On globally defined semianalytic sets (Q330871)

Developing Cartan's concept of \(C\)-analytic sets, the authors introduce a new class of real sets, \(C\)-semianalytic sets and show that they have the same boolean properties as subanalitic sets, the property of good projections included. By real sets we mean that the sets sit in a real analytic space or a real analytic manifold. All the properties of \(C\)-semianalytic sets are proved in the article. This new class of sets contains all coherent real analytic sets, but contains Whitney's umbrella too. Not every semianalytic set is \(C\)-semianalytic (counterexamples are given). \(C\)-semianalytic sets are globally described, thus the title of the article. The reviewer finds the Theorem 1.5 of the paper especially interesting. Among other things, it states that every subanalytic subset of a real analytic manifold is a proper analytic image of a basic \(C\)-semianalytic set. This is proved using the uniformization theorem (quoted as Theorem 6.5). It would be interesting to know under what assumptions the manifold in which the projected basic \(C\)-semianalytic set sits has a finite number of connected components. As it is, this manifold is obtained as a countable union of compact connected components. There are other interesting results in this article. For instance, Corollary 1.4, which states that the set of the points of non coherence of a \(C\)-analytic set \(X\) is a \(C\)-semianalytic set of dimension not greater than \(\dim X-2\). Reviewer's remark: The article is closely related to the problem of projections of relatively compact sets, whether they are analytic, \(C\)-analytic, semianalytic, subanalytic or \(C\)-semianalytic, or else definable. An attempt of giving a panorama of this problem is made, but it hasn't been very successful, as the paper isn't exactly reader friendly. First of all, the reader has to seek definitions in the text, as the word ``Definition'' is never used. Then, even if the bibliography is large, it is not complete, especially as far as subanalytic sets are concerned. As the authors mention, quoting the famous example of Osgood, a projection of a relatively compact semianalytic set needn't be semianalytic. They rightly remind that this was the reason for which subanalytic sets (which are locally projections of relatively compact semianalytic sets) were introduced. But they don't mention a quite important fact that while the theory of semianalytic sets is that of S. Łojasiewicz, the theory of subanalytic sets has three independent fathers: \textit{A. M. Gabrielov} [Funct. Anal. Appl. 2, 282--291 (1968; Zbl 0179.08503); translation from Funkts. Anal. Prilozh. 2, No. 4, 18--30 (1968)], \textit{H. Hironaka} [Astérisque 7--8, 13--20 (1974; Zbl 0287.14005)] and Łojasiewicz (no quotation). The property of subanalytic sets that was most difficult to prove was that of the complement of a subanalytic set being subanalytic. Hironaka proved it using his desingularization theorem. Gabrielov proved it in an elementary way (he was the first). Łojasiewicz and his group proved it using Łojasiewicz's normal partitions, yet not a single work of Łojasiewicz concerning subanalytic sets is quoted. It is worth noticing that the work of \textit{E. Bierstone} and \textit{P. D. Milman} [Publ. Math., Inst. Hautes Étud. Sci. 67, 5--42 (1988; Zbl 0674.32002)] presents (in its first, survey part) Łojasiewicz's approach to subanalytic sets, nicely clarified on the occasion. The full Łojasiewicz's approach can be found either in original works (Bull. Acad. Sci. Pol.) or in the textbook Denkowska, Stasica ``Ensembles sous-analytiques à la polonaise'' or in a short survey in Spanish written by Łojasiewicz himself. Definable sets are mentioned several times, but the basic bibliography concerning them isn't given either (van de Dries, Coste\dots). This makes the paper rather difficult to read for those who are not already familiar with all these theories and it is a pity because, as said above, the results obtained are interesting.