No label defined (Q4233754)

Let \(X\) be any (closed) subset of \(\mathbb{R}^n\). For \(k\in\mathbb{N} \cup \{\infty\}\), let \({\mathcal C}^k (X)=\{f: X\to\mathbb{R} \mid f=\widetilde f\) on \(X\) for some \({\mathcal C}^k\)-function \(\widehat f:\mathbb{R}^n \to\mathbb{R}\}\) and let \({\mathcal C}^{(\infty)} (X)=\cap_{k\in\mathbb{N}} {\mathcal C}^k(X)\). It is clear that \({\mathcal C}^\infty(X) \subset{\mathcal C}^{(\infty)} (X)\subset {\mathcal C}^k(X)\) for any \(k\in \mathbb{N}\). If \(n=1\), \textit{J. Merrien} [J. Math. Pures Appl., IX. Sér. 45, 291-309 (1966; Zbl 0163.06602)] showed that \({\mathcal C}^\infty(X)={\mathcal C}^{(\infty)}(X)\). If \(n>1\), it is, in general, not true which follows from a beautiful simple Example 1 of the paper under review. The author also discusses conditions under which the equality of both algebras \({\mathcal C}^\infty\) and \({\mathcal C}^{(\infty)}\) still holds. In particular, it is known that this is the case of \(X\) being a subanalytic subset of \(\mathbb{R}^n\) of dimension at most two or else of pure codimension one. In this paper, the author gives a (much more sophisticated) Example 2 of a subanalytic subset \(X\) of \(\mathbb{R}^5\) and \(f\in{\mathcal C}^{(\infty)} (X)\setminus {\mathcal C}^\infty(X)\). For the entire collection see [Zbl 0906.00013].