Semialgebraic sets and the Łojasiewicz-Siciak condition (Q327606)

This is an important paper, as it gives a full characterization of a very useful class of sets. The reviewer will call them (ŁS) sets for short. We are speaking about the subsets of the complex plane which fulfil the following Łojasiewicz-Siciak condition: \(\Phi_K(z)\geq 1+\mu(\operatorname{dist}(z;K))^K\) for \(z\in\mathbb C^N\), \(\operatorname{dist}(z;K)\leq 1\) where \(K\) is the Siciak extremal function: \[ \Phi_K(z):=\sup\big\{|p(z)|^{1/\deg p}:p\in\mathbb C[z_1,\dots,z_N], \;\deg p>0\text{ and }\| p\|_K\leq 1\big\}. \] Theorem 1.3 of the present paper gives a full geometric characterisation of compact, semi-algebraic (ŁS) sets in the complex plane. So far very few (ŁS) sets were known (as remarked by \textit{L. Gendre} in [Inégalités de Markov singulières et approximation des fonctions holomorphes de la classe \(M\). Toulouse: Université Toulouse Paul Sabatier (PhD Thesis) (2005)], mostly discs and cubes). This is related to the fact that few sets have an explicit Siciak extremal function. At the same time such sets are important, for instance in the theory of approximation (works of L. Gendre, J. Siciak, W. Plesniak, W. Pawlucki and the author are quoted in the comprehensive bibliography). As we teach our students, in the real domain, continuous functions can be very nicely approximated by polynomials. In the complex domain, things are not that simple. Even continuous functions on a circle are not good enough. Surprisingly, compact polynomially convex sets are not regular enough for many purposes, even if we add (as the author did) the condition of being semi-algebraic in \(\mathbb R^2\). The paper begins with an introduction, which is really helpful for the reader. Many interesting examples are given throughout the paper. Theorem 1.3. is proved in Chapter 2 and the proof is very elegant, although far from obvious. In Chapter 1, some more results are formulated (Theorems 1.5 and 1.6, this time in \(\mathbb C^n\), give sufficient conditions for the inverse image by a polynomial mapping and a union of compact (ŁS) sets that are also semi-algebraic to be (ŁS) again). Some examples explain why certain hypotheses in these theorems cannot be omitted. In Chapter 3, the author presents subanalytic and definable versions of his theorems, explaining very clearly their meaning. In the appendix, the constraint in Theorem 1.6 (the set needs to be non-pluripolar) is slightly relaxed. The paper is self-contained and is a really nice read.