Electronic Journal of Differential Equations, Vol. 2013 (2013), No. 249, pp. 1–13. ISSN: 1072-6691. URL: http://ejde.math.txstate.edu or http://ejde.math.unt.edu ftp ejde.math.txstate.edu MULTIPLICITY OF SOLUTIONS FOR QUASILINEAR EQUATIONS INVOLVING CRITICAL ORLICZ-SOBOLEV NONLINEAR TERMS JEFFERSON A. SANTOS Abstract. In this work, we study the existence and multiplicity of solutions for a class of problems involving the φ-Laplacian operator in a bounded domain, where the nonlinearity has a critical growth. Our main tool is the variational method combined with the genus theory for even functionals. 1. Introduction In this article, we consider the existence and multiplicity of solutions for the quasilinear problem − div φ(|∇u|)∇u = λφ∗ (|u|)u + f (x, u), in Ω (1.1) u = 0, on ∂Ω where Ω ⊂ RN is a bounded domain in RN with smooth boundary, λ is a positive parameter and φ : (0, +∞) → R is a continuous function satisfying (φ(t)t)0 > 0 ∀t > 0. (1.2) There exist l, m ∈ (1, N ) such that l≤ φ(|t|)t2 ≤m Φ(t) ∀t 6= 0, (1.3) where Z |t| mN lN , m∗ = . N − l N −m 0 Moreover, φ∗ (t)t is such that Sobolev conjugate function Φ∗ of Φ is its primitive; R |t| that is, Φ∗ (t) = 0 φ∗ (s)sds. Related to function f : Ω × R → R, we assume the following: (F1) f ∈ C(Ω × R, R) is odd with respect t and f (x, t) = o φ(|t|)|t| , as |t| → 0 uniformly in x; f (x, t) = o φ∗ (|t|)|t| , as |t| → +∞ uniformly in x; Φ(t) = φ(s)s ds, l ≤ m < l∗ , l∗ = 2000 Mathematics Subject Classification. 35J62, 35B33, 35A15. Key words and phrases. Quasilinear elliptic equations; critical growth; variational methods. c 2013 Texas State University - San Marcos. Submitted October 8, 2013. Published November 20, 2013. Partially supported supported by PROCAD/CAPES-Brazil. 1 2 J. A. SANTOS EJDE-2013/249 (F2) There is θ ∈ (m, l∗ ) such that F (x, t) ≤ θ1 f (x, t)t, for all t > 0 and a.e. in Rt Ω, where F (x, t) = 0 f (x, s)ds. Problem (1.1) associated with nonhomogenous nonlinear Φ arises in various fields of physics [16]: (i) in nonlinear elasticity, Φ(t) = (1 + |t|2 )γ − 1 for γ ∈ (1, NN−2 ). (ii) in plasticity, Φ(t) = |t|p ln(1 + |t|) for 1 < p0 < p < N − 1 with p0 = √ −1+ 1+4N . 2 Rt (iii) in generalized Newtonian fluids, Φ(t) = 0 s1−α (sinh−1 s)β ds, 0 ≤ α ≤ 1, β > 0. Our main result reads as follows. Theorem 1.1. Assume that (1.2), (1.3), (F1) and (F2) are satisfied. Then, there exist a sequence {λk } ⊂ (0, +∞) with λk+1 < λk , such that, for λ ∈ (λk+1 , λk ), problem (1.1) has at least k pairs of nontrivial solutions. The main difficulty to prove Theorem 1.1 is related to the fact that the nonlinearity f has a critical growth. In this case, it is not clear that functional energy associated with (1.1) satisfies the well known (PS) condition, once that the embedding W 1,Φ (Ω) ,→ LΦ∗ (Ω) is not compact. To overcome this difficulty, we use a version of the concentration compactness lemma due to Lions for Orlicz-Sobolev space found in Fukagai, Ito and Narukawa [14]. We would like to mention that Theorem 1.1 improves the main result found in [25]. We cite the papers of Alves and Barreiro [3], Alves, Gonçalves and Santos [4], Bonano, Bisci and Radulescu [5], Cerny [7], Clément, Garcia-Huidobro and Manásevich [9], Donaldson [10], Fuchs and Li [12], Fuchs and Osmolovski [13], Fukagai, Ito and Narukawa [14, 15], Gossez [17], Mihailescu and Raduslescu [19, 20], Mihailescu and Repovs [21], Pohozaev [22] and references therein, where quasilinear problems like (1.1) have been considered in bounded and unbounded domains of RN . In some those papers, the authors have mentioned that this class of problem arises in applications, such as, nonlinear elasticity, plasticity and non-Newtonian fluids. This paper is organized as follows: In Section 2, we collect some preliminaries on Orlicz-Sobolev spaces that will be used throughout the paper, which can be found in [1], [2], [11] and [23]. In Section 3, we recall an abstract theorem involving genus theory that will use in the proof of Theorem 1.1 and prove some technical lemmas, and then we prove Theorem 1.1. 2. Preliminaries on Orlicz-Sobolev spaces First of all, we recall that a continuous function A : R → [0, +∞) is a N -function if: (A1) A is convex. (A2) A(t) = 0 if and only if t = 0. (A3) A(t) t → 0 as t → 0, and A(t)/t → ∞ as t → +∞. (A4) A is an even function. In what follows, we say that a N -function A satisfies the ∆2 -condition if, there exists t0 ≥ 0 and k > 0 such that A(2t) ≤ kA(t) ∀t ≥ t0 . EJDE-2013/249 MULTIPLICITY OF SOLUTIONS 3 This condition can be rewritten of the following way: For each s > 0, there exists Ms > 0 and t0 ≥ 0 such that A(st) ≤ Ms A(t), ∀t ≥ t0 . (2.1) Fixed an open set Ω ⊂ RN and a N-function A satisfying ∆2 -condition, the space LA (Ω) is the vectorial space of the measurable functions u : Ω → R such that Z A(u) < ∞. Ω The space LA (Ω) endowed with Luxemburg norm, Z n o u |u|A = inf α > 0 : A ≤1 , α Ω e is a Banach space. The complement function of A, denoted by A(s), is given by the Legendre transformation, e = max{st − A(t)} A(s) t≥0 for s ≥ 0. e are complementary each other. Moreover, we have the The functions A and A Young’s inequality e st ≤ A(t) + A(s), ∀t, s ≥ 0. (2.2) Using this inequality, it is possible to prove the Hölder type inequality Z uv ≤ 2|u|A |v|Ae, ∀u ∈ LA (Ω) and v ∈ LAe(Ω). (2.3) Ω Another important function related to function A is the Sobolev’s conjugate function A∗ of A, defined by Z t A−1 (s) ds, for t > 0. A−1 (t) = ∗ (N +1)/N 0 s ∗ ∗ When A(t) = |t|p for 1 < p < N , we have A∗ (t) = p∗ p |t|p , where p∗ = pN N −p . Hereafter, we denote by W01,A (Ω) the Orlicz-Sobolev space obtained by the completion of C0∞ (Ω) with respect to norm kuk = |∇u|A + |u|A . e satisfy the ∆2 -condition, then the An important property is that: If A and A spaces LA (Ω) and W 1,A (Ω) are reflexive and separable. Moreover, the ∆2 -condition also implies that Z un → u in LA (Ω) ⇐⇒ A(|un − u|) → 0 (2.4) Ω and un → u in W 1,A Z (Ω) ⇐⇒ Z A(|un − u|) → 0 and Ω A(|∇un − ∇u|) → 0. (2.5) Ω Another important inequality was proved by Donaldson and Trudinger [10], which establishes that for all open Ω ⊂ RN and there is a constant SN = S(N ) > 0 such that |u|A∗ ≤ SN |∇u|A , u ∈ W01,A (Ω). (2.6) 4 J. A. SANTOS EJDE-2013/249 Moreover, exist C0 > 0 such that Z Z A(u) ≤ C0 A(|∇u|), u ∈ W01,A (Ω). Ω (2.7) Ω This inequality shows the following embedding is continuous W01,A (Ω) ,→ LA∗ (Ω). If Ω is a bounded domain and the two limits hold B(t) B(t) < +∞, lim sup = 0, lim sup A(t) A t→0 ∗ (t) |t|→+∞ (2.8) then the embedding W01,A (Ω) ,→ LB (Ω) (2.9) is compact. e and Φ∗ and theirs proofs can The next four lemmas involving the functions Φ, Φ e Φ∗ be found in [14]. Hereafter, Φ is the N -function given in the introduction and Φ, are the complement and conjugate functions of Φ respectively. Lemma 2.1. Assume (1.2) and (1.3). Then Z |t| Φ(t) = sφ(s)ds, 0 e ∈ ∆2 . Hence, LΦ (Ω), W 1,Φ (Ω) and W 1,Φ (Ω) are reflexive is a N -function with Φ, Φ 0 and separable spaces. e and Φ e ∗ satisfy the inequality Lemma 2.2. The functions Φ, Φ∗ , Φ e e ∗ (φ∗ (|t|)t) ≤ Φ∗ (2t), ∀t ≥ 0. Φ(φ(|t|)t) ≤ Φ(2t), Φ (2.10) l m Lemma 2.3. Assume that (1.2) and (1.3) hold and let ξ0 (t) = min{t , t }, ξ1 (t) = max{tl , tm }, for all t ≥ 0. Then ξ0 (ρ)Φ(t) ≤ Φ(ρt) ≤ ξ1 (ρ)Φ(t) for ρ, t ≥ 0, Z ξ0 (|u|Φ ) ≤ Φ(u) ≤ ξ1 (|u|Φ ) for u ∈ LΦ (Ω). Ω Lemma 2.4. The function Φ∗ satisfies the inequality Φ0 (t)t ≤ m∗ for t > 0. l∗ ≤ ∗ Φ∗ (t) As an immediate consequence of the Lemma 2.4, we have the following result ∗ ∗ Lemma 2.5. Assume that (1.2) and (1.3) hold and let ξ2 (t) = min{tl , tm }, ∗ ∗ ξ3 (t) = max{tl , tm } for all t ≥ 0. Then ξ2 (ρ)Φ∗ (t) ≤ Φ∗ (ρt) ≤ ξ3 (ρ)Φ∗ (t) for ρ, t ≥ 0, Z ξ2 (|u|Φ∗ ) ≤ Φ∗ (u)dx ≤ ξ3 (|u|Φ∗ ) for u ∈ LA∗ (Ω). Ω e be the complement of Φ and put Lemma 2.6. Let Φ l m ξ4 (s) = min{s l−1 , s m−1 }, l m ξ5 (s) = max{s l−1 , s m−1 }, Then the following inequalities hold e e e ξ4 (r)Φ(s) ≤ Φ(rs) ≤ ξ5 (r)Φ(s), r, s ≥ 0; s ≥ 0. EJDE-2013/249 MULTIPLICITY OF SOLUTIONS 5 Z ξ4 (|u|Φ e) ≤ Ω e Φ(u)dx ≤ ξ5 (|u|Φ e ), u ∈ LΦ e (Ω). 3. An abstract theorem and technical lemmas In this section we recall an important abstract theorem involving genus theory, which will use in the proof of Theorem 1.1. After, we prove some technical lemmas that will use to show that the energy functional associated with problem (1.1) satisfies the hypotheses of the abstract theorem. 3.1. An abstract theorem. Let E be a real Banach space and Σ the family of sets Y ⊂ E\{0} such that Y is closed in E and symmetric with respect to 0; that is, Σ = {Y ⊂ E\{0}; Y is closed in E and Y = −Y }. Hereafter, let us denote by γ(Y ) the genus of Y ∈ Σ (see [24, pp. 45]). Moreover, we set Kc = {u ∈ E; I(u) = c and I 0 (u) = 0}, Ac = {u ∈ E; I(u) ≤ c}. Next, we recall a version of the Mountain Pass Theorem for even functionals, whose proof can be found in [24]. Theorem 3.1. Let E be an infinite dimensional Banach space with E = V ⊕ X, where V is finite dimensional and let I ∈ C 1 (E, R) be a even function with I(0) = 0, and satisfying: (I1) there are constants β, ρ > 0 such that I(u) ≥ β > 0, for each u ∈ ∂Bρ ∩ X; (I2) there is Υ > 0 such that I satisfies the (P S)c condition, for 0 < c < Υ; e ⊂ E, there is R = R(E) e > 0 such (I3) for each finite dimensional subspace E e R (0). that I(u) ≤ 0 for all u ∈ E\B Suppose V is k dimensional and V = span{e1 , . . . , ek }. For m ≥ k, inductively choose em+1 6∈ Em := span{e1 , . . . , em }. Let Rm = R(Em ) and Dm = BRm ∩ Em . Define Gm := h ∈ C(Dm , E); h is odd and h(u) = u, ∀u ∈ ∂BRm ∩ Em }, (3.1) (3.2) Γj := h(Dm \Y ); h ∈ Gm , m ≥ j, Y ∈ Σ, and γ(Y ) ≤ m − j . For each j ∈ N, let cj = inf max I(u). (3.3) K∈Γj u∈K Then, 0 < β ≤ cj ≤ cj+1 for j > k, and if j > k, cj < Υ and cj is critical value of I. Moreover, if cj = cj+1 = · · · = cj+l = c < Υ for j > k, then γ(Kc ) ≥ l + 1. 3.2. Technical lemmas. Associated with problem (1.1), we have the energy functional Jλ : W01,Φ (Ω) → R defined by Z Z Z Jλ (u) = Φ(|∇u|) − λ Φ∗ (u) − F (x, u). Ω Ω 1 Ω W01,Φ (Ω), R with By conditions (F1) and (F2), Jλ ∈ C Z Z Z 0 Jλ (u) · v = φ(|∇u|)∇u∇v − λ φ∗ (|u|)uv − f (x, u)v, Ω Ω Ω 6 J. A. SANTOS EJDE-2013/249 for any u, v ∈ W01,Φ (Ω). Thus, critical points of Jλ are weak solutions of problem (1.1). Lemma 3.2. Under the conditions (F1) and (F2), the functional Jλ satisfies (I1). Proof. On the one hand, from (F1) and (F2), for a given > 0, there exists C > 0 such that |F (x, t)| ≤ Φ(t) + C Φ∗ (t), ∀(x, t) ∈ Ω̄ × R. (3.4) Combining (2.7) with (3.4), Z Jλ (u) ≥ (1 − C0 ) Z Φ(|∇u|) − (1 + C ) Φ∗ (u). Ω Ω For is small enough and kuk = ρ ' 0, from (2.6) and Lemma 2.4, it follows that ∗ ∗ l l Jλ (u) ≥ C1 |∇u|m Φ − C2 SN |∇u|Φ for some positive constants C1 and C2 . For m < l∗ , if ρ is small enough, there is β > 0 such that Jλ (u) ≥ β > 0 ∀u ∈ ∂Bρ (0), which completes the proof. Lemma 3.3. Under conditions (F1) and (F2), the functional Jλ satisfies (I3). Proof. Suppose (I3) does not hold. Then, there is a finite dimensional subspace e ⊂ W 1,Φ (Ω) and a sequence (un ) ⊂ E\B e n (0) satisfying E 0 Jλ (un ) > 0, ∀n ∈ N. (3.5) A direct computation shows that given > 0, there is a constant M > 0 such that − M − Φ∗ (t) ≤ F (x, t) ≤ M + Φ∗ (t), ∀(x, t) ∈ Ω × R. (3.6) Consequently, Z Jλ (un ) ≤ Z Φ(|∇un |)dx − λ Ω Z Φ∗ (un ) + M |Ω|. Φ∗ (un ) + Ω Ω Fixing = λ/2, and using Lemma 2.5, we obtain Z λ Jλ (un ) ≤ Φ(|∇un |) − ξ3 (|un |Φ∗ ) + M |Ω|. 2 Ω (3.7) Using that dim Ẽ < ∞, we know that any two norms are equivalent in Ẽ. Then, using that kun k → ∞, we can assume that |un |Φ∗ > 1. Thereby, from Lemmas 2.3 and 2.5, λ l∗ Jλ (un ) ≤ |∇un |m Φ − |un |Φ∗ + M |Ω|. 2 Using again the equivalence of the norms in Ẽ, there is C > 0 such that ∗ λ Jλ (un ) ≤ kun km − Ckun kl + M |Ω|. 2 Recalling that m < l∗ , the above inequality implies that there is n0 ∈ N such that Jλ (un ) < 0, which contradicts (3.5). ∀n ≥ n0 , Lemma 3.4. Under conditions (F1)–(F2), any (PS) sequence for Jλ is bounded in W01,Φ (Ω). EJDE-2013/249 MULTIPLICITY OF SOLUTIONS 7 Proof. Let {un } be a (P S)d sequence of Jλ . Then, Jλ (un ) → d, Jλ0 (un ) → 0 as n → +∞. We claim that {un } is bounded. Indeed, note that Z Z 1 0 1 Jλ (un ) − Jλ (un )un = Φ(|∇un |) − φ(|∇un |)|∇un |2 θ θ Ω Ω Z Z λ −λ φ∗ (|un |)u2n Φ∗ (un ) + θ Ω Ω Z Z 1 f (x, un )un . − F (x, un ) + θ Ω Ω Consequently, Z Z 1 1 λ Φ(|∇un |) φ∗ (|un |)u2n − Φ∗ (un ) = Jλ (un ) − Jλ0 (un )un − θ Ω θ Ω Z 1 + φ(|∇un |)|∇un |2 θ Ω Z 1 + F (x, un ) − f (x, un )un . θ Ω Then, by (1.3), (F2) and Lemma 2.4, for n sufficiently large, Z Z l∗ m λ( − 1) Φ∗ (un ) ≤ C + 1 + kun k + ( − 1) Φ(|∇un |), θ θ Ω Ω which implies that [λ( l∗ − 1)] θ Z Φ∗ (un ) ≤ C + kun k, Ω where C is a positive constant, and so Z Φ∗ (un )dx ≤ C(1 + kun k). (3.8) Ω By (3.6) and (3.8), Z Z Z Φ(|∇un |) ≤ Jλ (un ) + λ Φ∗ (un ) + F (x, un )dx Ω Ω Ω Z ≤ C + on (1) + (λ + ) Φ∗ (un ) Ω ≤ C(1 + kun k) + on (1). Therefore, for n sufficiently large, Z Φ(|∇un |) ≤ C (1 + kun k) . Ω If kun k > 1, from Lemma 2.5, it follows that kun kl ≤ C(1 + kun k). Using that l > 1, the above inequality gives that {un } is bounded in W01,Φ (Ω). As a consequence of the above result, if {un } is a (PS) sequence for Jλ , we can extract a subsequence of {un }, still denoted by {un } and u ∈ W01,Φ (Ω), such that 8 J. A. SANTOS • • • • EJDE-2013/249 un * u in W01,Φ (Ω); un * u in LΦ∗ (Ω); un → u in LΦ (Ω); un (x) → u(x) a.e. in Ω; From the concentration compactness lemma of Lions in Orlicz-Sobolev space found in [14], there exist two nonnegative measures µ, ν ∈ M(RN ), a countable set J , points {xj }j∈J in Ω and sequences {µj }j∈J , {νj }j∈J ⊂ [0, +∞), such that X Φ(|∇un |) → µ ≥ Φ(|∇u|) + µj δxj in M(RN ) (3.9) j∈J Φ∗ (un ) → ν = Φ∗ (u) + X in M(RN ) νj δxj (3.10) j∈J ∗ l∗ m∗ ∗ ∗ l∗ ∗ m∗ m l m l µj l , SN µjm , SN µjm }, νj ≤ max{SN µjl , SN (3.11) where SN satisfies (2.6). Next, we will show an important estimate for {νi }, from below. Firs, we prove a technical lemma. Lemma 3.5. Under the conditions of Lemma 3.4. If {un } is a (P S) sequence for Jλ and {νj } as above, then for each j ∈ J , β − α l β−1 SN β−1 ∗ λm νj ≥ ∗ ∗ ∗ or νj = 0, ∗ l for some α ∈ {l∗ , m∗ } and β ∈ { ll , ml , m ,m m }. Proof. Let ψ ∈ C0∞ (RN ) such that ψ(x) = 1 in B1/2 (0), supp ψ ⊂ B1 (0), 0 ≤ ψ(x) ≤ 1 ∀x ∈ RN . For each j ∈ Γ and > 0, let us define ψ (x) = ψ x − xj , ∀x ∈ RN . Then {ψ un } is bounded in W01,Φ (Ω). Since Jλ0 (un ) → 0, we have Jλ0 (un )(ψ un ) = on (1), or equivalently, Z Z Z 2 φ(|∇un |)∇un ∇ (un ψ ) = on (1) + λ φ∗ (|un |)un ψ + f (x, un )un ψ Ω Ω Z ZΩ (3.12) ≤ on (1) + λm∗ Φ∗ (un )ψ + f (x, un )un ψ . Ω Ω Using that lim t→+∞ f (x, t)tψ (x) = 0, Φ∗ (t) uniformly in x ∈ Ω and that limn→+∞ f (x, un )un ψ = 0 a.e. on Ω, we have by compactness Lemma of Strauss [8] (note that this result is still true when we replace RN by Ω) Z Z lim f (x, un )un ψ = f (x, u)uψ . (3.13) n→∞ Ω Ω EJDE-2013/249 MULTIPLICITY OF SOLUTIONS 9 On the other hand, by (1.3) Z Z Z 2 φ(|∇un |)∇un ∇(un ψ ) = φ(|∇un |)|∇un | ψ + φ(|∇un |)(∇un ∇ψ )un Ω Ω ZΩ Z ≥l Φ(|∇u|)ψ + φ(|∇un |)(∇un ∇ψ )un . Ω Ω (3.14) By Lemmas 2.2 and 2.6, the sequence {|φ(|∇un |)∇un |Φ e } is bounded. Thus, there is a subsequence {un } such that φ(|∇un |)∇un * w e1 N weakly in LΦ e (Ω, R ), N for some w e1 ∈ LΦ e (Ω, R ). Since un → u in LΦ (Ω), Z Z φ(|∇un |)(∇un ∇ψ )un → (w e1 ∇ψ )u. Ω Ω Thus, combining (3.12), (3.13), (3.14), and letting n → ∞, we have Z Z Z Z l ψ dµ + (w e1 ∇ψ )u ≤ λm∗ ψ dν + f (x, u)uψ . Ω Ω Ω (3.15) Ω Now we show that the second term of the left-hand side converges 0 as → 0. Claim 1: {f (x, un )} is bounded in LΦ e ∗ (Ω). In fact, by (F1) and Lemma 2.2 we have Z Z Z e ∗ (f (x, un )) ≤ c1 e ∗ (φ∗ (|un |)un ) + c2 e ∗ (φ(|un |)un ) Φ Φ Φ Ω Ω [|un |>1] Z e ∗ (φ(|un |)un ) + c3 Φ [|un |≤1] Z Z e ∗ (φ(|un |)un ) + c3 |Ω|. ≤ c1 Φ∗ (un ) + c2 Φ [|un |>1] Ω ∗ Hence, by (1.3), Lemma 2.3 and m < l , Z Z Z e e e ∗ (|un |m−1 ) + C3 |Ω| Φ∗ (f (x, un )) ≤ C1 Φ∗ (φ∗ (|un |)un ) + C2 Φ Ω Ω [|un |>1] Z Z e ∗ (|un |l∗ −1 ) + C3 |Ω|. ≤ C1 Φ∗ (un ) + C2 Φ [|un |>1] Ω Now, by Lemmas 2.4 and 2.2, Z Z e ∗ (f (x, un )) ≤ K1 Φ Φ∗ (un ) + K2 |Ω| < +∞. Ω Ω From Claim 1, there is a subsequence {un } such that φ∗ (|un |)un + f (x, un ) * w e2 weakly in LΦ e ∗ (Ω), for some w e2 ∈ LΦ e ∗ (Ω). Since Z Z 0 Jλ (un )v = φ(|∇un |)∇un ∇v − (φ∗ (|un |)un + f (x, un )) v → 0, Ω as n → ∞ for any v ∈ Ω W01,Φ (Ω), Z Ω (w e1 ∇v − w e2 v) = 0, 10 J. A. SANTOS EJDE-2013/249 for any v ∈ W01,Φ (Ω). Substituting v = uψ we have Z (w e1 ∇(uψ ) − w e2 uψ ) = 0. Ω Namely, Z Z Ω (w e1 ∇ψ )u = − Ω (w e1 ∇u − w e2 u)ψ . Noting w e1 ∇u − w e2 u ∈ L1 (Ω), we see that right-hand side tends to 0 as → 0. Hence we have Z (w e1 ∇ψ )u → 0, Ω as → 0. Letting → 0 in (3.15), we obtain lµj ≤ λm∗ νj . Hence, lλ β β −α SN νj ≤ µβj ≤ νj , m∗ ∗ ∗ ∗ ∗ l ,m for some α ∈ {l∗ , m∗ }, β ∈ { ll , ml , m m }, and so β − α l β−1 or νj = 0. νj ≥ SN β−1 ∗ λm Lemma 3.6. Assume that (F1)–(F2). Then, Jλ satisfies (P S)d for d ∈ (0, dλ ) where n l∗ − θ β l l∗ m∗ l∗ m∗ o dλ = min ( ∗ ) β−1 ; α ∈ {l∗ , m∗ }, β ∈ { , , , } . α 1 l l m m θSNβ−1 λ β−1 m Proof. Using that Jλ (un ) = d + on (1) and Jλ0 (un ) = on (1), we have 1 d = lim I(un ) = lim Jλ (un ) − Jλ0 (un )un n→∞ n→∞ θ Z Z h l∗ m Φ(|∇un |) + λ( − 1) Φ∗ (un ) ≥ lim (1 − ) n→∞ θ Ω θ Ω Z i 1 − F (x, un ) − f (x, un )un θ Ω Z l∗ Φ∗ (un ). ≥ λ( − 1) θ Ω Recalling that Z hZ X i lim Φ∗ (un )dx = Φ∗ (u) + νj ≥ νj , n→∞ Ω Ω j∈J we derive that α β l β−1 − β−1 l∗ −1 S N θ λm∗ α β 1 − β−1 l∗ − θ l β−1 = SN λ 1−β , ∗ θ m ∗ ∗ l ∗ m∗ for some α ∈ {l∗ , m∗ }, β ∈ { ll , ml , m , m }, which is an absurd. From this, we must have νj = 0 for any j ∈ J , leading to Z Z Φ∗ (un ) → Φ∗ (u). (3.16) d≥λ Ω Ω EJDE-2013/249 MULTIPLICITY OF SOLUTIONS 11 Combining the last limit with Brézis and Lieb [6], we obtain Z Φ∗ (un − u) → 0 as n → ∞, Ω from where it follows by Lemma 2.5 un → u in LΦ∗ (Ω). Now, as Jλ0 (un )un = on (1), the last limit gives Z Z φ(|∇un |)|un |2 = λ φ∗ (|un |)u2n + f (x, un )un + on (1). Z Ω Ω Ω In what follows, let us denote by {Pn } the sequence Pn (x) = hφ(|∇un (x)|)∇un (x) − φ(|∇u(x)|)∇u(x), ∇un (x) − ∇u(x)i. Since Φ is convex in R and Φ(|.|) is C 1 class in RN , has Pn (x) ≥ 0. From definition of {Pn }, Z Z Z Z 2 Pn = φ(|∇un |)|∇un | − φ(|∇un |)∇un ∇u − φ(|∇u|)∇u∇(un − u). Ω Ω Ω Ω W01,Φ (Ω), Recalling that un * u in we have Z φ(|∇u|)∇u∇(un − u) → 0 as n → ∞, (3.17) Ω which implies that Z Z Z Pn = φ(|∇un |)|∇un |2 − φ(|∇un |)∇un ∇u + on (1). Ω Ω Ω = on (1) and Jλ0 (un )u = on (1), we derive On the other hand, from Z Z Z 2 0≤ Pn = λ φ∗ (|un |)|un | − λ φ∗ (|un |)un u Ω Ω ZΩ Z + f (x, un )un − f (x, un )u + on (1). Jλ0 (un )un Ω Ω Combining (3.16) with the compactness Lemma of Strauss [8], we deduce that Z Pn → 0 as n → ∞. Ω Using that Φ is convex, from a result due to Dal Maso and Murat [18], it follows that ∇un (x) → ∇u(x) a.e. Ω. Now, using Lebesgue’s Theorem, Z Φ(|∇un − ∇u|)dx → 0, Ω which shows that un → u in W01,Φ (Ω). (3.18) The next lemma is similar to [25, Lemma 5] and its proof will be omitted. 12 J. A. SANTOS EJDE-2013/249 Lemma 3.7. Under conditions (F1)–(F2), there is sequence {Mm } ⊂ (0, +∞) independent of λ with Mm ≤ Mm+1 , such that for any λ > 0, cλm = inf max Jλ (u) < Mm . K∈Γm u∈K (3.19) Proof of Theorem 1.1. For each k ∈ N, choose λk such that Mk < dλk . Thus, for λ ∈ (λk+1 , λk ), 0 < cλ1 ≤ cλ2 ≤ · · · ≤ cλk < Mk ≤ dλ . By Theorem 3.1, the levels cλ1 ≤ cλ2 ≤ · · · ≤ cλk are critical values of Jλ . Thus, if cλ1 < cλ2 < · · · < cλk , the functional Jλ has at least k critical points. Now, if cλj = cλj+1 for some j = 1, 2, . . . , k, it follows from Theorem 3.1 that Kcλj is an infinite set [24, Cap. 7]. Then, in this case, problem (1.1) has infinitely many solutions. References [1] A. Adams, J. F. Fournier; Sobolev spaces, 2nd ed., Academic Press, (2003). [2] A. Adams, L. I. Hedberg; Function spaces and Potential Theory, Grundlehren der Mathematischen Wissenschaften, vol. 314, Springer-Verlag, Berlin, 1996. [3] C. O. Alves, J. L. P. Barreiro; Existence and multiplicity of solutions for a p(x)-Laplacian equation with critical growth, J. Math. Anal. Appl. (Print), 2013. [4] C. O. Alves, J. V. Gonçalves, J. A. Santos; Strongly Nonlinear Multivalued Elliptic Equations on a Bounded Domain, J Glob Optim, DOI 10.1007/s10898-013-0052-3, 2013. [5] G. Bonano, G. M. Bisci, V. Radulescu; Quasilinear elliptic non-homogeneous Dirichlet problems through Orlicz-Sobolev spaces, Nonl. Anal. 75 (2012), 4441-4456. [6] H. Brezis, E. Lieb; A relation between pointwise convergence of functions and convergence of functinals, Proc. Amer. Math. Soc. 88 (1983), 486-490. [7] R. Cerný; Generalized Moser-Trudinger inequality for unbounded domains and its application, Nonlinear Differ. Equ. Appl. DOI 10.1007/s00030-011-0143-0. [8] J. Chabrowski; Weak covergence methods for semilinear elliptic equations. World Scientific Publishing Company, 1999. [9] Ph. Clément, M. Garcia-Huidobro, R. Manásevich, K. Schmitt; Mountain pass type solutions for quasilinear elliptic equations, Calc. Var. 11 (2000), 33-62. [10] T. K. Donaldson; Nonlinear elliptic boundary value problems in Orlicz-Sobolev spaces, J. Diff. Equat. 10 (1971), 507-528. [11] T. K. Donaldson, N. S. Trudinger; Orlicz-Sobolev spaces and imbedding theorems, J. Funct. Anal. 8 (1971) 52-75. [12] M. Fuchs, G. Li; Variational inequalities for energy functionals with nonstandard growth conditions, Abstr. Appl. Anal. 3 (1998), 405-412. [13] M. Fuchs, V. Osmolovski; Variational integrals on Orlicz-Sobolev spaces. Z. Anal. Anwendungen 17, 393-415 (1998) 6. [14] N. Fukagai, M. Ito, K. Narukawa; Positive solutions of quasilinear elliptic equations with critical Orlicz-Sobolev nonlinearity on RN , Funkcialaj Ekvacioj 49 (2006), 235-267. [15] N. Fukagai, M. Ito, K. Narukawa; Quasilinear elliptic equations with slowly growing principal part and critical Orlicz-Sobolev nonlinear term, Proc. R. S. Edinburgh 139 A (2009), 73-106. [16] N. Fukagai and K. Narukawa; On the existence of multiple positive solutions of quasilinear elliptic eigenvalue problems, Annali di Matematica, 186, 539-564, 2007. [17] J. P. Gossez; Nonlinear elliptic boundary value problem for equations with rapidly (or slowly) increasing coefficients, Trans. Am. Math. Soc. 190 (1974), 163-205. [18] G. Dal Maso, F. Murat; Almost everywhere convergence of gradients of solutions to nonlinear elliptic systems, Nonlinear Anal. 31 (1998), 405-412. [19] M. Mihailescu, V. Radulescu; Nonhomogeneous Neumann problems in Orlicz-Sobolev spaces, C .R. Acad. Sci. Paris, Ser. I 346 (2008), 401-406. [20] M. Mihailescu, V. Radulescu; Existence and multiplicity of solutions for a quasilinear nonhomogeneous problems: An Orlicz-Sobolev space setting, J. Math. Anal. Appl. 330 (2007), 416-432. EJDE-2013/249 MULTIPLICITY OF SOLUTIONS 13 [21] M. Mihailescu, D. Repovs; Multiple solutions for a nonlinear and non-homogeneous problems in Orlicz-Sobolev spaces, Appl. Math. Comput. 217 (2011), 6624-6632. [22] S. L. Pohozaev; Eingenfunctions for the equation ∆u + λf (u) = 0, Soviet Math. Dokl. 6 (1965) 1408-1411. [23] R. O’Neill; Fractional integration in Orlicz spaces, Trans. Amer. Math. Soc. 115 (1965) 300328. [24] P. H. Rabinowitz; Minimax methods in critical point theory with applications to diferential equations. CBMS Reg. Conf. series in math 65, 1984. [25] W. Zhihui, W. Xinmin; A multiplicity result for quasilinear elliptic equations involving critical sobolev exponents. Nonlinear Analysis, Theory, Methods and Applications 18 (1992), 559–567. Jefferson A. Santos Universidade Federal de Campina Grande, Unidade Acadêmica de Matemática e Estatı́stica, CEP: 58429-900, Campina Grande - PB, Brazil E-mail address: jefferson@dme.ufcg.edu.br