Homogenization of non-linear variational problems with thin low-conducting layers (Q870503): Difference between revisions
From MaRDI portal
Removed claim: reviewed by (P1447): Item:Q536424 |
Changed an Item |
||
Property / reviewed by | |||
Property / reviewed by: Alain Brillard / rank | |||
Normal rank |
Revision as of 00:23, 16 February 2024
scientific article
Language | Label | Description | Also known as |
---|---|---|---|
English | Homogenization of non-linear variational problems with thin low-conducting layers |
scientific article |
Statements
Homogenization of non-linear variational problems with thin low-conducting layers (English)
0 references
12 March 2007
0 references
The purpose of the paper is to prove some homogenization result for non-linear functionals of the type \(\mathcal{F}_{\varepsilon }(u)=\int_{\Omega }f_{\varepsilon }(x/\varepsilon ,Du)\,dx\), defined on \( W^{1,p}(\Omega ;\mathbb R^{m})\), where \(f_{\varepsilon }(y,\lambda )\) is periodic with respect to \(y\) and \(p\)-homogeneous with respect to \(\lambda\), plus some further \(p\)-growth conditions and a local Lipschitz continuity condition. \(f_\varepsilon\) is supposed to behave like as \(a_{\varepsilon }(x)| \lambda | ^{p}\) for some function \(a_{\varepsilon }\) which is equal to 1 in hard phases and tends to 0 in soft phases. The underlying domain \(\Omega \) is indeed the union of hard and soft phases. The hard phases are shared in \( N\geq 2\) two by two disjoint-closure regions \(E_{i}^{\varepsilon }\) of conductivity given by \(f_{\varepsilon }\) supposed to be here independent of \( \varepsilon \), these regions being separated by thin layers region \(\omega ^{\varepsilon }\) with conductivity \(a_{\varepsilon }\). The authors use the \( \Gamma \)-convergence method in order to describe this asymptotic behaviour. The topology associated to this problem corresponds to a multi-phase limit \( (u_{1},\dots ,u_{N})\) in some strong \(L^{p}\)-sense. The authors prove that the limit energy can be decomposed as \[ \sum_{i}\int_{\Omega }f_{i}^{\hom }(Du_{i})\,dx+\int_{\Omega }\phi (u_{1},\cdots u_{N})\,dx, \] for every \(u_{i}\in W^{1,p}(\Omega ,\mathbb R^{n})\). Here \(f_{i}^{\hom }\) can be computed through a classical homogenization procedure on a \(K\)-cell (\(K\in \mathbb N ^{\ast }\)), then letting \(K\) grow to \(+\infty \). The functional \(\phi \) can be defined using the mean of \(f_{\varepsilon }\) on a \(K\)-cell, then letting \( \varepsilon \) tend to 0 and \(K\) grow to \(+\infty \). The number of phases can be evaluated computing the \(\Gamma \)-limit of the functional associated to the density \(\varepsilon ^{-p}a_{\varepsilon }| Dv| ^{p}\) on the torus. The proof is first based on the determination of the domain of the limit energy, through adapted Poincaré-Wirtinger inequalities, then building appropriate test-functions requiring cut-off functions. The paper ends with an example with intricated connected channels.
0 references
\(\Gamma\)-convergence
0 references
multi-phase limit
0 references
double-porosity
0 references
adapted Poincaré-Wirtinger inequalities
0 references