International Journal of Analysis and Applications ISSN 2291-8639 Volume 9, Number 2 (2015), 83-89 http://www.etamaths.com QUASI-COMPACT PERTURBATIONS OF THE WEYL ESSENTIAL SPECTRUM AND APPLICATION TO SINGULAR TRANSPORT OPERATORS LEILA MEBARKI1, MOHAMMED BENHARRAT2 AND BEKKAI MESSIRDI1,∗ Abstract. This paper is devoted to the investigation of the stability of the Weyl essen- tial spectrum of closed densely defined linear operator A subjected to additive perturbation K such that (λ − A − K)−1K or K(λ − A − K)−1 is a quasi-compact operator. The ob- tained results are used to describe the Weyl essential spectrum of singular neutron transport operator. 1. Introduction and preliminaries Let X and Y be complex infinite dimensional Banach spaces, and let C(X,Y ) (resp. L(X,Y )) denote the set of all closed, densely defined linear operators from X into Y (resp. the Banach algebra of all bounded operators), abbreviate C(X,X) (resp. L(X,X)) to C(X) (resp. L(X)). For A ∈C(X), write D(A) ⊂ X,N(A), R(A), σ(A) and ρ(A) respectively, the domain, the null space, the range, the spectrum and the resolvent set of A. The subset of all compact operators of L(X,Y ) is noted by K(X,Y ) and if X = Y we write K(X). Let I denote the identity operator in X. Let A ∈C(X), we know that D(A) provided with the graph norm ‖x‖A = ‖x‖+‖Ax‖ is a Banach space denoted by XA. Recall that an operator B is relatively bounded with respect to A or simply A-bounded if D(A) ⊆D(B) and B is bounded on XA. Definition 1.1. An operator V ∈L(X) is said to be quasi-compact operator if there exists a compact operator K and an integer m such that ‖V m −K‖ < 1. We denote by QK(X) the set of all quasi-compact operators. If r = r(V ) is the spectral radius of a bounded linear operator V on X, another equivalent definition is given in [2], to quasi-compactness, is that if there exists M and N closed V -invariant subspaces of X such that X = M ⊕N with r(V/M) < r and dimN < ∞. We refer the reader to [2] for a detailed presentation of the quasi-compactness. For A ∈ C(X,Y ), the nullity, α(A), of A is defined as the dimension of N(A) and the deficiency, β(A), of A is defined as the codimension of R(A) in Y . The set of Fredholm operators from X into Y is defined by Φ(X,Y ) = {A ∈C(X,Y ) : R(A) is closed in Y,β(A) < ∞ and α(A) < ∞}, If A ∈ Φ(X,Y ), the number ind(A) = α(A) −β(A) is called the index of A. The operator A is Weyl if it is Fredholm of index zero. The Fredholm essential spectrum σef (A) and the Weyl essential spectrum σew(A) are defined by: σef (A) := {λ ∈ C : (λ−A) is not Fredholm } and σew(A) := {λ ∈ C : (λ−A) is not Weyl }. 2010 Mathematics Subject Classification. 47A10. Key words and phrases. quasi-compact operators; Weyl essential spectrum; transport operators. c©2015 Authors retain the copyrights of their papers, and all open access articles are distributed under the terms of the Creative Commons Attribution License. 83 84 MEBARKI, BENHARRAT AND MESSIRDI For A ∈L(X), the Fredholm essential spectrum of A is also equal to the spectrum of A+K(X) in the Calkin algebra L(X)/K(X). Accordingly, the essential spectral radius of A, denoted by re(A), is given by the formula re(A) = lim n→+∞ dist(An,K(X)) 1 n = inf n∈N dist(An,K(X)) 1 n where dist(A,K(X)) = infK∈K(X) ‖A−K‖. Thus A is quasi-compact if and only if re(A) < 1. We say that A is a Riesz operator if λ − A is Fredholm for all non-zero complex numbers λ. Thus A is Riesz if and only if re(A) = 0. A useful property of a bounded linear operator A on a Banach space is that each spectral element of A which lies in the unbounded component of the complement of the essential spectrum of A is an eigenvalue of finite multiplicity. Further, if there are infinitely many of them, then they cluster only on the essential spectrum. Certainly we have the implications: compact ⇒ Riesz ⇒ quasi-compact. The following theorem gives an important and useful characterization of quasi-compact operators: Theorem 1.2. [1, Theorem I.6] If V ∈ QK(X) then for all for all complex number λ such that |λ| ≥ 1 then (λ−A) is a Weyl operator. As a consequence of this theorem, V is quasi-compact if (and only if) the peripheral spectrum of V contains only many poles of the resolvent R(.,V ) of V , and if V is power-bounded operator the residue of R(.,V ) at each peripheral pole is of finite rank. However, the concept of quasi- compact operator plays a crucial role and seems to be more appropriate because it evokes not only the special configuration of the spectrum of V , but also the fact that the spectral values of greatest modulus are eigenvalues associated with finite dimensional generalized eigenspaces. The purpose of this paper is to point out how, by means of the concept of the quasi- compactness it is possible to improve the definition of the Weyl essential spectrum. More precisely, we establish that σew(A) = σew(A + K) for all closed densely defined linear operator A, with K ∈ C(X) such that K is A-bounded and (λ−A−K)−1K or K(λ−A−K)−1 is a quasi-compact operator for all λ ∈ ρ(A + K). Our results generalize many known ones in the literature. The next Section is concerned with the definition of the Weyl essential spectrum of closed densely defined linear operator A subjected to additive perturbation K, A-bounded, such that (λ − A − K)−1K or K(λ − A − K)−1 is a quasi-compact operator for some λ ∈ ρ(A + K) and we establish some properties of quasi-compact operators. In the last Section we apply the results obtained in the second section to investigate the Weyl essential spectrum of the following singular neutron transport operator Aψ(x,ξ) = −ξ∇xψ(x,ξ) −σ(ξ)ψ(x,ξ) + ∫ Rn κ(x,ξ′)ψ(x,ξ′) dξ′ (x,ξ) ∈ Ω ×V = Tψ(x,ξ) + Kψ(x,ξ), with the vacum boundary conditions ψ|Γ− (x,ξ) = 0 (x,ξ) ∈ Γ− = {(x,ξ) ∈ ∂Ω ×V ; ξ ·n(x) < 0}. where Ω is a smooth open subset of Rn (n ≥ 1), V is the support of a positive Radon measure dµ on Rn and n(x) stands for the outward normal unit at x ∈ ∂Ω. The operator A describes the transport of particle (neutrons, photons, molecules of gas, etc.) in the domain Ω. The function ψ ∈ Lp(Ω×V,dxdµ(ξ)) (1 ≤ p < ∞) represents the number (or probability) density of particles having the position x and the velocity ξ. The functions σ(.) and κ(., .) are called, respectively, the collision frequency and the scattering kernel and will be assumed to be unbounded. More precisely, we will assume that there exist a closed subset E ⊂ Rn with zero dµ measure and a constant σ0 > 0 such that (1.1a) σ(.) ∈ L∞loc(R n \E), σ(ξ) > σ0; QUASI-COMPACT PERTURBATIONS OF THE WEYL ESSENTIAL SPECTRUM 85 (1.1b) [∫ Rn ( κ(.,ξ′) σ(ξ′ 1 p )qdµ(ξ′) ]1 q ∈ Lp(Rn). where q denotes the conjugate exponent of p. We describe here the Weyl essential spectrum of the operator A subjected to assumptions (1.1a), (1.1b) and the scattering kernel taking the form κ(., .) = κ1(., .) + κ2(., .) with κi(., .) are non-negative i = 1, 2. Theorem 3.3 assert that if conditions (1.1a) and (1.1b) are satisfied, the hyperplanes of Rn have zero dµ-measure, (i.e. for each e ∈ Sn−1,dµ{ξ ∈ Rn,ξ.e = 0} = 0, where Sn−1 denotes the unit ball of Rn), the collision operator K1 : ψ(x,ξ) −→ ∫ Rn κ1(x,ξ ′)ψ(x,ξ′)dξ′ is compact from Lp(Rn,σ(ξ)dµ(ξ)) into Lp(Rn,dµ(ξ)), and if further sup ξ∈Rn σ(ξ) (Reλ + σ(ξ))p ∥∥∥∥∥∥ [∫ Rn ( κ2(.,ξ ′) σ(ξ′ 1 p )qdµ(ξ′) ]1 q ∥∥∥∥∥∥ Lp(Rn) < 1 for Reλ > −σ0. Then K(λ−A−K)−1 is a sum of a compact operator K1(µ−T)−1[I+(µ−λ+K)(λ−T−K)−1]+ K2(µ−T)−1K1(λ−T−K)−1 and a pure contraction K2(µ−T)−1[I +(µ−λ+K2)(λ−T−K)−1] on Xp. Hence K(λ − A − K)−1 is a quasi-compact operator on Xp. Now, by the knowledge of the Weyl essential spectrum of the streaming operator T and Theorem 2.3 we assert that σew(A) = {λ ∈ C : Reλ ≤−σ0}. 2. Invariance of the Weyl essential spectrum If A ∈C(X) we define the sets RA(X) = {K ∈L(X) such that (λ−A−K)−1K ∈QK(X) for some λ ∈ ρ(A + K)}, LA(X) = {K ∈L(X) such that K(λ−A−K)−1 ∈QK(X) for some λ ∈ ρ(A + K)}, Theorem 2.1. Let A ∈C(X) with nonempty resolvent set. Then (2.1) σew(A) = ⋂ K∈RA(X) σ(A + K) = ⋂ K∈LA(X) σ(A + K). Proof. Set Σ1 = ⋂ K∈RA(X) σ(A + K) (resp. Σ2 = ⋂ K∈LA(X) σ(A + K)). We first claim that σew(A) ⊆ Σ1 (resp. σew(A) ⊆ Σ2). Indeed, if λ /∈ Σ1 (resp. λ /∈ Σ2), then there exists K ∈RA(X) (resp. K ∈LA(X)) such that λ ∈ ρ(A + K) and (λ−A−K)−1K ∈QK(X) (resp. K(λ−A−K)−1 ∈ QK(X)). Therefore Theorem 1.2 proves that I + (λ−A−K)−1K (resp. I + K(λ−A−K)−1) is a Weyl operator. Next, using the relation λ−A = (λ−A−K)(I + (λ−A−K)−1K) ( resp. λ−A = (I + K(λ−A−K)−1)(λ−A−K)) together with Atkinson’s theorem we get (λ−A) is a Weyl operator. This shows that λ /∈ σew(A). The opposite inclusion follows from K(X) ⊆RA(X) (resp. K(X) ⊆LA(X)). � It follows, immediately, from Theorem 2.1 that Corollary 2.2. let U(X) a subset of L(X) (not necessarily an ideal). If K(X) ⊆ U(X) ⊆ RA(X) or K(X) ⊆ U(X) ⊆LA(X). Then σew(A) = σew(A + K) for all K ∈ U(X). Proof. We have (2.2) ⋂ K∈RA(X) σ(A + K) ⊆ ⋂ K∈U(X) σ(A + K) ⊆ σew(A). and (2.3) ⋂ K∈LA(X) σ(A + K) ⊆ ⋂ K∈U(X) σ(A + K) ⊆ σew(A). 86 MEBARKI, BENHARRAT AND MESSIRDI Now, the result follows from Theorem 2.1. � Let A ∈ C(X) and let J be an arbitrary A-bounded operator on X. Hence we can regard A and J as operators from XA into X. They will be denoted by  and Ĵ, respectively. These belong to L(XA,X). Furthermore, we have the obvious relations (2.4)   α(Â) = α(A), β(Â) = β(A), R(Â) = R(A) α( + Ĵ) = α(A + J) β( + Ĵ) = β(A + J), and R( + Ĵ) = R(A + J) Theorem 2.3. Let A ∈C(X) with nonempty resolvent set. Then (2.5) σew(A) = ⋂ K∈∆A(X) σ(A + K). where ∆A(X) = {K ∈C(X),K is A-bounded and K(λ−A−K)−1 ∈QK(X) for all λ ∈ ρ(A + K)}. Proof. Since K(X) ⊂ ∆A(X), we refer that ⋂ K∈∆A(X) σ(A+K) ⊂ σew(A). Conversely, suppose that there exists K ∈ ∆A(X), hence by using (2.4) and Theorem 2.1, we infer that I + K(λ− A − K)−1 is a Weyl operator for all λ ∈ ρ(A + K). The fact that λ − A = (I + K(λ − A − K)−1)(λ − A − K) and by using the Atkinson′s theorem we get (λ − A) is a Weyl operator. This shows that λ /∈ σew(A). � Remark 2.4. Since K(X) ⊂ ∆A(X), K(X) is the minimal subset of C(X) (in the sense of inclusion) which characterize the essential Weyl spectrum . Hence Theorem 2.3 provides an improvement of the definition of σew(A) and is valid for a somewhat large variety of subsets of C(X). Furthermore, σew(A + K) = σew(A), for all K ∈ ∆A(X). It follows, immediately, from Theorem 2.3 that Corollary 2.5. Let A ∈C(X) and let M(X) be any subset of QK(X) (not necessarily an ideal) satisfying the condition (2.6) K(X) ⊆ M(X) ⊆QK(X). Then σew(A) = σew(A + K) for all K ∈ HA(X). HA(X) = {K ∈C(X),K is A-bounded and K(λ−A−K)−1 ∈ M(X) for all λ ∈ ρ(A + K)}. Proof. We have from (2.6) that K(X) ⊆ HA(X) ⊆ ∆A(X). From this we infer that (2.7) ⋂ K∈∆A(X) σ(A + K) ⊆ ⋂ K∈HA(X) σ(A + K) ⊆ σew(A). Now, the result follows from Theorem 2.3. � Note that in applications (transport operators, operators arising in dynamic populations, etc.), we deal with operators A and B such that B = A + K where A ∈ C(X) and K is, in general, a closed (or closable) A-bounded linear operator. The operator K does not necessarily satisfy the hypotheses of the previous results. For some physical conditions on K, we have information about the operator (λ − A)−1 − (λ − B)−1 (λ ∈ ρ(A) ∩ ρ(B)). So we have the following useful stability result. Theorem 2.6. Let A,B ∈ C(X) such that ρ(A) ∩ρ(B) 6= ∅. If for some λ ∈ ρ(A) ∩ρ(B) the operator (λ−A)−1 − (λ−B)−1 ∈ ∆A(X), then σef (A) = σef (B) and σew(A) = σew(B). QUASI-COMPACT PERTURBATIONS OF THE WEYL ESSENTIAL SPECTRUM 87 Proof. Without loss of generality, we suppose that λ = 0. Hence 0 ∈ ρ(A) ∩ρ(B). Therefore, we can write for µ 6= 0 µ−A = −µ(µ−1 −A−1)A. Since, A is one to one and onto, then α(µ−A) = α(µ−1 −A−1) and β(µ−A) = β(µ−1 −A−1). This shows that (µ − A) is a Fredholm operator if and only if (µ−1 − A−1) is a Fredholm operator and ind(µ−A) = ind(µ−1 −A−1). Now, assume that A−1 −B−1 ∈ ∆A(X). Hence using Theorem 2.3 we conclude that (µ − A) is a Fredholm operator if and only if (µ − B) is a Fredholm operator and ind(µ − A) = ind(µ − B) for each µ /∈ σef (A). This proves σef (A) = σef (B) and σew(A) = σew(B). � 3. Application to transport operator In this section we are concerned with the Weyl essential spectrum of singular transport operators (3.1a) Aψ(x,ξ) = −ξ ·∇xψ(x,ξ) −σ(ξ)ψ(x,ξ) + ∫ Rn κ(x,ξ′)ψ(x,ξ′) dξ′ (x,ξ) ∈ Ω ×V, (3.1b) ψ|Γ− (x,ξ) = 0 (x,ξ) ∈ Γ−. where Ω is a smooth open subset of Rn (n ≥ 1), V is the support of a positive Radon measure dµ on Rn and ψ ∈ Lp(Ω×V,dxdµ(ξ)) (1 ≤ p < ∞). In (3.1b) Γ− denotes the incoming part of the boundary of the phase space Ω ×V Γ− = {(x,ξ) ∈ ∂Ω ×V ; ξ ·n(x) < 0}, where n(x) stands for the outward normal unit at x ∈ ∂Ω. The operator A describes the transport of particle (neutrons, photons, molecules of gas, etc.) in the domain Ω. The function ψ represents the number (or probability) density of particles having the position x and the velocity ξ. The functions σ(.) and κ(., .) are called, respectively, the collision frequency and the scattering kernel. Let us first introduce the functional setting we shall use in the sequel. Let Xp = L p(Ω ×V,dxdµ(ξ)), Xσp = L p(Ω ×V,σ(ξ)dxdµ(ξ)) 1 ≤ p < ∞, Lpσ(R n) = Lp(Rn,σ(ξ)dµ(ξ)). We define the partial Sobolev space Wp = {ψ ∈ Xp ; ξ ·∇xψ ∈ Xp}. For any ψ ∈ Wp, one can define the space traces ψ|Γ− on Γ−, W̃p = {ψ ∈ Wp ; ψ|Γ− = 0}. The streaming operator T associated with the boundary condition (3.1b) is{ T : D(T) ⊂ Xp → Xp ψ 7→ Tψ(x,ξ) := −ξ ·∇xψ(x,ξ) −σ(ξ)ψ(x,ξ), with domain D(T) := W̃p ∩Xσp . The transport operator (3.1) can be formulated as follows A = T + K, where K denotes the following collision operator K : Xp → Xp ψ 7→ ∫ Rn κ(x,ξ′)ψ(x,ξ′) dξ′ . 88 MEBARKI, BENHARRAT AND MESSIRDI We will assume that the scattering kernel κ(., .) = κ1(., .) + κ2(., .), κi(., .) are non-negative i = 1, 2 and there exist a closed subset E ⊂ Rn with zero dµ measure and a constant σ0 > 0 such that (3.2a) σ(.) ∈ L∞loc(R n \E), σ(ξ) > σ0; (3.2b) [∫ Rn ( κi(.,ξ ′) σ(ξ′ 1 p ) )qdµ(ξ′) ]1 q ∈ Lp(Rn), i = 1, 2. where q denotes the conjugate exponent of p. Denote by Kiψ(x,ξ) = ∫ Rn κi(x,ξ ′)ψ(x,ξ′) dξ′ i = 1, 2. Using boundedness of Ω and the assumption (3.2b) we can fined that Ki ∈L(Xσp ,Xp) with (3.3) ‖Ki‖L(Xσp ,Xp) ≤ ∥∥∥∥∥∥ [∫ Rn ( κi(.,ξ ′) σ(ξ′ 1 p ) )qdµ(ξ′) ]1 q ∥∥∥∥∥∥ Lp(Rn) i = 1, 2. Note that a simple calculation using the assumption (3.2a) shows that Xσp is a subset of Xp and the the embedding Xσp ↪→ Xp is continuous. Let us now consider the resolvent equation (3.4) (λ−T)ψ = ϕ, where ϕ is a given element of Xp and the unknown ψ must be founded in D(T). For Reλ > −σ0, the solution of (3.4) reads as follows (3.5) ψ(x,ξ) = ∫ t(x,ξ) 0 e−(λ+σ(ξ))sϕ(x−sξ,ξ)ds, where t(x,ξ) = sup{t > 0 ; x−sξ ∈ Ω, ∀0 < s < t} = inf{s > 0 ; x−sξ /∈ Ω}. An immediate consequence of these facts is that σ(T) ⊆{λ ∈ C : Reλ ≤−σ0}, and in [8, 7] shows that σ(T) is reduced to the continuous spectrum σc(T) of T, that is (3.6) σ(T) = σc(T) = {λ ∈ C : Reλ ≤−σ0}, Since all essential spectra are enlargement of the continuous spectrum we infer that (3.7) σew(T) = σef (T) = {λ ∈ C : Reλ ≤−σ0}. Lemma 3.1. The collision operator K is T -bounded. Proof. Let λ ∈ C be such that Reλ > −σ0 and consider ψ ∈ Xp. It follows from (3.5) that∫ Ω ∣∣(λ−T)−1ψ(x,ξ)∣∣p dx ≤ 1 (Reλ + σ(ξ))p ∫ Ω |ψ(x,ξ)|p dx Therefore, (3.8) ∥∥(λ−T)−1ψ∥∥ Xσp ≤ sup ξ∈Rn σ(ξ) (Reλ + σ(ξ))p ‖ψ‖Xp Hence, (λ−T)−1 ∈ L(Xp,Xσp ). Using now the equation (3.3) to deduce that the operator K is T-bounded. � Proposition 3.2 ([7, Proposition 4.1]). Let Ω be a bounded subset of Rn and 1 < p < ∞. If the hypotheses (3.2a) and (3.2b) are satisfied for κ1(., .), the measure dµ satisfies (3.9) { the hyperplanes have zero dµ-measure, i.e., for each, e ∈ Sn−1,dµ{ξ ∈ Rn,ξ.e = 0} = 0, where Sn−1 denotes the unit ball of Rn and the collision operator K1 : Lpσ(R n) −→ Lp(Rn) is compact. Then for any λ ∈ C such that Reλ > −σ0, the operator K1(λ−T)−1 is compact on Xp. QUASI-COMPACT PERTURBATIONS OF THE WEYL ESSENTIAL SPECTRUM 89 Now we are in position to state the main result of this section. Theorem 3.3. Assume that the hypotheses of Proposition 3.2 are satisfied and (3.10) sup ξ∈Rn σ(ξ) (Reλ + σ(ξ))p ∥∥∥∥∥∥ [ ∫ Rn ( κ2(.,ξ ′) σ(ξ′ 1 p ) )qdµ(ξ′) ]1 q ∥∥∥∥∥∥ Lp(Rn) < 1 for Reλ > −σ0. Then (1) For all λ ∈ ρ(A + K), we have K(λ−A−K)−1 is quasi-compact on Xp, (2) σew(A) = σew(T) = {λ ∈ C : Reλ ≤−σ0}. Proof. (1) Let λ ∈ ρ(A + K) and µ ∈ ρ(T) such that∥∥I + (µ−λ + K2)(λ−T −K)−1∥∥ < 1. By this estimate and (3.10) we can deduce that K2(µ−T)−1[I + (µ−λ + K2)(λ−T −K)−1] is a pure contraction. On other hand, we have K(λ−A−K)−1 = K(µ−T)−1[I + (µ−λ + K)(λ−T −K)−1] = K1(µ−T)−1[I + (µ−λ + K)(λ−T −K)−1] + K2(µ−T)−1K1(λ−T −K)−1 + K2(µ−T)−1[I + (µ−λ + K2)(λ−T −K)−1]. By this equation and Proposition 3.2 we have K(λ−A−K)−1 is a sum of a compact operator and a pure contraction on Xp. Hence K(λ−A−K)−1 is a quasi-compact operator on Xp. (2) The first item together with hypotheses on K implies that K ∈ ∆A(Xp). Now the second item follows from Theorem 2.3, Remark 2.4 and the relation (3.7). � References [1] A. Brunel and D. Revuz, Quelques applications probabilistes de la quasi-compacité. Ann. Inst. Henri et Poincaré, B 10(3) (1974), 301-337. [2] H. Hennion and L. Hervé, Limit Theorems for Markov Chains and Stochastic Properties of Dynamical Systems by Quasi-Compactness. Springer-Verlag Berlin Heidelberg (2001). [3] K. Latrach, Essential spectra on spaces with the Dunford-Pettis property , J. Math. Anal. Appl., 233, (1999), 607-622. [4] K. Latrach, A. Dehici, Fredholm, Semi-Fredholm Perturbations and Essential spectra, J. Math. Anal. Appl., 259 (2001), 277-301. [5] K. Latrach, A. Dehici, Relatively strictly singular perturbations, Essential spectra, and Application to transport operators, J. Math. Anal. Appl., 252(2001), 767-789. [6] K. Latrach, A. Jeribi, Some results on Fredholm operators, essential spectra, and application, J. Math. Anal. Appl., 225, (1998), 461-485. [7] K. Latrach, J. Martin Paoli, Relatively compact-like perturbations, essensial spectra and application, J. Aust. Math. Soc. 77 (2004), 73-89. [8] M. Mokhtar-Kharroubi, Time asymptotic behaviour and compactness in neutron transport theory, Euro. J. Mech., B Fluids, 11(1992), 39-68. [9] M. Schechter, Invariance of essential spectrum, Bull. Amer. Math. Soc. 71 (1965), 365-367. [10] K. Yosida, Quasi-completely continuous linear functional operations, Jap. Journ. of Math., 15 (1939), 297– 301. [11] K. Yosida, Functional Analysis, Springer-Verlag Berlin Heiderlberg, 1980. [12] K. Yosida, S. Kakutani, Operator-theoretical treatment of Markov’s process and mean ergodic theorem, Ann. of Math., 42, (1941), 188-228. 1Département de Mathématiques, Université d’Oran 1, Algérie 2Département de Mathématiques et Informatique, Ecole Nationale Polytechnique d’Oran (Ex. ENSET d’Oran); B.P. 1523 Oran-El M’Naouar, Oran, Algérie ∗Corresponding author