Electronic Journal of Differential Equations, Vol. 2015 (2015), No. 45, pp. 1–11. ISSN: 1072-6691. URL: http://ejde.math.txstate.edu or http://ejde.math.unt.edu ftp ejde.math.txstate.edu THREE SOLUTIONS FOR A FOURTH-ORDER BOUNDARY-VALUE PROBLEM GHASEM A. AFROUZI, SAEID SHOKOOH Abstract. Using two three-critical points theorems, we prove the existence of at least three weak solutions for one-dimensional fourth-order equations. Some particular cases and two concrete examples are then presented. 1. Introduction In this note, we consider the fourth-order boundary-value problem u0000 h(x, u0 ) − u00 = [λf (x, u) + g(u)]h(x, u0 ), 00 in (0, 1), 00 u(0) = u(1) = 0 = u (0) = u (1), (1.1) where λ is a positive parameter, f : [0, 1] × R → R is an L1 -Carathéodory function, g : R → R is a Lipschitz continuous function with the Lipschitz constant L > 0, i.e., |g(t1 ) − g(t2 )| ≤ L|t1 − t2 | for every t1 , t2 ∈ R, with g(0) = 0, and h : [0, 1] × R → [0, +∞) is a bounded and continuous function with m := inf (x,t)∈[0,1]×R h(x, t) > 0. Due to the importance of fourth-order two-point boundary value problems in describing a large class of elastic deflection, many researchers have studied the existence and multiplicity of solutions for such a problem, we refer the reader to [1, 2, 3, 6, 11] and references therein. For example, authors in [2], using Ricceri’s Variational Principle [10, Theorem 1], established the existence three weak solutions for the problem u0000 + αu00 + βu = λf (x, u) + µg(x, u), 00 in (0, 1), 00 u(0) = u(1) = 0 = u (0) = u (1), where α, β are real constants, f, g : [0, 1] × R → R are Carathéodory functions and λ, µ > 0. In this article, employing two three-critical points theorems which we recall in the next section (Theorems 2.1 and 2.2), we establish the existence three weak solutions for (1.1). A special case of Theorem 3.1 is the following theorem. 2000 Mathematics Subject Classification. 34B15, 34B18, 58E05. Key words and phrases. Dirichlet boundary condition; variational method; critical point. c 2015 Texas State University - San Marcos. Submitted November 14, 2014. Published February 17, 2015. 1 2 G. A. AFROUZI, S. SHOKOOH EJDE-2015/45 Theorem 1.1. Let f : R → R be a non-negative continuous function such that Z 2 Z 3√3 12 2 f (x) dx < f (x) dx, 0 0 Rξ lim sup |ξ|→+∞ 0 f (x) dx ≤ 0. ξ2 Then, for each λ∈ i 213 (π 4 + π 2 + 1) 2(π 4 + π 2 + 1) h , , R2 R 3√3 π 4 0 f (x) dx π 4 0 f (x) dx the problem u0000 − u00 + u = f (u), in (0, 1), 00 u(0) = u(1) = 0 = u (0) = u00 (1) admits at least three weak solutions. The following result is a consequence of Theorem 3.6. Theorem 1.2. Let f : R → R be a non-negative continuous function such that Z 2 Z 3 211 f (x) dx < f (x) dx, 0 Z 210 f (x) dx < 27 0 0 Z 3 f (x) dx, 0 Then, for each n i 213 (π 4 + π 2 + 1) (π 4 + π 2 + 1) oh 2 220 min R 2 , , R 1024 λ∈ , R3 4 π π 4 0 f (x) dx f (x) dx f (x) dx 0 0 the problem u0000 − u00 − u = f (u), in (0, 1), 00 u(0) = u(1) = 0 = u (0) = u00 (1) admits at least three weak solutions. 2. Preliminaries We now state two critical point theorems established by Bonanno and coauthors [4, 5] which are the main tools for the proofs of our results. The first result has been obtained in [5] and it is a more precise version of Theorem 3.2 of [4]. The second one has been established in [4]. In the first one the coercivity of the functional Φ − λΨ is required, in the second one a suitable sign hypothesis is assumed. Theorem 2.1 ([5, Theorem 2.6]). Let X be a reflexive real Banach space; Φ : X → R be a sequentially weakly lower semicontinuous, coercive and continuously Gâteaux differentiable functional whose Gâteaux derivative admits a continuous inverse on X ∗ , Ψ : X → R be a sequentially weakly upper semicontinuous, continuously Gâteaux differentiable functional whose Gâteaux derivative is compact, such that Φ(0) = Ψ(0) = 0 . Assume that there exist r > 0 and x̄ ∈ X, with r < Φ(x̄) such that EJDE-2015/45 THREE SOLUTIONS FOR A BOUNDARY-VALUE PROBLEM 3 (i) supΦ(x)≤r Ψ(x) < rΨ(x̄)/Φ(x̄), (ii) for each λ in h i Φ(x̄) r , Λr := , Ψ(x̄) supΦ(x)≤r Ψ(x) the functional Φ − λΨ is coercive. Then, for each λ ∈ Λr the functional Φ − λΨ has at least three distinct critical points in X. Theorem 2.2 ([4, Theorem 3.2]). Let X be a reflexive real Banach space; Φ : X → R be a convex, coercive and continuously Gâteaux differentiable functional whose Gâteaux derivative admits a continuous inverse on X ∗ , Ψ : X → R be a continuously Gâteaux differentiable functional whose Gâteaux derivative is compact, such that inf Φ = Φ(0) = Ψ(0) = 0 . X Assume that there exist two positive constants r1 , r2 > 0 and x̄ ∈ X, with 2r1 < Φ(x̄) < r2 /2, such that (j) supΦ(x)≤r1 Ψ(x)/r1 < (2/3)Ψ(x̄)/Φ(x̄), (jj) supΦ(x)≤r2 Ψ(x)/r2 < (1/3)Ψ(x̄)/Φ(x̄), (jjj) for each λ in i 3 Φ(x̄) h r2 r1 , , min Λ∗r1 ,r2 := 2 Ψ(x̄) supΦ(x)≤r1 Ψ(x) 2 supΦ(x)≤r2 Ψ(x) and for every x1 , x2 ∈ X, which are local minima for the functional Φ−λΨ, and such that Ψ(x1 ) ≥ 0 and Ψ(x2 ) ≥ 0, one has inf t∈[0,1] Ψ(tx1 + (1 − t)x2 ) ≥ 0. Then, for each λ ∈ Λ∗r1 ,r2 the functional Φ − λΨ has at least three distinct critical points which lie in Φ−1 (] − ∞, r2 [). Let us introduce some notation which will be used later. Define H01 ([0, 1]) := u ∈ L2 ([0, 1]) : u0 ∈ L2 ([0, 1]), u(0) = u(1) = 0 , H 2 ([0, 1]) := u ∈ L2 ([0, 1]) : u0 , u00 ∈ L2 ([0, 1]) . Take X = H 2 ([0, 1]) ∩ H01 ([0, 1]) endowed with the usual norm Z 1 1/2 kuk := |u00 (x)|2 dx . 0 We recall the following Poincaré type inequalities (see, for instance, [8, Lemma 2.3]): 1 kuk2 , π2 1 ≤ 4 kuk2 π ku0 k2L2 ([0,1]) ≤ (2.1) kuk2L2 ([0,1]) (2.2) for all u ∈ X. For the norm in C 1 ([0, 1]), n o kuk∞ := max max |u(x)|, max |u0 (x)| , x∈[0,1] we have the following relation. x∈[0,1] 4 G. A. AFROUZI, S. SHOKOOH EJDE-2015/45 Proposition 2.3. Let u ∈ X. Then 1 kuk. (2.3) 2π Proof. Taking (2.1) into account, the conclusion follows from the well-known in equality kuk∞ ≤ 21 ku0 kL2 ([0,1]) . kuk∞ ≤ For an excellent overview of the most significant mathematical methods employed in this paper we refer to [7, 9]. Let g : R → R is a Lipschitz continuous function with the Lipschitz constant L > 0, i.e., |g(t1 ) − g(t2 )| ≤ L|t1 − t2 | for every t1 , t2 ∈ R, and g(0) = 0, h : [0, 1] × R → [0, +∞) is a bounded and continuous function with m := inf (x,t)∈[0,1]×R h(x, t) > 0, and f : [0, 1] × R → R be an L1 -Carathéodory function. We recall that f : [0, 1] × R → R is an L1 -Carathéodory function if (a) the mapping x 7→ f (x, ξ) is measurable for every ξ ∈ R; (b) the mapping ξ 7→ f (x, ξ) is continuous for almost every x ∈ [0, 1]; (c) for every ρ > 0 there exists a function lρ ∈ L1 ([0, 1]) such that sup |f (x, ξ)| ≤ lρ (x) |ξ|≤ρ for almost every x ∈ [0, 1]. Corresponding to f, g and h we introduce the functions F : [0, 1] × R → R, G : R → R and H : [0, 1] × R → [0, +∞), respectively, as follows Z t Z t F (x, t) := f (x, ξ) dξ, G(t) := − g(ξ) dξ, 0 0 Z tZ τ 1 dδ dτ H(x, t) := 0 0 h(x, δ) for all x ∈ [0, 1] and t ∈ R. In the following, we let M := sup(x,t)∈[0,1]×R h(x, t) and suppose that the Lipschitz constant L of the function g satisfies 0 < L < π 4 . We say that a function u ∈ X is a weak solution of (1.1) if Z 1 Z 1 Z u0 (x) 1 00 00 u (x)v (x) dx + dτ v 0 (x) dx h(x, τ ) 0 0 0 Z 1 Z 1 −λ f (x, u(x))v(x) dx − g(u(x))v(x) dx = 0 0 0 holds for all v ∈ X. 3. Main results Put π4 − L π 2 + m(π 4 + L) , B := , 2π 4 2mπ 4 and suppose that B ≤ 4Aπ 2 . We formulate our main results as follows. A := Theorem √ 3.1. Assume that there exist two positive constants c, d, satisfying c < 32d/(3 3π), such that EJDE-2015/45 THREE SOLUTIONS FOR A BOUNDARY-VALUE PROBLEM 5 (A1) F (x, t) ≥ 0 for all (x, t) ∈ ([0, 3/8] ∪ [5/8, 1]) × [0, d]; (A2) R 5/8 R1 max|t|≤c F (x, t) dx 27 3/8 F (x, d) dx 0 < ; c2 4096 d2 (A3) R1 supx∈[0,1] F (x, ξ) π 4 A 0 max|t|≤c F (x, t) dx lim sup ≤ . ξ2 B c2 |ξ|→+∞ Then, for every λ in i h 4096Bd2 Bc2 Λ := , R1 , R 5/8 27 3/8 F (x, d) dx 0 max|t|≤c F (x, t) dx problem (1.1) has at least three distinct weak solutions. Proof. Fix λ as in the conclusion. Our aim is to apply Theorem 2.1 to our problem. To this end, for every u ∈ X, we introduce the functionals Φ, Ψ : X → R by setting Z 1 Z 1 1 Φ(u) := kuk2 + H(x, u0 (x)) dx + G(u(x)) dx, 2 0 0 Z 1 Ψ(u) := F (x, u(x)) dx, 0 and putting Iλ (u) := Φ(u) − λΨ(u) ∀ u ∈ X. Note that the weak solutions of (1.1) are exactly the critical points of Iλ . The functionals Φ, Ψ satisfy the regularity assumptions of Theorem 2.1. Indeed, by standard arguments, we have that Φ is Gâteaux differentiable and sequentially weakly lower semicontinuous and its Gâteaux derivative is the functional Φ0 (u) ∈ X ∗ , given by Z 1 Z 1 Z 1 Z u0 (x) 1 dτ v 0 (x) dx− Φ0 (u)(v) = g(u(x))v(x) dx u00 (x)v 00 (x) dx+ h(x, τ ) 0 0 0 0 for any v ∈ X. Furthermore, the differential Φ0 : X → X ∗ is a Lipschitzian operator. Indeed, taking (2.1) and (2.2) into account, for any u, v ∈ X, there holds kΦ0 (u) − Φ0 (v)kX ∗ = sup |(Φ0 (u) − Φ0 (v), w)| kwk≤1 1 Z |u00 (x) − v 00 (x)||w00 (x)| dx ≤ supkwk≤1 0 Z 1 + sup kwk≤1 0 Z v 0 (x) u0 (x) 1 dτ |w0 (x)| dx h(x, τ ) 1 |g(u(x)) − g(v(x))||w(x)| dx + sup kwk≤1 Z 0 ≤ ku − vk + 1 sup ku0 − v 0 kL2 (0,1) kw0 kL2 (0,1) m kwk≤1 + L sup ku − vkL2 (0,1) kwkL2 (0,1) kwk≤1 ≤ 1+ 1 L + 4 ku − vk = 2Bku − vk. 2 mπ π 6 G. A. AFROUZI, S. SHOKOOH EJDE-2015/45 Recalling that g is Lipschitz continuous and h is bounded away from zero, the claim is true. In particular, we derive that Φ is continuously differentiable. Also, for any u, v ∈ X, we have Z 1 Z v0 (x) 1 (Φ0 (u) − Φ0 (v), u − v) = ku − vk2 + dτ (u0 (x) − v 0 (x)) dx 0 u0 (x) h(x, τ ) Z 1 (g(u(x)) − g(v(x)))(u(x) − v(x)) dx − 0 1 0 ku − v 0 k2L2 (0,1) − Lku − vk2L2 (0,1) M L ≥ ku − vk2 − 4 ku − vk2 = 2Aku − vk2 . π By the assumption L < π 4 , it turns out that Φ0 is a strongly monotone operator. So, by applying Minty-Browder theorem [12, Theorem 26.A], Φ0 : X → X ∗ admits a Lipschitz continuous inverse. On the other hand, the fact that X is compactly embedded into C 0 ([0, 1]) implies that the functional Ψ is well defined, continuously Gâteaux differentiable and with compact derivative, whose Gâteaux derivative is given by Z 1 0 Ψ (u)(v) = f (x, u(x))v(x) dx ≥ ku − vk2 + 0 for any v ∈ X. Since g is Lipschitz continuous and satisfies g(0) = 0, while h is bounded away from zero, the inequalities (2.1) and (2.2) yield for any u ∈ X the estimate Akuk2 ≤ Φ(u) ≤ Bkuk2 . (3.1) 2 We will verify (i) and (ii) of Theorem 2.1. Put r = Bc . Taking (2.3) into account, for every u ∈ X such that Φ(u) ≤ r, one has maxx∈[0,1] |u(x)| ≤ c. Consequently, Z 1 sup Ψ(u) ≤ max F (x, t) dx; Φ(u)≤r 0 |t|≤c that is, supΦ(u)≤r Ψ(u) ≤ r R1 0 max|t|≤c F (x, t) dx . Bc2 Hence, supΦ(u)≤r Ψ(u) 1 < . r λ (3.2) Put 64d 3 2 − 9 (x − 4 x), w(x) = d, 64d 2 5 − 9 (x − 4 x + 41 ), x ∈ [0, 83 [, x ∈ [ 38 , 58 ], x ∈] 58 , 1]. It is easy to verify that w ∈ X and, in particular, 4096 2 kwk2 = d . 27 So, taking (3.1) into account, we deduce 4096 2 4096 2 Ad ≤ Φ(w) ≤ Bd . 27 27 EJDE-2015/45 THREE SOLUTIONS FOR A BOUNDARY-VALUE PROBLEM 7 Hence, from c < 3√323π d and B ≤ 4Aπ 2 , we obtain r < Φ(w). Since 0 ≤ w(x) ≤ d for each x ∈ [0, 1], assumption (A1) ensures that Z 1 Z 3/8 F (x, w(x)) dx ≥ 0, F (x, w(x)) dx + 5/8 0 and so 5/8 Z Ψ(w) ≥ F (x, d) dx. 3/8 Therefore, we obtain Ψ(w) 27 ≥ Φ(w) 4096 R 5/8 3/8 F (x, d) dx Bd2 > 1 . λ (3.3) Therefore, from (3.2) and (3.3), condition (i) of Theorem 2.1 is fulfilled. Now, to prove the coercivity of the functional Iλ . By (A3), we have supx∈[0,1] F (x, ξ) π4 A < . ξ2 λ |ξ|→+∞ lim sup So, we can fix ε > 0 satisfying supx∈[0,1] F (x, ξ) π4 A < ε < . ξ2 λ |ξ|→+∞ lim sup Then, there exists a positive constant θ such that F (x, t) ≤ ε|t|2 + θ ∀x ∈ [0, 1], ∀t ∈ R . Taking into account (2.2) and (3.1), we have λε kuk2 − λθ π4 for all u ∈ X. So, the functional Iλ is coercive. Now, the conclusion of Theorem 2.1 can be used. It follows that for every i Φ(w) h r λ∈Λ⊆ , , Ψ(w) supΦ(u)≤r Ψ(u) Iλ (u) = Φ(u) − λΨ(u) ≥ Akuk2 − λεkuk2L2 [0,1] − λθ ≥ A − the functional Iλ has at least three distinct critical points in X, which are the weak solutions of the problem (1.1). This completes the proof. Now, we present a consequence of Theorem 3.1. Corollary 3.2. Let α ∈ L1 ([0, 1]) be a non-negative and non-zero function and let R1 R 5/8 γ : R → R be a continuous function. Put α0 := 3/8 α(x)dx, kαk1 := 0 α(x) dx Rt and Γ(t) = 0 γ(ξ)dξ for all t ∈ R, and assume that there exist two positive constants c, d, with c < 3√323π d, such that (A1’) Γ(t) ≥ 0 for all t ∈ [0, d]; (A2’) max|t|≤c Γ(t) 27 α0 Γ(d) < ; c2 4096 kαk1 d2 (A3’) lim sup|ξ|→+∞ Γ(ξ)/ξ 2 ≤ 0. 8 G. A. AFROUZI, S. SHOKOOH EJDE-2015/45 Then, for every i 4096 Bd2 h Bc2 , , 27 α0 Γ(d) kαk1 max|t|≤c Γ(t) λ∈ the problem u0000 h(x, u0 ) − u00 = [λα(x)γ(u) + g(u)]h(x, u0 ), 00 in (0, 1), 00 u(0) = u(1) = 0 = u (0) = u (1) (3.4) has at least three weak solutions. The proof of the above corollary follows from Theorem 3.1 by choosing f (x, t) := α(x)γ(t) for all (x, t) ∈ [0, 1] × R. Remark 3.3. Clearly, if γ is non-negative then assumption (A1’) is satisfied and (A2’) becomes Γ(c) 27 α0 Γ(d) < . c2 4096 kαk1 d2 Remark 3.4. Theorem 1.1 in the introduction is an immediate consequence of √ Corollary 3.2, on choosing g(u) = −u, h ≡ 1, c = 2 and d = 3 3. The following lemma will be crucial in our arguments. Lemma 3.5. Assume that f (x, t) ≥ 0 for all (x, t) ∈ [0, 1] × R. If u is a weak solution of (1.1), then u(x) ≥ 0 for all x ∈ [0, 1]. Proof. Arguing by contradiction, if we assume that u is negative at a point of [0, 1], the set Ω := {x ∈ [0, 1] : u(x) < 0}, is non-empty and open. Let us consider v̄ := min{u, 0}, one has, v̄ ∈ X. So, taking into account that u is a weak solution and by choosing v = v̄, from our assumptions, one has Z 0≥λ f (x, u(x))u(x) dx Ω Z 00 2 |u (x)| dx + = Ω 4 Z Z Ω 0 u0 (x) 1 dτ u0 (x) dx − h(x, τ ) Z g(u(x))u(x) dx Ω π −L kuk2H 2 (Ω)∩H 1 (Ω) . 0 π4 Therefore, kukH 2 (Ω)∩H01 (Ω) = 0 which is absurd. Hence, the conclusion is achieved. ≥ Our other main result is as follows. Theorem 3.6. √Assume that there exist three positive constants c1 , c2 , d, satisfying √ 3π 3 √3 3 √ c < d < 64 c , such that 1 16 2 2 2 (B1) f (x, t) ≥ 0 for all (x, t) ∈ [0, 1] × [0, c2 ]; (B2) R 5/8 R1 F (x, d) dx F (x, c ) dx 9 1 3/8 0 < ; 2 c1 2048 d2 EJDE-2015/45 THREE SOLUTIONS FOR A BOUNDARY-VALUE PROBLEM 9 (B3) R1 0 R 5/8 F (x, c2 ) dx 9 < 2 c2 4096 3/8 F (x, d) dx d2 . Let Λ0 := Bd2 i 2048 9 R 5/8 3/8 F (x, d) dx n , B min R 1 0 c21 oh c22 , R1 . F (x, c1 ) dx 2 0 F (x, c2 ) dx 0 Then, for every λ ∈ Λ the problem (1.1) has at least three weak solutions ui , i = 1, 2, 3, such that 0 < kui k∞ ≤ c2 . Proof. Without loss of generality, we can assume f (x, t) ≥ 0 for all (x, t) ∈ [0, 1]×R. Fix λ as in the conclusion and take X, Φ and Ψ as in the proof of Theorem 3.1. Put w as in Theorem 3.1, r1 = Bc21 and r2 = Bc22 . Therefore, one has 2r1 < Φ(w) < r22 and we have Z 1 1 1 1 sup Ψ(u) ≤ F (x, c1 ) dx < 2 r1 Φ(u)<r1 Bc1 0 λ R 5/8 9 3/8 F (x, d) dx < 2048 Bd2 2 Ψ(w) , ≤ 3 Φ(w) and 2 r2 sup Ψ(u) ≤ Φ(u)<r2 2 Bc22 Z 1 F (x, c2 ) dx < 0 1 λ R 5/8 9 3/8 F (x, d) dx 2048 Bd2 2 Ψ(w) ≤ . 3 Φ(w) < So, conditions (j) and (jj) of Theorem 2.2 are satisfied. Finally, let u1 and u2 be two local minima for Φ − λΨ. Then, u1 and u2 are critical points for Φ − λΨ, and so, they are weak solutions for the problem (1.1). Hence, owing to Lemma 3.5, we obtain u1 (x) ≥ 0 and u2 (x) ≥ 0 for all x ∈ [0, 1]. So, one has Ψ(su1 + (1 − s)u2 ) ≥ 0 for all s ∈ [0, 1]. From Theorem 2.2 the functional Φ−λΨ has at least three distinct critical points which are weak solutions of (1.1). This complete the proof. Now, we present a consequence of Theorem 3.6. Corollary 3.7. Let α ∈ L1 ([0, 1]) be such that α(x) ≥ 0 a.e. x ∈ [0, 1], α 6≡ 0, and R1 R 5/8 let γ : R → R be a continuous function. Put α0 := 3/8 α(x)dx, kαk1 := 0 α(x) dx Rt and Γ(t) = 0 γ(ξ)dξ for all t ∈ R, and assume that there exist three positive √ √ 3 √3 constants c1 , c2 , d, with 316√3π2 c1 < d < 64 c , such that 2 2 (B1’) γ(t) ≥ 0 for all t ∈ [0, c2 ]; (B2’) Γ(c1 ) 9 α0 Γ(d) < ; c21 2048 kαk1 d2 10 G. A. AFROUZI, S. SHOKOOH EJDE-2015/45 (B3’) Γ(c2 ) 9 α0 Γ(d) . < c22 4096 kαk1 d2 Then, for every λ∈ oh i 2048 Bd2 n c21 c22 , , B min , 9 α0 Γ(d) kαk1 Γ(c1 ) 2kαk1 Γ(c2 ) the problem (3.4) has at least three weak solutions ui , i = 1, 2, 3, such that 0 < kui k∞ ≤ c2 . The proof of the above corollary follows from Theorem 3.6 by choosing f (x, t) := α(x)γ(t) for all (x, t) ∈ [0, 1] × R. Remark 3.8. Theorem 1.2 in the introduction is an immediate consequence of Corollary 3.7, on choosing g(u) = u, h ≡ 1, c1 = 2, c2 = 210 , and d = 3. Finally, we present the following examples to illustrate our results. Example 3.9. Consider the following problem u0000 − u00 (2 + x + cos u0 ) + u = λf (u), 00 in (0, 1), 00 u(0) = u(1) = 0 = u (0) = u (1), (3.5) where f : R → R is defined by −10 2 f (t) = 2−10 t4 20 −2 2 t if |t| ≤ 1, if 1 < |t| ≤ 32, if |t| > 32. Here, g(t) = −t and h(x, t) = (2 + x + cos t)−1 for all x ∈ [0, 1] and t ∈ R. It is easy to verify that (A2’) and (A3’) are satisfied with c = 1 and d = 32. From Corollary 3.2, for each parameter i 48(π 4 + 4π 2 + 1) 512(π 4 + 4π 2 + 1) h λ∈ , , π4 π4 problem (3.5) admits at least three weak solutions. Example 3.10. Consider the problem u0000 − u00 (3 + sin u0 ) − 2u = λf (u), 00 in (0, 1), 00 u(0) = u(1) = 0 = u (0) = u (1), (3.6) where f : R → R is defined by −20 2 4 f (t) = t −2 t if |t| ≤ 2−5 , if 2−5 < |t| ≤ 1, if |t| > 1. Here, g(t) = 2t and h(x, t) = (3 + sin t)−1 for all x ∈ [0, 1] and t ∈ R. It is easy to verify that (B2’) and (B3’) are satisfied with c1 = 2−5 , d = 1 and c2 = 210 . From Corollary 3.7, for each parameter i 2276(π 4 + 4π 2 + 2) 214 (π 4 + 4π 2 + 2) h , , λ∈ π4 π4 problem (3.6) admits at least three weak solutions ui , i = 1, 2, 3, such that 0 < kui k∞ ≤ 1024. EJDE-2015/45 THREE SOLUTIONS FOR A BOUNDARY-VALUE PROBLEM 11 References [1] G. Afrouzi, A. Hadjian, V. D. Rădulescu; Variational approach to fourth-order impulsive differential equations with two control parameters, Results in Mathematics, 65 (2014), 371384. [2] G. A. Afrouzi, S. Heidarkhani, D. O’Regan; Existence of three solutions for a doubly eigenvalue fourth-order boundary value problem, Taiwanese J. Math., 15 (2011), 201-210. [3] G. Afrouzi, M. Mirzapour, V. D. Rădulescu; Nonlocal fourth-order Kirchhoff systems with variable growth: low and high energy solutions, Collectanea Mathematica, in press (DOI: 10.1007/s13348-014-0131-x). [4] G. Bonanno, P. Candito; Non-differentiable functionals and applications to elliptic problems with discontinuous nonlinearities, J. Differential Equations, 244 (2008), 3031-3059. [5] G. Bonanno, S. A. Marano; On the structure of the critical set of non-differentiable functionals with a weak compactness condition, Appl. Anal., 89 (2010), 1-10. [6] G. Chai; Existence of positive solutions for fourth-order boundary value problem with variable parameters, Math. Comput. Modelling, 66 (2007), 870-880. [7] P. G. Ciarlet; Linear and Nonlinear Functional Analysis with Applications, Society for Industrial and Applied Mathematics (SIAM), Philadelphia, 2013. [8] L. A. Peletier, W. C. Troy, R. C. A. M. Van der Vorst; Stationary solutions of a fourth order nonlinear diffusion equation, (Russian) Translated from the English by V. V. Kurt. Differentsialnye Uravneniya 31 (1995), 327-337. English translation in Differential Equations 31 (1995), 301-314. [9] V. D. Rădulescu; Qualitative Analysis of Nonlinear Elliptic Partial Differential Equations: Monotonicity, Analytic, and Variational Methods, Contemporary Mathematics and Its Applications, Vol. 6, Hindawi Publ. Corp., 2008. [10] B. Ricceri; A three critical points theorem revisited, Nonlinear Anal., 70 (2009), 3084-3089. [11] V. Shanthi, N. Ramanujam; A numerical method for boundary value problems for singularly perturbed fourth-order ordinary differential equations, Appl. Math. Comput., 129 (2002). 269-294. [12] E. Zeidler; Nonlinear Functional Analysis and its Applications, vol. II/B and III, BerlinHeidelberg-New York, 1990. Ghasem A. Afrouzi Department of Mathematics, Faculty of Mathematical Sciences, University of Mazandaran, Babolsar, Iran E-mail address: afrouzi@umz.ac.ir Saeid Shokooh Department of Mathematics, Faculty of Mathematical Sciences, University of Mazandaran, Babolsar, Iran E-mail address: saeid.shokooh@stu.umz.ac.ir