Electronic Journal of Differential Equations, Vol. 2013 (2013), No. 225, pp. 1–9. ISSN: 1072-6691. URL: http://ejde.math.txstate.edu or http://ejde.math.unt.edu ftp ejde.math.txstate.edu WEAK HETEROCLINIC SOLUTIONS OF ANISOTROPIC DIFFERENCE EQUATIONS WITH VARIABLE EXPONENT ABOUDRAMANE GUIRO, BLAISE KONE, STANISLAS OUARO Abstract. In this article, we prove the existence of heteroclinic solutions for a family of anisotropic difference equations. The proof of the main result is based on a minimization method, a change of variables and a discrete Hölder type inequality. 1. Introduction In this article we study the existence of heteroclinic solutions for the nonlinear discrete anisotropic problem −∆(a(k − 1, ∆u(k − 1))) + g(k, u(k)) = f (k), k ∈ Z∗ (1.1) u(0) = 0, lim u(k) = −1, lim u(k) = 1, k→−∞ k→+∞ where ∆u(k) = u(k + 1) − u(k) is the forward difference operator. The study of heteroclinic connections for boundary value problems had a certain impulse in recent years, motivated by applications in various biological, physical and chemical models, such has phase-transition, physical processes in which the variable transits from an unstable equilibrium to a stable one, or front-propagation in reaction-diffusion equations. Indeed, heteroclinic solutions are often called transitional solutions (see [2, 6] and the references therein). In this article, we show that the solvability of (1.1) is connected to the behavior of g(k, s) as k ∈ Z+ and as k ∈ Z− . Problem (1.1) involves variable exponents due to their use in image restoration (see [3]), in electrorheological and thermorheological fluids dynamic (see [4, 7, 8]). The paper is organized as follows: In section 2, we introduce hypotheses on f , g and a, we define the functional spaces and some of their useful properties and in section 3, we prove the existence of heteroclinic solutions of (1.1). 2. Auxiliary results For the rest of this article, we will use the notation: p− = inf p(k). p+ = sup p(k), k∈Z k∈Z 2000 Mathematics Subject Classification. 47A75, 35B38, 35P30, 34L05, 34L30. Key words and phrases. Difference equations; heteroclinic solutions; anisotropic problem; discrete Hölder type inequality. c 2013 Texas State University - San Marcos. Submitted December 11, 2012. Published October 11, 2013. 1 2 A. GUIRO, B. KONE, S. OUARO EJDE-2013/225 We assume that p(.) : Z → (1, +∞) and 1 < p− ≤ p(.) < p+ < +∞. (2.1) We introduce the spaces: l1 = {u : Z → R, kukl1 := X |u(k)| < ∞}, k∈Z l01 = {u : Z → R; u(0) = 0 and kukl01 := X |u(k)| < ∞}, k∈Z p(.) l0 = {u : Z → R; u(0) = 0 and ρp(.) (u) := X |u(k)|p(k) < ∞}, k∈Z p(.) l0,+ X + = {u : Z → R; u(0) = 0 and ρp+ (.) (u) := |u(k)|p(k) < ∞}, k∈Z+ p(.) X l0,− = {u : Z− → R; u(0) = 0 and ρp− (.) (u) := |u(k)|p(k) < ∞}, k∈Z− 1,p(.) W0,+ = {u : Z+ → R; u(0) = 0 and ρ1,p+ (.) (u) := X |u(k)|p(k) k∈Z+ + X |∆u(k)|p(k) < ∞} k∈Z+ p(.) p(.) = {u : Z+ → R; u ∈ l+ , ∆u(k) ∈ l+ 1,p(.) W0,− and u(0) = 0} X = {u : Z− → R; u(0) = 0 and ρ1,p− (.) (u) := |u(k)|p(k) k∈Z− + X |∆u(k)|p(k) < ∞} k∈Z− p(.) p(.) = {u : Z− → R; u ∈ l− , ∆u(k) ∈ l− and u(0) = 0}. p(.) On l0,+ we introduce the Luxemburg norm X u(k) kukp+ (.) := inf λ > 0 : | |p(k) ≤ 1 λ + k∈Z and we deduce that X u(k) X ∆u(k) kuk1,p+ (.) := inf λ > 0; | |p(k) + | |p(k) ≤ 1 λ λ + + k∈Z k∈Z = kukp+ (.) + k∆ukp+ (.) 1,p(.) p(.) is a norm on the space W0,+ . We replace Z+ by Z− to get the norms on l0,− and 1,p(.) W0,− denoted respectively k · kp− (.) and k · k1,p− (.) . For the data f , g and a, we assume the following: a(k, .) : R → R for all k ∈ Z and there exists a mapping A : Z × R → R such that a(k, ξ) = ∂ A(k, ξ) for all k ∈ Z and A(k, 0) = 0 for all k ∈ Z. ∂ξ |ξ|p(k) ≤ a(k, ξ)ξ ≤ p(k)A(k, ξ) ∀k ∈ Z and ξ ∈ R. (2.2) (2.3) EJDE-2013/225 WEAK HETEROCLINIC SOLUTIONS 3 There exists a positive constant C1 such that |a(k, ξ)| ≤ C1 (j(k) + |ξ|p(k)−1 ), 0 for all k ∈ Z and ξ ∈ R where j ∈ lp (.) with 1 p(k) + 1 p0 (k) (2.4) = 1. f ∈ l1 , p(k)−2 g(k, t) = |t − 1| (2.5) p(k)−2 (t − 1)χZ+ (k) + |t + 1| (t + 1)χZ− (k), (2.6) where χA (k) = 1 if k ∈ A and χA (k) = 0 if k ∈ / A. p(.) p(.) p(.) p(.) 1,p(.) Remark 2.1. Note that l0,+ ⊂ l0 , l0,− ⊂ l0 , W0,+ 1,p(.) 1,p(.) ⊂ W0 and W0,− ⊂ 1,p(.) W0 . p(.) p(.) p(.) If u ∈ l0,+ (or u ∈ l0,− or u ∈ l0 ) then limk→+∞ u(k) = 0 (or limk→−∞ u(k) = 0 P p(.) or lim|k|→+∞ u(k) = 0). Indeed, for instance, if u ∈ l0,+ then k∈Z+ |u(k)|p(k) < ∞. Let X X X |u(k)|p(k) = |u(k)|p(k) + |u(k)|p(k) , k∈Z+ k∈S1 k∈S2 where S1 = {k ∈ Z+ ; |u(k)| < 1} and S2 = {k ∈ Z+ ; |u(k)| ≥ 1}. The set S2 is p(.) necessarily finite, and |u(k)| < ∞ for any k ∈ S2 since u ∈ l0,+ . We also have that X X + |u(k)|p ≤ |u(k)|p(k) , k∈Z+ k∈S1 + then k∈S1 |u(k)|p < ∞. As S2 is a finite set then implies that X + |u(k)|p < ∞. P P k∈S2 + |u(k)|p < ∞, which k∈Z+ Thus, limk→+∞ u(k) = 0. We now give useful properties of the spaces defined above which are similar to those in [5]. p(.) Proposition 2.2. Assume that (2.1) is fulfilled. Then l01 ⊂ l0 . Proposition 2.3. Under conditions (2.1), ρp+ (.) satisfies p(.) (a) ρp+ (.) (u + v) ≤ 2p+ (ρp+ (.) (u) + ρp+ (.) (v)), for all u, v ∈ l0,+ . p(.) (b) For u ∈ l0,+ , if λ > 1 we have − + ρp+ (.) (u) ≤ λρp+ (.) (u) ≤ λp ρp+ (.) (u) ≤ ρp+ (.) (λu) ≤ λp ρp+ (.) (u) and if 0 < λ < 1, we have − + λp ρp+ (.) (u) ≤ ρp+ (.) (λu) ≤ λp ρp+ (.) (u) ≤ λρp+ (.) (u) ≤ ρp+ (.) (u). p(.) (c) For every fixed u ∈ l0,+ \ {0}, ρp+ (.) (λu) is a continuous convex even function in λ and it increases strictly when λ ∈ [0, ∞). p(.) Proposition 2.4. Let u ∈ l0,+ \ {0}, then kukp+ (.) = a if and only if ρp+ (.) ( ua ) = 1. p(.) Proposition 2.5. If u ∈ l0,+ and p+ < +∞, then the following properties hold: (1) kukp+ (.) < 1 (= 1; > 1) if and only if ρp+ (.) (u) < 1 (= 1; > 1); − (2) kukp+ (.) > 1 implies kukpp+ (.) ≤ ρp+ (.) (u) ≤ kukp+ p+ (.) ; 4 A. GUIRO, B. KONE, S. OUARO EJDE-2013/225 + (3) kukp+ (.) < 1 implies kukpp+ (.) ≤ ρp+ (.) (u) ≤ kukp− p+ (.) ; (4) kun kp+ (.) → 0 if and only if ρp+ (.) (un ) → 0 as n → +∞. 1,p(.) Proposition 2.6. Let u ∈ W0,+ ρ1,p+ (.) (u/a) = 1. \ {0}. Then kuk1,p+ (.) = a if and only if 1,p(.) Proposition 2.7. If u ∈ W0,+ and p+ < +∞, then the following properties hold: (1) kuk1,p+ (.) < 1 (= 1; > 1) if and only if ρ1,p+ (.) (u) < 1 (= 1; > 1); − + + − (2) kuk1,p+ (.) > 1 implies kukp1,p+ (.) ≤ ρ1,p+ (.) (u) ≤ kukp1,p+ (.) ; (3) kuk1,p+ (.) < 1 implies kukp1,p+ (.) ≤ ρ1,p+ (.) (u) ≤ kukp1,p+ (.) ; (4) kun k1,p+ (.) → 0 if and only if ρ1,p+ (.) (un ) → 0 as n → +∞. p(.) Theorem 2.8 (Discrete Hölder type inequality). Let u ∈ l+ 1 1 + q(k) = 1 for all k ∈ Z+, then that p(k) +∞ X |uv| ≤ ( k=0 q(.) and v ∈ l+ be such 1 1 + − )kukp+ (.) kvkq+ (.) . p− q p(.) Remark 2.9. All the properties above hold for the spaces lp(.) , l− 1,p(.) and W0,− . 3. Existence of weak heteroclinic solutions In this section, we study the existence of weak heteroclinic solutions of problem (1.1). Definition 3.1. A weak heteroclinic solution of (1.1) is a function u : Z → R such that X X X a(k − 1, ∆u(k − 1))∆v(k − 1) + g(k, u(k))v(k) = f (k)v(k), (3.1) k∈Z k∈Z k∈Z for any v : Z → R, with u(0) = 0, limk→+∞ u(k) = 1 and limk→−∞ u(k) = −1. Theorem 3.2. Assume that (2.1)–(2.6) hold. Then, there exists at least one weak heteroclinic solution of (1.1). To prove Theorem 3.2, we first prove that the problem −∆(a(k − 1, ∆u(k − 1))) + |u(k)|p(k)−2 u(k) = f (k), u(0) = 0, k ∈ Z+ ∗ lim u(k) = 0, (3.2) k→+∞ admits a weak solution in the following sense. 1,p(.) Definition 3.3. A weak solution of (3.2) is a function u ∈ W0,+ +∞ X a(k − 1, ∆u(k − 1))∆v(k − 1) + k=1 +∞ X k=1 |u(k)|p(k)−2 u(k)v(k) = +∞ X such that f (k)v(k), (3.3) k=1 1,p(.) for any v ∈ W0,+ . We have the following result. Theorem 3.4. Assume that (2.1)–(2.5) hold. Then, there exists at least one weak solution of problem (3.2). EJDE-2013/225 WEAK HETEROCLINIC SOLUTIONS 5 1,p(.) The energy functional corresponding to problem (3.2) is J : W0,+ → R defined by J(u) = +∞ X A(k − 1, ∆u(k − 1)) + k=1 +∞ +∞ X X 1 |u(k)|p(k) − f (k)u(k). p(k) (3.4) k=1 k=1 We first present some basic properties of J. 1,p(.) Proposition 3.5. The functional J is well-defined on the space W0,+ class C 1 1,p(.) (W0,+ , R), and is of with the derivative given by hJ 0 (u), vi = +∞ X a(k − 1, ∆u(k − 1))∆v(k − 1) k=1 +∞ X + |u(k)| p(k)−2 u(k)v(k) − k=1 +∞ X (3.5) f (k)v(k), k=1 1,p(.) for all u, v ∈ W0,+ . Proof. We denote I(u) = +∞ X A(k −1, ∆u(k −1)), +∞ X 1 |u(k)|p(k) , L(u) = p(k) Λ(u) = k=1 k=1 +∞ X f (k)u(k). k=1 Using Young inequality, from assumptions (2.2) and (2.4) it follows that |I(u)| = | ≤ ≤ ≤ +∞ X A(k − 1, ∆u(k − 1))| k=1 +∞ X |A(k − 1, ∆u(k − 1))| k=1 +∞ X k=1 +∞ X C1 j(k − 1) + 1 |∆u(k − 1)|p(k−1)−1 |∆u(k − 1)| p(k − 1) C1 j(k − 1)|∆u(k − 1)| + k=1 +∞ X k=1 C1 |∆u(k − 1)|p(k−1) < ∞, p(k − 1) +∞ 1 X |u(k)|p(k) < ∞, |L(u)| ≤ p− k=1 +∞ +∞ X X |Λ(u)| = f (k)u(k) ≤ |f (k)||u(k)| < ∞. k=1 k=1 Therefore, J is well-defined. 1,p(.) 1,p(.) Clearly I, L and Λ are in C 1 (W0,+ , R). Let us now choose u, v ∈ W0,+ . We have I(u + δv) − I(u) L(u + δv) − L(u) hI 0 (u), vi = lim+ , hL0 (u), vi = lim+ , δ δ δ→0 δ→0 Λ(u + δv) − Λ(u) hΛ0 (u), vi = lim+ . δ δ→0 6 A. GUIRO, B. KONE, S. OUARO Let us denote gδ = ity, +∞ X k=1 EJDE-2013/225 A(k−1,∆u(k−1)+δ∆v(k−1))−A(k−1,∆u(k−1)) . δ +∞ +∞ k=1 k=1 Using Young inequal- 1X 1X |A(k−1, ∆u(k−1)+δ∆v(k−1))|+ |A(k−1, ∆u(k−1))| < +∞. |gδ | ≤ δ δ Thus, I(u + δv) − I(u) δ δ→0 +∞ X A(k − 1, ∆u(k − 1) + δ∆v(k − 1)) − A(k − 1, ∆u(k − 1)) = lim+ δ δ→0 lim+ k=1 = = +∞ X k=1 +∞ X lim+ δ→0 A(k − 1, ∆u(k − 1) + δ∆v(k − 1)) − A(k − 1, ∆u(k − 1)) δ a(k − 1, ∆u(k − 1))∆v(k − 1). k=1 By the same method, we deduce that +∞ lim δ→0+ X |u(k) + δv(k)|p(k) − |u(k)|p(k) L(u + δv) − L(u) = lim δ p(k)δ δ→0+ k=1 = = +∞ X k=1 +∞ X lim δ→0+ |u(k) + δv(k)|p(k) − |u(k)|p(k) p(k)δ |u(k)|p(k)−2 u(k)v(k) k=1 and +∞ lim δ→0+ X f (k)(u(k) + δv(k)) − f (k)u(k) Λ(u + δv) − Λ(u) = lim δ δ δ→0+ k=1 = = +∞ X k=1 +∞ X lim δ→0+ f (k)(u(k) + δv(k)) − f (k)u(k) δ f (k)v(k) k=1 Lemma 3.6. The functional I is weakly lower semi-continuous. Proof. From (2.2), I is convex with respect to the second variable. Thus, by [1, corollary III.8], it is sufficient to show that I is lower semi-continuous. For this, we 1,p(.) 1,p(.) fix u ∈ W0,+ and > 0. Since I is convex, we deduce that for any v ∈ W0,+ , I(v) ≥ I(u) + hI 0 (u), v − ui ≥ I(u) + +∞ X k=1 a(k − 1, ∆u(k − 1))(∆v(k − 1) − ∆u(k − 1)) EJDE-2013/225 WEAK HETEROCLINIC SOLUTIONS ≥ I(u) − C( 7 1 1 + 0− )kgkp0+ (.) k∆(u − v)kp+ (.) , − p p with g(k) = j(k) + |∆u(k)|p(k)−1 ≥ I(u) − K ku − vkp+ (.) + k∆(u − v)kp+ (.) ≥ I(u) − Kku − vk1,p+ (.) ≥ I(u) − , 1,p(.) for all v ∈ W0,+ with ku − vk1,p+ (.) < δ = /K. Hence, we conclude that I is weakly lower semi-continuous. Proposition 3.7. The functional J is bounded from below, coercive and weakly lower semi-continuous. Proof. By Lemma 3.6, J is weakly lower semi-continuous. We shall only prove the coerciveness of the energy functional J and its boundedness from below. J(u) = ≥ +∞ X k=1 +∞ X k=1 A(k − 1, ∆u(k − 1)) + +∞ +∞ X X 1 |u(k)|p(k) − f (k)u(k) p(k) k=1 k=1 +∞ +∞ X X 1 1 |∆u(k − 1)|p(k−1) + |u(k)|p(k) − |f (k)u(k)| p(k − 1) p(k) k=1 k=1 +∞ +∞ +∞ X X 1 X p(s) p(k) |∆u(s)| + |u(k)| )− |f (k)||u(k)| ≥ +( p s=1 k=1 k=1 1 ≥ + ρ1,p+ (.) (u) − c0 kf kp0+ (.) kukp+ (.) p 1 ≥ + ρ1,p+ (.) (u) − Kkuk1,p+ (.) . p To prove the coerciveness of J, we may assume that kuk1,p+ (.) > 1 and we get from the above inequality that J(u) ≥ − 1 kukp1,p+ (.) − Kkuk1,p+ (.) . p+ Thus, J(u) → +∞ as kuk1,p+ (.) → +∞. As J(u) → +∞ when kuk1,p+ (.) → +∞, then for kuk1,p+ (.) > 1, there exists c ∈ R such that J(u) ≥ c. For kuk1,p+ (.) ≤ 1, we have J(u) ≥ 1 ρ1,p+ (.) (u) − Kkuk1,p+ (.) ≥ −Kkuk1,p+ (.) ≥ −K > −∞. p+ Thus J is bounded below. We can now give the proof of the main result. Proof of Theorem 3.4. By Proposition 3.7, J has a minimizer which is a weak solution of (3.2) 8 A. GUIRO, B. KONE, S. OUARO EJDE-2013/225 Now, we consider the problem −∆(a(k − 1, ∆u(k − 1))) + |u(k)|p(k)−2 u(k) = f (k), u(0) = 0, k ∈ Z− (3.6) lim u(k) = 0. k→−∞ A weak solution of problem (3.6) is defined as follows. 1,p(.) Definition 3.8. A weak solution of (3.6) is a function u ∈ W0,− 0 X a(k − 1, ∆u(k − 1))∆v(k − 1) + k=−∞ 0 X |u(k)|p(k)−2 u(k)v(k) = k=−∞ such that 0 X f (k)v(k), k=−∞ (3.7) for any v ∈ 1,p(.) W0,− . By mimicking the proof of Theorem 3.4, we prove the following result. Theorem 3.9. Assume that (2.1)–(2.5) hold. Then, there exists at least one weak solution of problem (3.6). Let us now show the existence of weak heteroclinic solutions of problem (1.1). Proof of Theorem 3.2. We define v1 = u1 +1, where u1 is a weak solution of problem (3.2) and v2 = u2 − 1, where u2 is a weak solution of problem (3.6). Therefore, we deduce that u = v1 χZ+ + v2 χZ− is an heteroclinic solution of problem (1.1). Acknowledgments. This work was done while the authors were visiting the Abdus Salam International Centre for Theoretical Physics (ICTP), Trieste, Italy. The authors want to thank the mathematics section of ICTP for their hospitality for providing financial support. References [1] H. Brezis; Analyse fonctionnelle: thorie et applications. Paris, Masson (1983). [2] Alberto Cabada, Stepan Tersian; Existence of Heteroclinic solutions for discrete p-laplacian problems with a parameter, Nonlinear Anal. RWA 12 (2011), 2429-2434. [3] Y. Chen, S. Levine, M. Rao; Variable exponent, linear growth functionals in image processing, SIAM J. Appl. Math. 66(4) (2006), 1383-1406. [4] L. Diening; Theoretical and numerical results for electrorheological fluids, Ph.D. thesis, University of Freiburg, Germany, 2002. [5] A. Guiro, B. Koné, S. Ouaro; Weak Homoclinic solutions of anisotropic difference equations with variable exponents. Adv. Differ. Equ. 154 (2012), 1-13. [6] C. Marcelli, F. Papalini; Heteroclinic connections for fully non-linear non-autonomous second-order differential equations. J. Differential Equations. 241 (2007), 160-183. [7] K. R. Rajagopal, M. Ruzicka; Mathematical Modeling of Electrorheological materials, Contin. Mech. Thermodyn. 13 (2001), 59–78. [8] M. Ruzicka; Electrorheological fluids: modelling and mathematical theory, Lecture Notes in Mathematics 1748, Springer-Verlag, Berlin, 2002. Aboudramane Guiro Laboratoire d’Analyse Mathématique des Equations (LAME), UFR Sciences et Techniques, Université Polytechnique de Bobo Dioulasso, 01 BP 1091 Bobo-Dioulasso 01, Bobo Dioulasso, Burkina Faso E-mail address: abouguiro@yahoo.fr, aboudramane.guiro@univ-ouaga.bf EJDE-2013/225 WEAK HETEROCLINIC SOLUTIONS 9 Blaise Kone Laboratoire d’Analyse Mathématique des Equations (LAME), Institut Burkinabé des Arts et Métiers, Université de Ouagadougou, 03 BP 7021 Ouaga 03, Ouagadougou, Burkina Faso E-mail address: leizon71@yahoo.fr, blaise.kone@univ-ouaga.bf Stanislas Ouaro Laboratoire d’Analyse Mathématique des Equations (LAME), UFR. Sciences Exactes et Appliquées, Université de Ouagadougou, 03 BP 7021 Ouaga 03, Ouagadougou, Burkina Faso E-mail address: ouaro@yahoo.fr, souaro@univ-ouaga.bf