Shape optimization by the homogenization method (Q5950926)

During the past two decades we have seen a revived interest in theoretical study of the macroscopic behaviour of microscopically heterogeneous materials, partly driven by applications in the optimal design. The book under review presents a comprehensive introduction to the homogenisation method applied to optimal design, including many proofs which were hitherto only scattered throughout the literature. While primarily based on the results of the French school, where the author's contribution has been significant, the results of other schools were presented as well. The physical motivation stems from the problems originating in conductivity theory and linearised elasticity. It must be stressed that in comparison to the above mentioned treatises, this one provides the most complete treatment of numerical methods. The book is divided into five chapters; the first offers an introduction to the homogenisation theory. Brief review of periodic homogenisation is followed by general theory of H-convergence and G-convergence, discussing the topics as div-curl lemma, Tartar's oscillating test function method, compactness and locality of H-convergence, energy convergence, correctors, homogenisation of eigenvalue problems and laminated structures. The mathematical modelling of two-phase composite materials is the topic of second chapter. Using the fundamental construction of sequential laminates, the optimality of Hashin-Shtrikman bounds is proved for two well-ordered phases in both conductivity and linearised elasticity (only for isotropic materials in this case) setting. Particular attention has been paid to the G-closure problem including a proof of the known result of DalMaso and Kohn on the local character of G-closure for two-phase materials. In the conductivity setting, with two isotropic materials, a characterisation of G-closure is proved in detail. Chapters three and four are concerned with the optimal design in conductivity and elasticity respectively, following the same programme. As there is no classical solution to the optimal design problem, for which some counterexamples are presented, a relaxation is necessary. The homogenised materials turn out to be the right answer to this need. Using it, necessary conditions of optimality are obtained in several settings, including the multiple state equation for conductivity. For elasticity, due to the lack of a good characterisation for G-closure, the results are obtained only for compliance optimisation, which is of particular importance in applications, allowing the author to write down the necessary conditions of optimality, important for numerical treatment. Shape optimisation problems are explained as the limit of two phase problems, where one phase degenerates as to become void. The final chapter is devoted to numerical algorithms based on the above described methods, paying attention to convergence questions. A number of realistic examples, mostly for elasticity, has been developed in detail. The results for both conductivity and elasticity have been separately developed in chapters three and four. A unified treatment might have better put to fore the similarities and differences, while offering some economy in the exposition. While nicely written, the first chapter still requires from the reader certain familiarity with weak convergence methods in partial differential equations. For a better prepared reader, we would recommend reading it in order to familiarise oneself with the notation and terminology, while for a novice, we suggest to have an introductory text at hand, in order to clarify certain points. We hope that in a subsequent edition the publisher will better fit the text on the pages (the margins are barely 1cm wide). In conclusion, we would like to warmly recommend this book to anybody working in optimal shape design, composites and homogenisation, as well to those who wish to enter these fields.