ = 0, then ν1 = λ1 , ν2 = λ2 and λ ∈ (ν1 , ν2 ). If G∞ 6= 0, since µ < µG,λ , we have λ pT p + p µ G∞ < 1, λ2 2 and so λ2 > λ, pT p µ 1 + 2p λ λ2 G∞ EJDE-2013/97 EXISTENCE OF INFINITELY MANY ANTI-PERIODIC SOLUTIONS 7 namely, λ < ν2 . Hence, taking into account that λ > λ1 = ν1 , one has λ ∈ (ν1 , ν2 ). Now, set µ J(x, s) := F (s) + G(x, s) λ for all (x, s) ∈ [0, T ] × R. Assume j identically zero in X and for each u ∈ X put Z u Z T m Z u(xi ) X 1 p J(x, s) ds dx, N (u) := kukX − min Ii (s)ds, Υ(u) := p 0 0 i=1 0 M(u) := Υ(u) − j(u) = Υ(u), Iλ (u) := N (u) − λM(u) = N (u) − λΥ(u). It is a simple matter to verify that N is sequentially weakly lower semicontinuous on X. Clearly, N ∈ C 1 (X). By Lemma 2.1, N is locally Lipschitz on X. By Lemma 2.11, F and G are locally Lipschitz on Lp ([0, T ]). So, Υ is locally Lipschitz on Lp ([0, T ]). Moreover, X is compactly embedded into Lp ([0, T ]). So Υ is locally Lipschitz on X. Furthermore, Υ is sequentially weakly upper semicontinuous. For all u ∈ X, by (I1 ), Z u(xi ) Ii (s)ds < 0, i = 1, 2, . . . , m. 0 So, we have m X 1 N (u) = kukpX − p i=1 Z u(xi ) Ii (s)ds > 0 1 kukpX p for all u ∈ X. Hence, N is coercive and inf X N = N (0) = 0. We want to prove that, under our hypotheses, there exists a sequence {un } ⊂ X of critical points for the functional Iλ , that is, every element un satisfies Iλ◦ (un , v − un ) ≥ 0, for every v ∈ X. Now, we claim that γ < +∞. To see this, let {ξn } be a sequence of positive numbers such that limn→+∞ ξn = +∞ and Rt Rt sup|t|≤ξn min 0 J(x, s)ds sup|t|≤ξ min 0 J(x, s)ds lim = lim inf . (3.1) n→+∞ ξ→+∞ ξnp ξp Put p 1 2ξn rn := , for all n ∈ N. p T 1/q Then, for all v ∈ X with N (v) < rn , taking into account that kvkpX < prn and kvkC 0 ≤ 12 T 1/q kvkX , one has |v(x)| ≤ ξn for every x ∈ [0, T ]. Therefore, for all n ∈ N, supv∈N −1 ((−∞,r)) M(v) − M(u) ϕ(rn ) = inf r − N (u) u∈N −1 ((−∞,r)) µ supkvkpX <prn F(v) + λ G(v) ≤ rn Rt RT Rt RT sup|t|≤ξn 0 min 0 F (s) ds dx + µλ 0 min 0 G(x, s) ds dx ≤ rn Rt Rt h T p sup|t|≤ξn min 0 F (s)ds µ sup|t|≤ξn min 0 G(x, s)ds i ≤p + . 2 ξnp ξnp λ 8 S. HEIDARKHANI, G. A. AFROUZI, A. HADJIAN, J. HENDERSON EJDE-2013/97 Moreover, from Assumptions (F4) and (G5), we have Rt Rt sup|t|≤ξn min 0 F (s)ds µ sup|t|≤ξn min 0 G(x, s)ds lim + lim < +∞, n→+∞ n→+∞ λ ξnp ξnp which follows Rt sup|t|≤ξn min 0 J(x, s)ds lim < +∞. n→+∞ ξnp Therefore, Rt sup|t|≤ξ min 0 J(x, s)ds T p lim inf < +∞. (3.2) γ ≤ lim inf ϕ(rn ) ≤ p n→+∞ 2 ξ→+∞ ξp Since Rt Rt Rt sup|t|≤ξ min 0 J(x, s)ds sup|t|≤ξ min 0 F (s)ds µ sup|t|≤ξ min 0 G(x, s)ds ≤ + , ξp ξp ξp λ and taking (G5) into account, we get Rt Rt sup|t|≤ξ min 0 J(x, s)ds sup|t|≤ξ min 0 F (s)ds µ lim inf ≤ lim inf + G∞ . (3.3) ξ→+∞ ξ→+∞ ξp ξp λ Moreover, from Assumption (G4) we obtain RT R ξ( T −x) J(x, s) ds dx min 0 2 0 lim sup P m R ξ( T2 −xi ) ξ→+∞ 1 ξ p T + 2M ( T )p+1 − Ii (s)ds i=1 0 p p+1 2 R ξ( T2 −x) F (s) ds dx 0 . Pm R ξ( T2 −xi ) 2M T p+1 T + p+1 ( 2 ) − i=1 0 Ii (s)ds RT ≥ lim sup 0 ξ→+∞ 1 ξ p p min Therefore, from (3.3) and (3.4), we observe that 1 λ ∈ (ν1 , ν2 ) ⊆ R lim supξ→+∞ T 0 1 p pξ min 2M T p+1 T + p+1 (2) − 1 lim inf ξ→+∞ , R ξ( T2 −x) 0 (3.4) J(x,s) ds dx Pm R ξ( T2 −xi ) i=1 0 Ii (s)ds Rt J(x,s)ds 0 1 2ξ p ( ) p T sup|t|≤ξ min ⊆ (0, 1/γ). For the fixed λ, the inequality (3.2) ensures that the condition (b) of Theorem 2.6 can be applied and either Iλ has a global minimum or there exists a sequence {un } of weak solutions of the problem (1.1) such that limn→∞ kun k = +∞, The other step is to show that for the fixed λ the functional Iλ has no global minimum. Let us verify that the functional Iλ is unbounded from below. Since RT R ξ( T −x) min 0 2 F (s) ds dx 1 0 < lim sup Pm R ξ( T2 −xi ) 1 2M T p p+1 λ ξ→+∞ Ii (s)ds − i=1 0 p ξ T + p+1 ( 2 ) RT R ξ( T2 −x) min 0 J(x, s) ds dx 0 ≤ lim sup , P m R ξ( T2 −xi ) ξ→+∞ 1 ξ p T + 2M ( T )p+1 − I (s)ds i i=1 0 p p+1 2 EJDE-2013/97 EXISTENCE OF INFINITELY MANY ANTI-PERIODIC SOLUTIONS 9 there exists a sequence {ηn } of positive numbers and a constant τ such that limn→+∞ ηn = +∞ and RT R η ( T −x) min 0 n 2 J(x, s) ds dx 1 0 <τ < (3.5) P m R ηn ( T2 −xi ) 1 λ ηn p T + 2M ( T )p+1 − Ii (s)ds p p+1 i=1 0 2 for each n ∈ N large enough. For all n ∈ N, set T −x . 2 For any fixed n ∈ N, it is easy to see that wn ∈ X and, in particular, one has 2M T p+1 , kwn kpX = ηnp T + ( ) p+1 2 wn (x) := ηn and so T m Z 1 p 2M T p+1 X ηn ( 2 −xi ) Ii (s)ds. ηn T + ( ) − p p+1 2 i=1 0 N (wn ) = (3.6) By (3.5) and (3.6), we see that Iλ (wn ) = N (wn ) − λM(wn ) T m Z 2M T p+1 X ηn ( 2 −xi ) 1 ( ) − Ii (s)ds = ηnp T + p p+1 2 i=1 0 Z T Z ηn ( T2 −x) −λ J(x, s) ds dx min 0 0 < 1 p ηnp T m Z 2M T p+1 X ηn ( 2 −xi ) ( ) T+ − Ii (s)ds (1 − λτ ) p+1 2 i=1 0 for every n ∈ N large enough. Since λτ > 1 and limn→+∞ ηn = +∞, we have lim I (wn ) n→+∞ λ = −∞. Then, the functional Iλ is unbounded from below, and it follows that Iλ has no global minimum. Therefore, from part (b) of Theorem 2.6, the functional Iλ admits a sequence of critical points {un } ⊂ X such that limn→+∞ N (un ) = +∞. Since N is bounded on bounded sets, and taking into account that limn→+∞ N (un ) = +∞, then {un } has to be unbounded, i.e., lim kun kX = +∞. n→+∞ Moreover, if un ∈ X is a critical point of Iλ , clearly, by definition, one has Iλ◦ (un , v − un ) ≥ 0, for every v ∈ X. Finally, by Lemma 2.12, the critical points of Iλ are weak solutions for the problem (1.1), and by Lemma 2.10, every weak solution of (1.1) is a solution of (1.1). Hence, the assertion follows. Remark 3.2. Under the conditions sup|t|≤ξ min lim inf ξ→+∞ ξp Rt 0 F (s)ds = 0, 10 S. HEIDARKHANI, G. A. AFROUZI, A. HADJIAN, J. HENDERSON RT lim sup ξ→+∞ 1 ξ p p R ξ( T2 −x) min 0 0 2M T p+1 p+1 ( 2 ) T+ EJDE-2013/97 − F (s) ds dx Pm R ξ( T2 −xi ) i=1 0 = +∞, Ii (s)ds p from Theorem 3.1, we see that for every λ > 0 and for each µ ∈ 0, pT 2p G∞ , problem (1.1) admits a sequence of solutions which is unbounded in X. Moreover, if G∞ = 0, the result holds for every λ > 0 and µ ≥ 0. The following result is a special case of Theorem 3.1 with µ = 0. Theorem 3.3. Assume that (F1)–(F4), (I1) hold. Then, for each 1 λ∈ RT lim supξ→+∞ min 0 0 2M T p+1 T + p+1 (2) − 1 p pξ F (s) ds dx Pm R ξ( T2 −xi ) i=1 0 Ii (s)ds ! 1 lim inf ξ→+∞ , R ξ( T2 −x) sup|t|≤ξ min Rt 0 1 2ξ p p( T ) F (s)ds , the problem −(φp (u0 (x)))0 + M φp (u(x)) ∈ λF (u(x)) 0 −∆φp (u (xk )) = Ik (u(xk )), in [0, T ] \ Q, k = 1, 2, . . . , m, 0 u (0) = −u0 (T ) u(0) = −u(T ), has an unbounded sequence of solutions in X. Now, we present the following example to illustrate our results. Example 3.4. Consider the problem −(φ3 (u0 (x)))0 + φ3 (u(x)) ∈ λF (u(x)) 0 −∆φ3 (u (x1 )) = I1 (u(x1 )), 0 u(0) = −u(2), in [0, 2] \ {1}, x1 = 1, 0 u (0) = −u (2), where, for s ∈ R, {0}, [0, 1], F (s) = {s − 2−1/3 + 1}, {s + 2−1/3 + 1}, if if if if |s| < 2−1/3 , |s| = 2−1/3 , s > 2−1/3 , s < −2−1/3 . Simple calculations show that Z sup min |t|≤2−1/3 and R2 min R ξ(1−x) F (s) ds dx 0 R ξ(1−x1 ) − 0 I1 (s)ds Z 1 Z ξx 6 1 = min F (s) ds dx 5 ξ 3 −1 0 0 5 3 6ξ t F (s)ds = 0 0 (3.7) EJDE-2013/97 EXISTENCE OF INFINITELY MANY ANTI-PERIODIC SOLUTIONS 11 Z ξx Z 0 Z −1/3 Z ξx 6 1 −2 max F (s) ds dx max F (s) ds dx + = 5 ξ 3 −1 −2−1/3 0 0 Z 2−1/3 Z ξx Z 1 Z ξx + max F (s) ds dx + max F (s) ds dx > 0 2−1/3 0 0 0 for some ξ ∈ R. So, sup|t|≤ξ min Rt F (s)ds 0 = 0, 1 3 ξ→+∞ ξ 3 R2 R ξ(1−x) min 0 F (s) ds dx 0 > 0. lim sup R ξ(1−x1 ) 5 3 ξ→+∞ I1 (s)ds 6ξ − 0 lim inf Hence, using Theorem 3.3, problem (3.7), for λ lying in a convenient interval, has an unbounded sequence of solutions in X := {u ∈ W 1,3 ([0, 2]) : u(0) = −u(2)}. Here we point out the following consequences of Theorem 3.3, using the assumptions R p sup|t|≤ξ min 0t F (s)ds (F5) lim inf ξ→+∞ < p1 T2 ; ξp RT (F6) lim supξ→+∞ 0 1 p pξ min 2M T T + p+1 (2 R ξ( T2 −x) 0 )p+1 − F (s) ds dx Pm R ξ( T2 −xi ) i=1 0 > 1. Ii (s)ds Corollary 3.5. Assume that (F1)–(F3), (F5)–(F6), (I1) hold. Then, the problem −(φp (u0 (x)))0 + M φp (u(x)) ∈ F (u(x)) 0 −∆φp (u (xk )) = Ik (u(xk )), in [0, T ] \ Q, k = 1, 2, . . . , m, 0 u (0) = −u0 (T ) u(0) = −u(T ), has an unbounded sequence of solutions in X. Remark 3.6. Theorem 1.1 in the Introduction is an immediate consequence of Corollary 3.5. Now, we give the following consequence of the main result. Corollary 3.7. Let F1 : R → 2R be an upper semicontinuous multifunction with compact convex values, such that min F1 , max F1 : R → R are Borel measurable and |ξ| ≤ a(1 + |s|r1 −1 ) for all s ∈ R, ξ ∈ F1 (s), r1 > 1 (a > 0). Furthermore, suppose that R (C1) lim inf ξ→+∞ sup|t|≤ξ min ξp RT (C2) lim supξ→+∞ 0 1 p pξ t 0 F1 (s)ds min < +∞; R ξ( T2 −x) 0 2M T p+1 T + p+1 (2) − F1 (s) ds dx Pm R ξ( T2 −xi ) i=1 0 = +∞. Ii (s)ds Then, for every multifunction F2 : R → 2R which is upper semicontinuous with compact convex values, min F2 , max F2 : R → R are Borel measurable and |ξ| ≤ b(1 + |s|r2 −1 ) for all s ∈ R, ξ ∈ F2 (s), r2 > 1 (b > 0), and satisfies the conditions Z t sup min F2 (s)ds ≤ 0 t∈R and RT 0 lim inf ξ→+∞ 1 p pξ T+ min 0 R ξ( T2 −x) 0 2M T p+1 p+1 ( 2 ) − F2 (s) ds dx Pm R ξ( T2 −xi ) i=1 0 > −∞, Ii (s)ds 12 S. HEIDARKHANI, G. A. AFROUZI, A. HADJIAN, J. HENDERSON EJDE-2013/97 for each 1 λ ∈ 0, lim inf ξ→+∞ sup|t|≤ξ min Rt 0 1 2ξ p p( T ) F1 (s)ds , and the problem −(φp (u0 (x)))0 + M φp (u(x)) ∈ λ(F1 (u(x)) + F2 (u(x))) 0 −∆φp (u (xk )) = Ik (u(xk )), in [0, T ] \ Q, k = 1, 2, . . . , m, 0 u (0) = −u0 (T ) u(0) = −u(T ), has an unbounded sequence of solutions in X. Proof. Set F (t) = F1 (t) + F2 (t) for all t ∈ R. Assumption (C2) along with the condition RT R ξ( T −x) min 0 2 F2 (s) ds dx 0 P lim inf > −∞ m R ξ( T2 −xi ) ξ→+∞ 1 p 2M T p+1 ξ T + ( ) − I (s)ds i i=1 0 p p+1 2 yield RT 0 lim sup ξ→+∞ 1 ξ p p T+ RT = lim sup ξ→+∞ 0 min R ξ( T2 −x) 0 2M T p+1 p+1 ( 2 ) − F (s) ds dx Pm R ξ( T2 −xi ) Ii (s)ds R ξ( T −x) F1 (s) ds dx + 0 min 0 2 F2 (s) ds dx min 0 = +∞. Pm R ξ( T2 −xi ) 1 p 2M T p+1 − i=1 0 Ii (s)ds p ξ T + p+1 ( 2 ) i=1 0 R ξ( T2 −x) RT Moreover, Assumption (C1) and the condition Z t sup min F2 (s)ds ≤ 0 t∈R 0 ensure that sup|t|≤ξ min lim inf ξ→+∞ ξp Rt 0 F (s)ds Rt sup|t|≤ξ min 0 F1 (s)ds ≤ lim inf < +∞. ξ→+∞ ξp Since 1 lim inf ξ→+∞ sup|t|≤ξ min Rt 0 1 2ξ p p( T ) F (s)ds 1 ≥ lim inf ξ→+∞ sup|t|≤ξ min by applying Theorem 3.3, we have the desired conclusion. Rt 0 1 2ξ p p( T ) F1 (s)ds , Remark 3.8. We observe that in Theorem 3.1 we can replace ξ → +∞ with ξ → 0+ , and then by the same argument as in the proof of Theorem 3.1, but using conclusion (c) of Theorem 2.6 instead of (b), problem (1.1) has a sequence of solutions, which strongly converges to 0 in X. Acknowledgments. Shapour Heidarkhani was supported by a grant 91470046 from IPM. EJDE-2013/97 EXISTENCE OF INFINITELY MANY ANTI-PERIODIC SOLUTIONS 13 References [1] G. Bonanno, G. Molica Bisci; Infinitely many solutions for a boundary value problem with discontinuous nonlinearities, Boundary Value Problems, 2009 (2009), 1-20. [2] G. Bonanno, G. Molica Bisci, V. Rădulescu; Existence of three solutions for a nonhomogeneous Neumann problem through Orlicz-Sobolev spaces, Nonlinear Anal., 74 (2011), 4785-4795. [3] H. L. Chen; Antiperiodic wavelets, J. Comput. Math., 14 (1996), 32-39. [4] F. H. Clarke; Optimization and Nonsmooth Analysis, Wiley, New York, 1983. [5] F. J. Delvos, L. Knoche; Lacunary interpolation by antiperiodic trigonometric polynomials, BIT, 39 (1999), 439-450. [6] P. Djakov, B. Mityagin; Simple and double eigenvalues of the Hill operator with a two-term potential, J. Approx. Theory, 135 (2005), 70-104. [7] L. H. Erbe, W. Krawcewicz; Existence of solutions to boundary value problems for impulsive second order differential inclusions, Rocky Mountain J. Math., 22 (1992), 519-539. [8] M. Frigon and D. O’Regan; First order impulsive initial and periodic problems with variable moments, J. Math. Anal. Appl., 233 (1999), 730-739. [9] A. Iannizzotto; Three critical points for perturbed nonsmooth functionals and applications, Nonlinear Anal., 72 (2010), 1319-1338. [10] A. Iannizzotto; Three periodic solutions for an ordinary differential inclusion with two parameters, Ann. Polon. Math., 103 (2012), 89-100. [11] A. Kristály; Infinitely many solutions for a differential inclusion problem in RN , J. Differential Equations, 220 (2006), 511-530. [12] A. Kristály, W. Marzantowicz, C. Varga; A non-smooth three critical points theorem with applications in differential inclusions, J. Glob. Optim., 46 (2010), 49-62. [13] S. A. Marano, D. Motreanu; Infinitely many critical points of non-differentiable functions and applications to a Neumann-type problem involving the p-Laplacian, J. Differential Equations, 182 (2002), 108-120. [14] D. Motreanu, P.D. Panagiotopoulos; Minimax Theorems and Qualitative Properties of the Solutions of Hemivariational Inequalities, Kluwer Academic Publishers, Dordrecht, 1999. [15] B. Ricceri; A general variational principle and some of its applications, J. Comput. Appl. Math., 113 (2000), 401-410. [16] Y. Tian, J. Henderson; Three anti-periodic solutions for second-order impulsive differential inclusions via nonsmooth critical point theory, Nonlinear Anal., 75 (2012), 6496-6505. Shapour Heidarkhani Department of Mathematics, Faculty of Sciences, Razi University, 67149 Kermanshah, Iran. School of Mathematics, Institute for Research in Fundamental Sciences (IPM), P.O. Box: 19395-5746, Tehran, Iran E-mail address: s.heidarkhani@razi.ac.ir Ghasem A. Afrouzi Department of Mathematics, Faculty of Mathematical Sciences, University of Mazandaran, Babolsar, Iran E-mail address: afrouzi@umz.ac.ir Armin Hadjian Department of Mathematics, Faculty of Mathematical Sciences, University of Mazandaran, Babolsar, Iran E-mail address: a.hadjian@umz.ac.ir Johnny Henderson Department of Mathematics, Baylor University, Waco, TX 76798-7328, USA E-mail address: Johnny Henderson@baylor.edu