Asymptotic Solution to a Mixed Boundary-Value Problem in a Thick Multi-Structure of Type 3:2:2

The leading terms of the asymptotic expansion for the solution to a mixed boundary value problem for the Poisson equation in a thick multi-structure, which is the union of some domain and a large number N of ε-periodically situated thin annular disks with variable thickness of order ε = O(N⁻¹), are...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Datum:2005
Hauptverfasser: De Maio, U., Mel'nyk, T.A.
Format: Artikel
Sprache:English
Veröffentlicht: Інститут прикладної математики і механіки НАН України 2005
Schriftenreihe:Український математичний вісник
Online Zugang:http://dspace.nbuv.gov.ua/handle/123456789/124606
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
Назва журналу:Digital Library of Periodicals of National Academy of Sciences of Ukraine
Zitieren:Asymptotic Solution to a Mixed Boundary-Value Problem in a Thick Multi-Structure of Type 3:2:2 / U. De Maio, T.A. Mel'nyk // Український математичний вісник. — 2005. — Т. 2, № 4. — С. 463-481. — Бібліогр.: 34 назв. — англ.

Institution

Digital Library of Periodicals of National Academy of Sciences of Ukraine
id irk-123456789-124606
record_format dspace
spelling irk-123456789-1246062017-09-30T03:04:08Z Asymptotic Solution to a Mixed Boundary-Value Problem in a Thick Multi-Structure of Type 3:2:2 De Maio, U. Mel'nyk, T.A. The leading terms of the asymptotic expansion for the solution to a mixed boundary value problem for the Poisson equation in a thick multi-structure, which is the union of some domain and a large number N of ε-periodically situated thin annular disks with variable thickness of order ε = O(N⁻¹), are constructed and the corresponding estimates in the Sobolev space H¹ are proved as ε → 0. 2005 Article Asymptotic Solution to a Mixed Boundary-Value Problem in a Thick Multi-Structure of Type 3:2:2 / U. De Maio, T.A. Mel'nyk // Український математичний вісник. — 2005. — Т. 2, № 4. — С. 463-481. — Бібліогр.: 34 назв. — англ. 1810-3200 2000 MSC. 35B27, 3540, 35J25, 35C20, 35B25. http://dspace.nbuv.gov.ua/handle/123456789/124606 en Український математичний вісник Інститут прикладної математики і механіки НАН України
institution Digital Library of Periodicals of National Academy of Sciences of Ukraine
collection DSpace DC
language English
description The leading terms of the asymptotic expansion for the solution to a mixed boundary value problem for the Poisson equation in a thick multi-structure, which is the union of some domain and a large number N of ε-periodically situated thin annular disks with variable thickness of order ε = O(N⁻¹), are constructed and the corresponding estimates in the Sobolev space H¹ are proved as ε → 0.
format Article
author De Maio, U.
Mel'nyk, T.A.
spellingShingle De Maio, U.
Mel'nyk, T.A.
Asymptotic Solution to a Mixed Boundary-Value Problem in a Thick Multi-Structure of Type 3:2:2
Український математичний вісник
author_facet De Maio, U.
Mel'nyk, T.A.
author_sort De Maio, U.
title Asymptotic Solution to a Mixed Boundary-Value Problem in a Thick Multi-Structure of Type 3:2:2
title_short Asymptotic Solution to a Mixed Boundary-Value Problem in a Thick Multi-Structure of Type 3:2:2
title_full Asymptotic Solution to a Mixed Boundary-Value Problem in a Thick Multi-Structure of Type 3:2:2
title_fullStr Asymptotic Solution to a Mixed Boundary-Value Problem in a Thick Multi-Structure of Type 3:2:2
title_full_unstemmed Asymptotic Solution to a Mixed Boundary-Value Problem in a Thick Multi-Structure of Type 3:2:2
title_sort asymptotic solution to a mixed boundary-value problem in a thick multi-structure of type 3:2:2
publisher Інститут прикладної математики і механіки НАН України
publishDate 2005
url http://dspace.nbuv.gov.ua/handle/123456789/124606
citation_txt Asymptotic Solution to a Mixed Boundary-Value Problem in a Thick Multi-Structure of Type 3:2:2 / U. De Maio, T.A. Mel'nyk // Український математичний вісник. — 2005. — Т. 2, № 4. — С. 463-481. — Бібліогр.: 34 назв. — англ.
series Український математичний вісник
work_keys_str_mv AT demaiou asymptoticsolutiontoamixedboundaryvalueprobleminathickmultistructureoftype322
AT melnykta asymptoticsolutiontoamixedboundaryvalueprobleminathickmultistructureoftype322
first_indexed 2025-07-09T01:42:39Z
last_indexed 2025-07-09T01:42:39Z
_version_ 1837131744469319680
fulltext Український математичний вiсник Том 2 (2005), № 4, 463 – 481 Asymptotic Solution to a Mixed Boundary-Value Problem in a Thick Multi-Structure of Type 3:2:2 Umberto De Maio and Taras A. Mel’nyk (Presented by E. Ya. Khruslov) Abstract. The leading terms of the asymptotic expansion for the solution to a mixed boundary value problem for the Poisson equation in a thick multi-structure, which is the union of some domain and a large number N of ε-periodically situated thin annular disks with variable thickness of order ε = O(N−1), are constructed and the corresponding estimates in the Sobolev space H1 are proved as ε → 0. 2000 MSC. 35B27, 3540, 35J25, 35C20, 35B25. Key words and phrases. Asymptotic expansions, homogenization, thick multi-structure. 1. Introduction and Statement of the Problem It is an interesting problem to study the behaviour of solutions to boundary-value problems when the domain is perturbed. There are many kinds of the domain perturbations and we need different asymptotic meth- ods to study boundary-value problems in perturbed domains. Numerous monographs and papers (see, e.g., [3,5,9,10,12,16,17,19,31–33] and ref- erences there) are devoted to asymptotic methods for the investigation of boundary-value problems in domains with complex dependence on a pa- rameter of perturbation (perforated domains, partly perforated domains, lattice frames, junctions of domains with different limit dimensions, etc.). Boundary-value problems in thick multi-structures (components of such junctions infinitely increases as the perturbation parameter ε → 0) have own specific difficulties and such problems deserve special attention. As it was shown in E. Sanchez-Palencia’s papers [13, 32], such problems lose the coercitivity as ε → 0 and this creates special difficulties in the Received 18.11.2004 ISSN 1810 – 3200. c© Iнститут прикладної математики i механiки НАН України 464 Asymptotic Solution... asymptotic investigation. In [21]– [28], a classification of such thick multi- structures was given and basic results were obtained for boundary-value problems in thick junctions of different types. It was shown that qual- itative properties of solutions essentially depend on the junction type and on the conditions given on the boundaries of the attached thin do- mains. There and in [11] a survey of results obtained in this direction is presented. Here we mention only the pioneer papers [15, 34], where the asymptotic behaviour of Green’s function of the Neumann problem for the Helmholtz equation in an unbounded junction body of type 3:2:1 was studied, and the papers [6,7], where the homogenization of nonlinear problems in thick junctions of types 3:2:1, 2:2:1 was made. Such thick junctions are prototypes of widely used engineering con- structions, industrial installations, spaceship grids as well as many other physical and biological systems with very different characteristic scales. Despite the enormous growth in computational power, it is often im- possible to represent a complete system at the finest scale for which the various constitutive elements may suitably be represented. Increase in the size of computational domains for thick multi-structures naturally leads to longer computing times and makes it very difficult to maintain an acceptable level of accuracy. Thus, asymptotic analysis of boundary value problems in such do- mains is an important task for applied mathematics. The aim of the analysis is to develop rigorous asymptotic methods for boundary value problems in thick junctions of different types as ε → 0, i.e., when the number of attached thin domains increases and their thickness decreases. In this paper we consider a model thick multi-structure (thick junc- tion) Ωε of type 3:2:2. It consists of a junction’s body Ω0 = { x ∈ R 3 : 0 < x2 < l, r := √ x2 1 + x2 3 < a0 } and a large number N of thin annular disks G(ε) = ⋃N−1 j=0 Gj(ε), Gj(ε) = {x ∈ R 3 : −ε h−(r) < x2−ε(j+1/2) < εh+(r), a0 ≤ r < a1}, i.e., Ωε = Ω0 ⋃ G(ε). Here h− and h+ are piecewise smooth functions on [a0, a1], 0 < h±(r) < 1 2 for r ∈ [a0, a1], and the functions h− and h+ are locally constant and equal at an enough small neighborhood of the point a0; the number of the thin disks is equal to a large even integer N, therefore, ε = l/N is a small parameter, which characterizes the distance between the neighboring thin U. De Maio, T. A. Mel’nyk 465 disks and their thickness. The type 3:2:2 of the junction refers to the limit dimensions of the junction’s body, the joint zone (the lateral surface of Ω0), and each attached thin disk. In Ωε we consider the mixed boundary value problem −∆xuε(x) = fε(x), x ∈ Ωε, uε(x) = 0, x ∈ S(0) ∪ S(l), ∂νuε(x) = 0, x ∈ ∂Ωε \ ( S(0) ∪ S(l) ) , (1.1) where ∂ν = ∂/∂ν is the outward normal derivative; S(0) = {x ∈ ∂Ω0 : x2 = 0}, S(l) = {x ∈ ∂Ω0 : x2 = l} and the right-hand side fε ∈ L2(Ωε). The aim of the paper is to construct the leading terms of the asymp- totic expansion for the solution to problem (1.1) and to prove the cor- responding asymptotic estimates as ε → 0, i.e., when the number of attached thin disks increases and their thickness decreases. There are two ways of investigation of boundary-value problems in perturbed domains. The first one lies in the proofs of convergence theo- rems with the help of special extension operators. The second one consists in the construction the leading terms of the asymptotic expansions for the solutions and proving the corresponding asymptotic estimates. For these two ways we need different assumptions for the right-hand sides and the geometrical structure of the perturbed domains. To construct the leading terms of asymptotics of solutions to boundary-value problems in thick junctions, the method of matched asymptotic expansions and asymptotic methods for thin domains were used in [21]– [28]. It is turned out that the corresponding limit problem is derived from limit problems for each domain forming the thick junc- tion with the help of the solutions to junction-layer problems around the joint zone. However, the junction-layer solutions behave as powers (or logarithm) at infinity and do not decrease exponentially. Therefore, they influence directly the leading terms of the asymptotics. Regarding the approximations of the solutions it should be mentioned that in [1,2] the correctors for the solutions to the Laplace equation in a plane thick junction of type 2:1:1 with the Dirichlet condition were con- structed outside a layer of width 2ε in a neighborhood of the joint zone. But for applied problems, it is very important to construct the asymp- totic expansion for the solution and to prove the asymptotic estimates in all thick junction, especially in a neighborhood of the joint zone, since the solution has singularities in the joint zone. It should be emphasized new moments in the investigation of problem (1.1). The first one is type 3:2:2; only spectral problems were considered in such multi-structures (see [4,27]). Secondly, only boundary value prob- 466 Asymptotic Solution... lems in thick junctions with attached thin domains whose thickness are unvarying or problems in domains with rapidly oscillating smooth bound- aries (see [8]) were considered till now. Our thick junction Ωε has the Lipschitz boundary and the thickness of each thin rings is equal to the value εh0(r), r ∈ [a0, a1]. Because of this, the coefficients of the corre- sponding limit problem depend on the function h0 = h− + h+. Finally, we require for the right-hand side fε of problem (1.1) only the minimal condition, namely, fε ∈ L2(Ωε). For comparison, in [8] the authors as- sumed that the right-hand side f does not depend on ε and f ∈ H1(Ωε) to prove only the convergence theorem for the solution of the Neumann problem in a plane thick junction of type 2:1:1; in [2] to construct the leading terms of asymptotics for the solution to the Dirichlet problem in a plane thick junction of type 2:1:1, the right-hand side f must vanish on the thin rods and f ∈ H4 in the junction’s body, in [23] fε has special form like in Corollary 2. The convergence theorem for the solution to the Neumann problem for the Ukawa equation in Ωε, when h− = h+, was proved by authors in [11]. 2. Formal Asymptotic Representation for the Solution In this section, we construct the leading terms of outer expansions both in the junction’s body and in each thin disks as well as the leading terms of an inner expansion in neighborhood of the joint zone. Then using the method of matched asymptotic expansions, the corresponding limit problem is derived. Also we assume in this section that the right- hand side in (1.1) is independent of ε, i.e., fε = f0 and f0 is smooth in Ω1, where Ω1 is the interior of Ω0 ∪D. The domain D = {x : 0 < x2 < l, a0 < r < a1} is filled up by the thin disks in the limit passage. 2.1. Outer Expansions We seek the leading terms of the asymptotics for the solution uε, restricted to Ω0, in the form uε(x) ≈ v+ 0 (x) + ∞∑ k=1 εkv+ k (x, ε), (2.1) and, restricted to Gj(ε), in the form uε(x) ≈ v−0 (x) + ∞∑ k=1 εkv−k (x, ξ2 − j), ξ2 = ε−1x2 . (2.2) U. De Maio, T. A. Mel’nyk 467 The expansions (2.1) and (2.2) are usually called outer expansions. Substituting the series (2.1) in the equation of problem (1.1) and in the boundary conditions on the bases S(0), S(l) of the cylinder Ω0 and collecting the coefficients of the same powers of ε, we get the following relations for the function v+ 0 : −∆xv + 0 (x) = f0(x), x ∈ Ω0, v+ 0 (x) = 0, x ∈ S(0) ∩ S(l). (2.3) Now we find limit relations in the domain D. Assuming for the mo- ment that the functions v−k in (2.2) are smooth, we write their Taylor series with respect to the x2 at the point x2 = ε(j + 1 2) and pass to the "fast" variable ξ2 = ε−1x2. Then (2.2) takes the form uε(x) = v−0 ( x1, ε(j + 1/2), x3 ) + +∞∑ k=1 εkV j k (x1, ξ2, x3), x ∈ Gj(ε), (2.4) where V j k = v−k ( x1, ε(j + 1/2), x3, ξ2 ) + k∑ m=1 (ξ2 − j − 1 2)m m! ∂mv−k−m ∂xm2 ( x1, ε(j + 1/2), x3, ξ2 ) . (2.5) Let us substitute (2.4) into (1.1) instead of uε. Since the Laplace operator takes the form ∆x = ε−2 ∂2 ∂ξ22 + ∆x̃, where x̃ = (x1, x3), and the outward normal to the lateral surface of the thin disk Gj(ε) (beside some set of zero measure) has the form ν±(x̃, ε) = 1√ 1 + ε2|h′±(r)|2 ( −εh′±(r) x1 r , ±1, −εh′±(r) x3 r ) , the collection of coefficients of the same power of ε gives us one dimen- sional Neumann problems with respect to ξ2. Write the first two ones: ∂2 ξ2ξ2V j 1 (x̃, ξ2) = 0, ξ2 ∈ Ij(r), ∂ξ2V j 1 ( x̃, j+2−1±h±(r) ) = 0; (2.6) − ∂2 ξ2ξ2V j 2 (x̃, ξ2) = ∆x̃ v − 0 (x̃, ε(j + 1/2)) + f0(x̃, ε(j + 1/2)), ξ2 ∈ Ij(r), (2.7) ∂ξ2V j 2 ( x̃, j + 2−1 ± h±(r) ) = ±∇x̃ h±(r) · ∇x̃ v − 0 (x̃, ε(j + 1/2)), (2.8) where Ij(r) := ( j + 1 2 − h−(r) , j + 1 2 + h+(r) ) , ∂ξ2 = ∂/∂ξ2, ∂ 2 ξ2ξ2 = ∂2/∂ξ22 . 468 Asymptotic Solution... From (2.6), it follows that the function V j 1 does not depend on ξ2. We restrict ourselves to the leading term of the asymptotics, and thus set V j 1 = 0. Then, by virtue of (2.5), we have v−1 ( x1, ε(j+1/2), x3, ξ2 ) = −∂x2v − 0 ( x1, ε(j+1/2), x3 )( ξ2−j−1/2 ) , (2.9) where ∂x2 = ∂/∂x2. The solvability condition for the problem (2.7)–(2.8) is given by the differential equation −divx̃ ( h0(r)∇x̃ v − 0 (x̃, ε(j + 1/2)) ) = f0(x̃, ε(j + 1/2)), r ∈ (a0, a1), (2.10) where h0(r) = h−(r) + h+(r), r ∈ [a0, a1]. Since these planes x2 = ε(j + 1/2), j = 0, 1, . . . , N − 1, make up the ε-net in D, we can spread this equation in all domain D. Due to the Neumann conditions for the solution to problem (1.1) we must require from v−0 to satisfy the condition ∂rv − 0 (x) = 0, r = a1, x2 ∈ (0, l), (2.11) where ∂r = ∂/∂r is the derivative with respect to polar radius r which coincides with the outward normal derivative in this case. So, it remains to provide the continuity of the asymptotic approxima- tion and their gradients in the joint zone Γ0 := ∂Ω0 \ ( S(0) ∪ S(l) ) . It is doubtless the condition v+ 0 (x) = v−0 (x), x ∈ Γ0. (2.12) To get the second transmission condition we should use the method of matched asymptotic expansions for the outer expansions (2.1), (2.2) and an inner expansion which we will construct in the following section. 2.2. Inner Expansion In a neighborhood of Γ0 we consider the Laplace operator in the cylindrical coordinates r, ϕ, x2, where r = √ x2 1 + x2 3 and tan(ϕ) = x3/x1, and then pass to the “rapid” coordinates ξ = (ξ1, ξ2), where ξ1 = −ε−1(r− a0) and ξ2 = ε−1x2. Then Laplace’s operator in the coordinates ξ = (ξ1, ξ2), ϕ has the following form ε−2 ( ∂2 ∂ξ21 + ∂2 ∂ξ22 ) − ε−1 a0 − εξ1 ∂ ∂ξ1 + 1 (a0 − εξ1)2 ∂2 ∂ϕ2 . (2.13) We seek the leading terms of the inner expansion in a neighborhood of Γ0 in the form uε(x) ≈ v+ 0 (x)|r=a0 + ε ( Z1(ξ)(∂x2v + 0 (x))|r=a0 + Z2(ξ)(∂rv + 0 (x))|r=a0 ) + . . . . (2.14) U. De Maio, T. A. Mel’nyk 469 Substituting (2.14) in (2.13) and in the Neumann condition, collecting the coefficients of the same power of ε, we arrive at junction-layer problems for the functions Z1 and Z2 : −∆ξ1ξ2 Zi(ξ) = 0, ξ ∈ Π, ∂ξ1Zi(0, ξ2) = 0, (0, ξ2) ∈ ∂Π+ \ Ih, ∂ξ2Zi(ξ) = −δ1i, ξ ∈ ∂Π− \ Ih, ∂kξ2Zi(ξ1, 0) = ∂kξ2Zi(ξ1, 1), ξ1 > 0, k = 0, 1. (2.15) Here Π is the union of semi-infinite strips Π+ = (0,+∞) × (0, 1) and Π− = (−∞, 0] × Ih, where Ih = ( (1 − h)/2 , (1 + h)/2 ) , the constant h is equal to h0(a0). The last periodic condition in (2.15) due to the periodicity of the thin disks {Gj(ε) : j = 0, . . . , N − 1}. The same junction-layer problems were investigated in [22]. The main asymptotic relations for the functions {Zi} can be obtained from general results about the asymptotic behaviour of solutions to elliptic problems in domains with different exits to infinity [14, 29]. However, using the symmetry of the domain Π, we can define more exactly the asymptotic relations and detect other properties of the junction-layer solutions Z1, Z2 similarly as in the papers [22,23]. Statement 2.1 ([22]). There exist solutions Zi ∈ H1 loc,η2 (Π), i = 1, 2, of problems (2.15), which have the following differentiable asymptotics Z1(ξ) = { O(exp(−2πξ1)), ξ1 → +∞, −ξ2 + 1 2 + O(exp(πh−1ξ1)), ξ1 → −∞; (2.16) Z2(ξ) = { −ξ1 + ch + O(exp(−2πξ1)), ξ1 → +∞, −h−1ξ1 + O(exp(πh−1ξ1)), ξ1 → −∞. (2.17) In addition, the function Z1 is odd in ξ2 and Z2 is even in ξ2 with respect to 1/2. Now we verify the matching conditions for the outer expansions (2.1), (2.2) and the inner expansion (2.14), namely, the leading terms of the asymptotics of the outer expansions as r → a0 ± 0 must coincide with the leading terms of the inner expansion as ξ1 → ∓∞ respectively. Near the point x ∈ Γ0 the function v+ 0 has the following asymptotics v+ 0 (x) = v+ 0 (x)|r=a0 − ε ξ1 (∂rv + 0 (x))|r=a0 + O(ε2ξ21), r → a0 − 0. Taking into account the asymptotics of Z1 and Z2 as ξ1 → +∞, we see that the matching conditions are satisfied for the expansion (2.1) and 470 Asymptotic Solution... (2.14). The asymptotics of (2.2) as r → a0 + 0 and the asymptotics of (2.14) as ξ1 → −∞ are the following v−0 (x)|r=a0 + ε ( Y (ξ2)(∂x2v − 0 (x))|r=a0 − ξ1 (∂rv − 0 (x))|r=a0 ) + . . . , r → a0 + 0; v+ 0 (x)|r=a0 + ε ( Y (ξ2)(∂x2v + 0 (x))|r=a0 − h−1ξ1 (∂rv + 0 (x))|r=a0 ) + . . . , ξ1 → −∞. Comparing the main terms of these asymptotics, we get the first trans- mission condition (2.12) and the second one ∂rv + 0 (x) = h ∂rv − 0 (x), x ∈ Γ0. (2.18) So, the function v0(x) = { v+ 0 (x), x ∈ Ω0, v−0 (x), x ∈ D, (2.19) must satisfy the relations (2.3), (2.10)–(2.12), (2.18), which form the limit problem −∆xv + 0 (x) = f0(x), x ∈ Ω0, −divx̃ ( h0(r)∇x̃ v − 0 (x) ) = h0(r) f0(x), x ∈ D, ∂rv − 0 (x) = 0, r = a1, x2 ∈ (0, l), v+ 0 (x) = 0, x ∈ S(0) ∪ S(l), v+ 0 (x) = v−0 (x), x ∈ Γ0, ∂rv + 0 (x) = h0(a0) ∂rv − 0 (x), x ∈ Γ0, (2.20) for problem (1.1). Let us show that there exist a unique weak solution v ∈ H0 to problem (2.20) if f0 ∈ L2(Ω1). Here H0 := { ϕ ∈ L2(Ω1) : ϕ ∈ H1(Ω0), ϕ = 0 on S(0) ∪ S(l), ∂x1ϕ ∈ L2(D), ∂x3ϕ ∈ L2(D), ϕ|r=a0−0 = ϕ|r=a0+0 on Γ0 } is an anisotropic Sobolev space with the scalar product ( ϕ,ψ ) H0 = ∫ Ω0 ∇xϕ · ∇xψ dx+ ∫ D h0(r)∇x̃ ϕ · ∇x̃ ψ dx. U. De Maio, T. A. Mel’nyk 471 A function v ∈ H0 is called a weak solution to problem (2.20) if it satisfies the following integral identity ( v, ψ ) H0 = ∫ Ω0 f0(x)ψ(x) dx+ ∫ D h0(r) f0(x)ψ(x) dx, ∀ψ ∈ H0. (2.21) Next using standard Hilbert space methods and the Lax–Milgram lemma, it is easy to prove the existence and uniqueness of the weak solution to problem (2.20). 3. Corrector and Asymptotic Estimates Let f0 be a function in H3(Ω1). Assume that f0 and ∂x2f0 vanish on S(0) ∪ S(l). Let v0 ∈ H0 be the unique weak solution to problem (2.20) with the right-hand side f0. With the help of v0 and the junction- layer solutions Z1, Z2 (see Statement 2.1), we define the leading terms in (2.1), (2.2) and (2.14). Then matching these expansions, we construct an asymptotic approximation Rε belonging to the Hilbert space Hε := {u ∈ H1(Ωε) : u = 0 on S(0) ∪ S(l)}. It is equal to R+ ε (x) := v+ 0 (x) + εχ0(r)N+ ( −r − a0 ε , x2 ε , ϕ, x2 ) , x ∈ Ω0; (3.1) and to R− ε (x) := v−0 (x) + ε ( Y (x2 ε ) ∂x2v − 0 (x) + χ0(r)N− ( −r − a0 ε , x2 ε , ϕ, x2 )) , x ∈ D, (3.2) where Y (ξ2) = −ξ2 + 1 2 + [ξ2]; χ0 ∈ C∞ 0 (R) is a cut-off function such that χ0(r) = { 1, |r − a0| ≤ σ/2, 0, |r − a0| ≥ σ, where σ is an enough small fixed positive number such that the functions h± are constant and equal on the segment [a0, a0 + σ]; N+(ξ, ϕ, x2) = Z1(ξ) (∂x2v + 0 (x))|r=a0 + ( Z2(ξ) + ξ1 ) (∂rv + 0 (x))|r=a0 , N−(ξ, ϕ, x2) = ( Z1(ξ) − Y (ξ2) ) (∂x2v + 0 (x))|r=a0 + ( Z2(ξ) + h−1ξ1 ) (∂rv + 0 (x))|r=a0 , ξ1 = −ε−1(r − a0), ξ2 = ε−1x2, r = ( x2 1 + x2 3 )1/2 , tan(ϕ) = x3/x1. 472 Asymptotic Solution... 3.1. Discrepancies in the Domain Ω0 Taking into account the properties of the functions Z1 and Z2, we conclude that R+ ε (x) = 0, x ∈ S(0) ∪ S(l), ∂rR + ε (x) = −∂ξ1Z1(0, x2/ε) ( ∂x2v + 0 (x) ) |r=a0 − ∂ξ1Z2(0, x2/ε) ( ∂rv + 0 (x) ) |r=a0 = 0 for any x ∈ ∂Ωε ∩ {r = a0}. Observing that [∆x̃ , χ0(r)]Y (x) = ∇x̃ · ( Y (x)∇x̃ χ0(r)) + ∇x̃Y (x) · ∇x̃ χ0(r), where [A,B] = AB − BA is the commutator of two operators A and B, and taking into account form (2.13) of Laplace’s operator, we get − ∆xR + ε (x) − fε(x) = f0(x) − fε(x) − χ0(r) ( ∂2 x2ξ2N+(ξ, ϕ, x2) − r−1∂ξ1N+(ξ, ϕ, x2) ) −∇ξN+(ξ, ϕ, x2) · ∇x̃ χ0(r) − ε ( ∇x̃ · ( N+∇x̃ χ0(r) ) + χ0(r)∂ 2 x2x2 N+(ξ, ϕ, x2) + r−2χ0(r)∂ 2 ϕϕN+(ξ, ϕ, x2) ) , ξ1 = −r − a0 ε , ξ2 = x2 ε , r = √ x2 1 + x2 3, tan(ϕ) = x3 x1 , x ∈ Ω0. (3.3) Further, the arguments of functions involved in calculations are in- dicated only if their absence may cause confusion. We multiply identity (3.3) by a test function ψ ∈ Hε and integrate by parts in Ω0 : − ∫ Qε ∂rR + ε ψ dSx + ∫ Ω0 ∇xR + ε · ∇xψ dx− ∫ Ω0 fεψ dx = I+ 1 (ε, ψ) + . . .+ I+ 4 (ε, ψ), (3.4) where Qε = Ωε ∩ {r = a0}, I+ 1 (ε, ψ) = ∫ Ω0 ( f0(x) − fε(x) ) ψ dx, I+ 2 (ε, ψ) = ∫ Ω0 χ0(r) ( r−1∂ξ1N+ − ∂2 x2ξ2N+ ) ψ dx, U. De Maio, T. A. Mel’nyk 473 I+ 3 (ε, ψ) = ε ∫ Ω0 N+ ∇x̃ χ0(r) · ∇x̃ψ dx− ∫ Ω0 ψ∇ξN+ · ∇x̃ χ0(r) dx, I+ 4 (ε, ψ) = ε ∫ Ω0 χ0(r) (∂x2ψ ∂x2N+ + r−2∂x2ψ ∂ϕN+) dx. 3.2. Discrepancies in the Thin Disks Denote by S+ j (ε) and S− j (ε) the right and left lateral surfaces of the thin disk Gj(ε) respectively; S+(ε) := N−1⋃ j=0 S+ j (ε), S−(ε) := N−1⋃ j=0 S− j (ε). It is easy to calculate that ( ∂rR − ε ) |r=a1 = 0, ∂rR − ε (x) = εY (x2/ε) ∂r(∂x2v − 0 (x)) + ∂rR + ε (x), x ∈ Qε, (3.5) and ∂νR − ε (x) = 1√ 1 + ε2|h′±(r)|2 ( −ε∇x̃(h±) · ∇x̃ ( v−0 + εY (ξ2)∂x2v − 0 ) ± ε ( Y (ξ2)∂ 2 x2x2 v−0 (x) + χ0(r)∂x2N−(ξ, ϕ, x2) )) , x ∈ S±(ε). (3.6) Putting R− ε in the differential equation of problem (1.1), we obtain − ∆xR − ε (x) − fε(x) = f0(x) − fε(x) + ∇x̃(lnh0) · ∇x̃v − 0 + χ0(r) ( r−1∂ξ1N− − ∂ξ2N− ) −∇ξN− · ∇x̃ χ0(r) − ε∂x2 ( Y (x2 ε ) ∂2 x2x2 v−0 + χ0(r) ( ∂x2N− ) |ξ2=x2/ε ) − ε ( Y (x2 ε ) ∆x̃ ( ∂x2v − 0 ) + ∇x̃ · ( N−∇x̃ χ0(r) ) + r−2χ0(r) ∂ 2 ϕϕN− ) , x∈G(ε). (3.7) Next we will use the following identity ∫ S± ε εh±(r)√ 1 + ε2|h′±(r)|2 ψ dSx = ∫ Gε ψ dx− ε ∫ Gε Y (x2 ε ) ∂x2ψ dx ∀ψ ∈ Hε. (3.8) To prove (3.8) it is enough to integrate by part the last integral. 474 Asymptotic Solution... Using (3.8) and taking into account the boundary values of ∂νR − ε (see (3.5), (3.6)), we multiply (3.7) by a test function ψ ∈ Hε and integrate by parts in Gε. This yields ∫ Qε ∂rR + ε ψ dSx+ ∫ Gε ∇xR − ε ·∇xψ dx− ∫ Gε fεψ dx = I−1 (ε, ψ)+. . .+I−5 (ε, ψ), (3.9) where I−1 (ε, ψ) = ∫ Gε ( f0(x) − fε(x) ) ψ dx, I−2 (ε, ψ) = ∫ Gε ψ χ0(r) ( r−1∂ξ1N− − ∂ξ2N− ) dx, I−3 (ε, ψ) = ε ∫ Gε N−∇x̃ χ0(r) · ∇x̃ ψ dx− ∫ Gε ψ∇ξN− · ∇x̃ χ0(r) dx, I−4 (ε, ψ) = ε ∫ Gε χ0(r) ( ∂x2ψ ∂x2N− + r−2∂ϕψ ∂ϕN− ) dx, I−5 (ε, ψ) = ε ∫ Gε Y (x2 ε )( ∇x ( ∂x2v − 0 ) · ∇xψ+ ∂x2 ( ψ∇x̃ ( lnh0 ) · ∇x̃v − 0 )) dx. 3.3. Asymptotic Estimates Summing (3.4) and (3.9), we see that the function Rε constructed by formulas (3.1) and (3.2) satisfies the following integral identity ∫ Ωε ∇xRε · ∇xψ dx− ∫ Ωε fεψ dx = Fε(ψ), ∀ψ ∈ Hε, where Fε(ψ) = I±1 (ε, ψ) + . . . + I±4 (ε, ψ) + I−5 (ε, ψ); I±i = I+ i + I−i , i = 1, . . . , 4. Since the weak solution uε ∈ Hε to problem (1.1) satisfies the integral identity ∫ Ωε ∇xuε · ∇xψ dx− ∫ Ωε fεψ dx = 0, ∀ψ ∈ Hε, we have ∫ Ωε ∇x ( Rε − uε ) · ∇xψ dx = Fε(ψ), ∀ψ ∈ Hε. (3.10) Now we are going to estimate the value Fε(ψ). U. De Maio, T. A. Mel’nyk 475 The sum I±1 (ε, ψ) is a linear bounded functional on Hε. Thus, |I±1 (ε, ψ)| = ‖fε − f0‖∗ ‖ψ‖H1(Ωε), where ‖fε − f0‖∗ = sup ψ∈Hε, ‖ψ‖ H1(Ωε) =1 |(fε − f0, ψ)L2(Ωε)|. Obviously, ‖fε − f0‖∗ ≤ C1‖fε − f0‖L2(Ωε). Here and in what follows, all constants in asymptotic inequalities are independent of the parameter ε. In order to estimate the terms I+ 2 (ε, ψ), I−2 (ε, ψ), we will use the following lemma. Lemma 3.1. Let N be an 1-periodic in ξ2 function belonging to the space L2(Π) and exponentially decreasing at infinity, i.e., there exist positive constants c,R, γ such that for any |ξ1| ≥ R ∣∣N (ξ) ∣∣ ≤ c exp(−γ|ξ1|). Then for any δ > 0 there exist positive constants c1, ε0 such that for all values ε ∈ (0, ε0) the following inequality is valid ∣∣∣∣∣ ∫ Ωε N ( −r − a0 ε , x2 ε ) ψ(x) dx ∣∣∣∣∣ ≤ c1ε 1−δ‖ψ‖H1(Ωε), ∀ψ ∈ Hε. Proof. Set Bε,δ = Ωε ∩ {x : |r − a0| ≤ ε1−2δ} for any δ > 0. Then ∣∣∣∣∣ ∫ Ωε N ( −r − a0 ε , x2 ε ) ψ dx ∣∣∣∣∣ ≤ ∣∣∣∣∣ ∫ Bε,δ N ( −r − a0 ε , x2 ε ) ψ dx ∣∣∣∣∣ + ∣∣∣∣∣ ∫ Ωε\Bε,δ N ( −r − a0 ε , x2 ε ) ψ dx ∣∣∣∣∣. The properties of the function N lead us to the conclusion that the second summand in this inequality decreases exponentially as ε → 0. Using Lemma 1.5 ([18]), we estimate the first summand: ∣∣∣∣∣ ∫ Bε,δ N ( −r − a0 ε , x2 ε ) ψ dx ∣∣∣∣∣ ≤ ( ∫ Bε,δ N 2 ( −r − a0 ε , x2 ε ) dx )1/2 ‖ψ‖L2(Bε,δ) ≤ c2 ε 1/2‖N‖L2(Π)c3ε −δ+1/2‖ψ‖H1(Ωε). The lemma is proved. 476 Asymptotic Solution... Since the functions ∂ξ1N±, ∂2 x2ξ2 N−, ∂ξ2N+ exponentially decrease as |ξ1| → +∞, we deduce from Lemma 3.1 that for any fixed δ > 0 |I+ 2 (ε, ψ) + I−2 (ε, ψ)| ≤ ε1−δ C2‖ψ‖H1(Ωε). (3.11) Integrals in I+ 3 (ε, ψ), I−3 (ε, ψ) are, in fact, over supp ( ∇x̃ χ0(r) ) ∩ Ωε = {x : σ/2 < |r − a0| < σ } ∩ Ωε, where, by virtue of Statement 2.1, the functions N−,∇ξN± are expo- nentially small and the function N+ is uniformly bounded with respect to ε. Thus, |I+ 3 (ε, ψ) + I−3 (ε, ψ)| ≤ εC3‖ψ‖H1(Ωε). Integrals in I+ 4 (ε, ψ), I−4 (ε, ψ) are over {x : |r − a0| < σ } ∩ Ωε and they can be estimated with extracting, if necessary, the exponentially decreasing part in the corresponding integrand and then with the help of the Cauchy–Bunyakovsky inequality. Consider for example the integral ∣∣∣∣∣ ∫ Ω0 χ0(r) ∂x2ψ ∂x2N+ dx ∣∣∣∣∣ = ∣∣∣∣∣ ∫ Ω0 χ0(r) ∂x2ψ ( Z1 (∂2 x2x2 v+ 0 (x))|r=a0 + ( Z2 − ε−1(r − a0) ) (∂x2∂rv + 0 (x))|r=a0 ) dx ∣∣∣∣∣ ≤ ∫ Ω0 χ0(r) |∂x2ψ| |Z1| |∂2 x2x2 v+ 0 (x)| ∣∣ r=a0 dx + ∫ Ω0 χ0|∂x2ψ| ( |Z2 − ε−1(r − a0) − ch| |∂x2∂rv + 0 (x)| ∣∣ r=a0 + |ch| |∂x2∂rv + 0 (x)| ∣∣ r=a0 ) dx ≤ c‖ψ‖H1(Ω0) (√∫ Ω0 χ0 |Z1|2 dx + √∫ Ω0 χ0 |Z2 − ε−1(r − a0) − ch|2 dx+ |ch| √ |Ω0| ) ≤ c‖ψ‖H1(Ω0) (√ 2πa0lε ∫ Π+ Z 2 1 (ξ) dξ + √ 2πa0lε ∫ Π+ Z 2 2 (ξ + ξ1 − ch) dξ + |ch| √ |Ω0| ) ≤ c‖ψ‖H1(Ω0) (√ ε ‖Z1‖L2(Π+) + √ ε ‖Z2+ξ1−ch‖L2(Π+) + |ch| √ |Ω0| ) , where |Ω0| is the measure of the domain Ω0. On the basis of (2.16) and (2.17) the value ‖Z1‖L2(Π+) and ‖Z2 + ξ1 − ch‖L2(Π+) are bounded. As a U. De Maio, T. A. Mel’nyk 477 result, we have |I+ 4 (ε, ψ) + I−4 (ε, ψ)| ≤ εC4‖ψ‖H1(Ωε). (3.12) Remark 3.1. The constants C2 and C3 in (3.11) and (3.12) respectively depend on the following quantities sup x∈Γ0 ∣∣Dα ( v+ 0 (x) )∣∣, |α| = α1 + α2 + α3 ≤ 2. (3.13) Applying the odd extension to the limit problem (2.20) with respect to the planes x2 = 0, x2 = l and taking into account the conditions for the function f0, we conclude that the function v+ 0 and its second derivatives have no singularities at the points on S(0) ∩ Γ0 and on S(l) ∩ Γ0. Thus, by virtue of classical results on the smoothness of solutions to boundary value problems, the quantities (3.13) are bounded. Since f0 ∈ H3(Ω1), the function ∂x2v − 0 ∈ H1(Ω1 \ Ω0). Therefore, |I−5 (ε, ψ)| ≤ εC5 ‖∂x2v − 0 ‖H1(Ω1\Ω0) ‖ψ‖H1(Ω1\Ω0). So, with regard to the inequalities obtained, we conclude that for the right-hand side in (3.10) the following inequality holds |Fε(ψ)| ≤ c(δ) ε1−δ‖ψ‖H1(Ωε), (3.14) where δ is an arbitrary fixed positive number. From (3.10) and (3.14) it follows the following results. Theorem 3.1. Suppose fε ∈ L2(Ωε), f0 ∈ H3(Ω1) and f0, ∂x2f0 vanish on S(0) ∪ S(l). Then for any δ > 0 there exist positive constants c1, ε0 such that for all values ε ∈ (0, ε0) the difference between the solution uε to prob- lem (1.1) and the approximation function Rε defined by (3.1) and (3.2) satisfies the following estimate ‖uε −Rε‖H1(Ωε) ≤ c1 ( ε1−δ + ‖fε − f0‖∗ ) . (3.15) Corollary 3.1. From (3.15) it follows that ‖uε − v0‖L2(Ωε) ≤ c2 ( ε1−δ + ‖fε − f0‖∗ ) , where v0 is the weak solution to the limit problem (2.20). Corollary 3.2. Assume fε(x) = f0(x) + εf1(x, ε), x ∈ Ωε, where the norm ‖f1(·, ε)‖L2(Ωε) = O(1) as ε → 0. Then for any δ > 0 there exist positive constants c3, ε0 such that for all values ε ∈ (0, ε0) ‖uε −Rε‖H1(Ωε) ≤ c3 ε 1−δ, ‖uε − v0‖L2(Ωε) ≤ c3 ε 1−δ. 478 Asymptotic Solution... Example. If the right-hand side f0 of the limit problem (2.20) depends only on the variable r and x2, then we can find the explicit solution in the domain D = {x : r ∈ (a0, a1), x2 ∈ (0, l)} and reduce problem (2.20) to a problem in the junction’s body Ω0. In this case the solution to the limit problem (2.20) depends only on the variable r and x2 as well. So, we can rewrite the limit problem in the following form −∂r ( r∂rv + 0 ) + r∂2 x2x2 v+ 0 = rf0(r, x2), x ∈ Ω0, −∂r ( rh0(r)∂rv − 0 ) = rh0(r)f0(r, x2), x ∈ D, ∂rv − 0 (a1, x2) = 0, x2 ∈ (0, l), v+ 0 (r, 0) = v+ 0 (r, l) = 0, x ∈ S(0) ∪ S(l), v+ 0 (a0, x2) = v−0 (a0, x2), x2 ∈ (0, l), ∂rv + 0 (a0, x2) = h0(a0) ∂rv − 0 (a0, x2), x2 ∈ (0, l), (3.16) By solving the ordinary equation of problem (3.16) in the domain D with regard to the Neumann condition at r = a1 and to the first transmission condition at r = a0 in the joint zone Γ0, we find that v−0 (r, x2) = v+ 0 (a0, x2) + r∫ a0 1 ρ h0(ρ) a1∫ ρ t h0(t) f0(t, x2) dt dρ. Now, according to the second transmission condition in problem (3.16), we obtain the classical mixed boundary-value problem −∂r ( r∂rv + 0 ) + r∂2 x2x2 v+ 0 = rf0(r, x2), x ∈ Ω0, v+ 0 (r, 0) = v+ 0 (r, l) = 0, x ∈ S(0) ∪ S(l), ∂rv + 0 (a0, x2) = F̂0(x2), x2 ∈ (0, l), (3.17) to find v+ 0 . Here F̂0(x2) = a−1 0 a1∫ a0 t h0(t)f0(t, x2) dt, x2 ∈ (0, l). Problem (3.17) is called resulting problem for problem (1.1). U. De Maio, T. A. Mel’nyk 479 Conclusion We assumed that the functions h− and h+ are locally constant and equal in an enough small neighborhood of the point a0. This is a technical condition which allows to avoid additional bulky calculations. Because of this, the junction-layer solutions are odd or even with respect to 1/2 (see Statement 2.1). As a result, the approximation function Rε sat- isfies exactly some boundary conditions and we do not need additional boundary-layer asymptotics. If the right-hand side has the following form fε = ∑∞ k=0 ε kfk(x), we can define the other terms in the asymptotic expansions (2.1), (2.2), (2.14) and construct an asymptotic approximation to any degree of ac- curacy. From results proved in the present paper it follows that for applied problems in thick junctions we can use the corresponding limit problem or resulting problem in the junction’s body, which are simpler, instead of the initial problem with the sufficient plausibility. References [1] Y. Amirat and O. Bodart, Boundary layer corrector for the solution of Laplace equation in a domain with oscillating boundary, Zeitschrift für Analysis und ihre Anwendungen // 20 (2001), N 4, 929–940. [2] Y. Amirat, O. Bodart, U. De Maio, and A. Gaudiello, Asymptotic approxima- tion of the solution of the Laplace equation in a domain with highly oscillating boundary, Preprint of Laboratoire de Mathematiques Appliquees Universite Blaise Pascal, Clemont-Ferrand, October, 2002 (to appear in SIAM, J. Math. Anal.) [3] N. Bachvalov and G. Panasenko, Homogenization: averaging processes in periodic media, Kluwer Academic Publishes, series “Mathematics and its Applications”. 1989. [4] A. Benkaddour and J. Sanchez-Hubert, Spectral study of a coupled compact- noncompact problem // M2AN, 26 (1992), N 6, 659–672. [5] A. Bensoussan, J. Lions and G. Papanicolau, Asymptotic analysis for periodic structure, Amsterdam: North Holland, 1978. [6] D. Blanchard, L. Carbone, and A. Gaudiello, Homogenization of a monotone problem in a domain with oscillating boundary // Mathematical Modelling and Numerical Analysis, 33 (1999), N 5, 1057–1070. [7] D. Blanchard and A. Gaudiello, Homogenization of highly oscillating boundaries and reduction of dimension for a monotone problem // ESAIM, Control. Optim. Calc. Var., 9 (2003) 449–460. [8] R. Brizzi and J. P. Chalot, Homogenization and Neumann Boundary Value Prob- lem // Ric. Mat. 46 (1997), 347–387. [9] P. G. Ciarlet, Plates and Junctions in Elastic Multi-structures, Masson, Paris, 1990. [10] D. Cioranescu and J. Saint Jean Paulin, Homogenization of reticulated structures, Springer-Verlag, series “Applied Mathematical Sciences” 136, 1999. 480 Asymptotic Solution... [11] U. De Maio and T. A. Mel’nyk, Homogenization of the Neumann problem in thick multi-structures of type 3:2:2 // Mathematical Methods in the Applied Sciences, 28 (2005), 865–879. [12] V. V. Jikov, S. M. Kozlov and O. A. Oleinik, Homogenization of differential operators and integral functionals, Springer-Verlag, Berlin, Heidelberg, 1994. [13] F. Fleury and E. Sanchez-Palencia, Asymptotic and spectral properties of the acoustic vibrations of body perforated by narrow channels // Bull. Sci. Math, 2 (1986), N 110, 149–176. [14] V. A. Kondrat’ev and O. A. Oleinik, Boundary-value problems for partial differ- ential equations in non-smooth domains // Uspekhi Mat. Nauk, 38 (1983), N 2, 3–76; English transl., Russian Math. Survays 38 (1983), N 2, 1–86. [15] V. P. Kotliarov and E. Ya. Khruslov, On a limit boundary condition of some Neumann problem // Theor. Funkts. i Funkts. Anal. Prilozhen, Izd-vo Kharkov Univ., 10 (1970), 83–96 (in Russian). [16] A. A. Kovalevsky, Averaging of the Neumann problems for nonlinear elliptic equa- tions in domains with accumulators // Ukrainian Math. J. 47 (1995), N 2, 227– 249. [17] V. A. Kozlov, V. G. Maz’ya and A. B. Movchan, Asymptotic representation of an elastic field in a multi-structure // Asymptotic Analysis, 11 (1995) 343–415. [18] V. A. Kozlov, V. G. Maz’ya and J. Rossmann, Elliptic boundary value problems in domains with point singulaities, Math. Surveys Monogr., 52, AMS, Providence, RI, 1997. [19] V. A. Marchenko and E. Ya. Khruslov, Boundary-value problems in domains with fine-grained boundary, Naukova Dumka, Kiev, 1974 (in Russian). [20] V. G. Maz’ya and B. A. Plamenevskii, The coefficients in the asymptotics of solutions of elliptic boundary value problem with conical points // Math. Nachr. 76 (1977), 29–60 (Russian). [21] T. A. Mel’nyk and S. A. Nazarov, Asymptotic structure of the spectrum of the Neumann problem in a thin comb-like domain, C. R. Acad. Sci., Paris // 319, Serie 1 (1994), 1343–1348. [22] T. A. Mel’nyk and S. A. Nazarov, Asymptotics of the Neumann spectral problem solution in a domain of “thick comb” type // Trudy Seminara imeni I. G. Petro- vskogo, 19 (1996), 138–173 (in Russian); and English translation: Journal of Math. Sci. 85 (1997), N 6, 2326–2346. [23] T. A. Mel’nyk, Homogenization of the Poisson equation in a thick periodic junc- tion // Zeitschrift für Analysis und ihre Anwendungen, 18 (1999), N 4, 953–975. [24] T. A. Mel’nyk, Asymptotic analysis of a spectral problem in a periodic thick junc- tion of type 3:2:1 // Mathematical Methods in the Applied sciences, 23 (2000), N 4, 321–346. [25] T. A. Mel’nyk and S. A. Nazarov, Asymptotic analysis of the Neumann problem of the junction of a body and thin heavy rods // Algebra i Analiz., 12 (2000), N 2, 188–238; English transl.: St.Petersburg Math.J., 12 (2001), N 2, 317–351. [26] T. A. Mel’nyk, Homogenization of a singularly perturbed parabolic problem in a thick periodic junction of the type 3:2:1 // Ukrainskii Matem. Zhurnal, 52 (2000), N 11, 1524–1534 (in Ukrainian); English translation: Ukrainian Math. Journal, 52 (2001), N 11, 1737–1749. U. De Maio, T. A. Mel’nyk 481 [27] T. A. Mel’nyk, Asymptotics of the eigenvalues and eigenfunctions of a bound- ary value problem in a thick periodic junction of type 3:2:2 // Bulletin of the University of Lviv, Series: Mathematics, 58 (2000), 153–160 (in Ukrainian). [28] T. A. Mel’nyk, Vibrations of a Thick Periodic Junction with Concentrated Masses // Mathematical Models and Methods in Applied Sciences, 11 (2001), N 6, 1001–1029. [29] S. A. Nazarov and B. A. Plamenevskii, Elliptic problems in domains with piecewise smooth boundaries, Walter de Gruyter, Berlin, 1994. [30] S. A. Nazarov The polynimial property selfadjoint elliptic boundary value problems and algebraic discribsion their atributes // Uspechi Mat. Nauk, 54 (1999), N 5, 77–142; English transl., Russian Math. Surveys, 54 (1999), N 5. [31] O. A. Oleinik, G. A. Yosifian, and A. S. Shamaev, Mathematical Problems in Elasticity and Homogenization, North-Holland, Amsterdam, 1992. [32] J. Sanchez-Hubert and E. Sanchez-Palencia, Vibration and Coupling of Continu- ous Systems, Springer-Verlag, Berlin, Heidelberg, 1989. [33] I. V. Skrypnik, Methods for investigation on nonlinear elliptic boundary-value problems, Nauka, Moscow, 1990 (in Russian). [34] G. V. Suzikov and E. Ya. Khruslov, On advancing sound waves through narrow channels in a reflecting layer // Theor. Funkts., Funkts. Anal. Prilozhen, Izd-vo Kharkov Univ., 5 (1967), 35–49 (in Russian). Contact information Umberto De Maio Department of Applied Mathematics “R. Caccioppoli” Federico II University of Naples Complesso Monte S. Angelo-Edificio “T” Via Cintia, 80126 Naples, Italy E-Mail: udemaio@unina.it Taras A. Mel’nyk Faculty of Mathematics and Mechanics Taras Shevchenko University of Kyiv Volodymyrska str. 64 01033 Kyiv, Ukraine E-Mail: melnyk@imath.kiev.ua