Polynomial bound on the local Betti numbers of a real analytic germ (Q1687868)

The paper proves the existence of a bound on the sum of local Betti numbers of a real analytic germ. The bound is given as a polynomial function of the multiplicity of the germ. It is extremly pleasurable to read the paper. In the introduction the reader is really introduced to the problem and warned that such a bound doesn't always exist. It is announced that Theorem 1.13 will give a set of sufficient conditions on the dimension of the singular locus of the tangent cone for such a bound to exist. Theorem 1.17 then shows that these conditions are optimal by giving a family of analytic germs with constant multiplicity and with arbitrarily large 0th Betti number. Of course, the germs of this family do not meet the sufficient conditions mentioned above. Section 1 is entitled `Statement of Main Results', but contains more than that. Local Betti numbers and tangent cones are introduced there, but not just by tossing in some definitions; there is a lot of geometric intuition and excellent explanations, examples and remarks (among them a very appreciated by the reviewer Remark 1.8 that evokes the paper of H. Whitney). Then the main result is discussed, again in a way that is precise and intuitive at the same time. In Section 2 a proof of Theorem 1.13 is given. It is very geometrical and the article is illustrated by some very much appreciated drawings. In Section 3 the optimality of the conditions is proved, once again in a very clear way, with illustrations. In Section 4 the bound is applied to some interesting invariants, among which the Kurdyka-Raby density of subanalytic sets. The introduction informs the reader which invariants will be discussed in Section 4. The proofs are very ingenious, much in the spirit of René Thom and his school. The reviewer is most impressed by the remarks concerning actual computation (like Remark 1.12 that refers the reader to [\textit{J.-C. Faugère}, in: Proceedings of the 2002 international symposium on symbolic and algebraic computation, Lille, France, July 2002. New York, NY: ACM Press. 75--83 (2002; Zbl 1072.68664)]. Such a thing is very rare in papers concerning pure mathematics.) It is a very good thing that the journal (Ann. Inst. Fourier, Grenoble) didn't ask, as many journals do nowadays, for a shortening of the paper. Thanks to this we have here not only an interesting, important, applicable result, but such a nice reading! For the curious: at the end of the paper the author thanks his mentors and it becomes obvious that he comes from a very good school.