Bloch wave homogenization of a non-homogeneous Neumann problem (Q2461530)

In this paper, the authors revisit a homogenization problem and derive a rigorous convergence result using Bloch wave decomposition. Specifically, the problem under consideration is the homogenization of periodically pierced (or perforate) media in bounded domains of \(\mathbb R^N\) and with non-homogeneous Neumann conditions on the boundary of the holes. Their convergence proof deals with the asymptotic behavior of the solution to Laplace equation in such domains as the size of the holes vanishes more rapidly than the small parameter \(\varepsilon\), which represents the periodicity. In this respect the paper is an important contribution to the understanding of various interesting phenomena (such as occurrence of strange terms at the limit) typically arising in the homogenization of perforate media with small and very small holes. Of course, it is based on the existing research in this field, however, using Bloch wave approach, as well as witty harmonic analysis techniques to treat strongly oscillating integrals, together with a clever combination of all above elements, make this piece of research considerably original. This may be a first attempt to deal with this kind of difficulties for the Bloch wave approach in homogenization. In a previous work [\textit{C. Conca} and \textit{P. Donato}, RAIRO, Modélisation Math. Anal. Numér. 22, No.~4, 561--607 (1988; Zbl 0669.35028)], it has been proven, depending on how small the holes are with respect to \(\varepsilon\), that in the homogenized equation can emerge a non expected new term, which is referred to as the ``strange term'' in the literature. In section \S4, the ideas and methods mentioned above have been combined and re-applied to produce a dual approach which gives rise, and clearly justifies, the appearance of the strange term in their situation (non-homogeneous Neumann boundary conditions on the boundary of the holes). This work certainly shows that the Bloch wave decomposition method is essentially independent of asymptotic expansion techniques. The authors provide useful explanations and motivations in appropriate places.