On Gabrielov's regularity condition for analytic mappings (Q1187738)

Gabrielov introduced three kinds of analytic invariants for an analytic mapping \(\varphi:X\to Y\), where \(X\) and \(Y\) are analytic manifolds over \(K=\mathbb{R}\) or \(\mathbb{C}\): \(r_ 1(x)=\) the generic rank of \(\varphi\) near \(x\), \(r_ 3(x)=\) the Krull dimension of \({\hat{\mathcal O}}_{Y,\varphi(x)}/\text{Ker} \hat\varphi^*_ x\), \(r_ 2(x)=\) the Krull dimension of \({\mathcal O}_{Y,\varphi(x)}/\text{Ker} \varphi^*_ x\), with \(\varphi_ x^*:{\mathcal O}_{Y,\varphi(x)}\to{\mathcal O}_{X,x}\) and ``\(^ \wedge\)'' standing for completion. It is easy to see that \(r_ 1(x)\leq r_ 2(x)\leq r_ 3(x)\). \textit{E. Bierstone} and \textit{P. D. Milman} studied these invariants [Ann. Math., II. Ser. 116, No. 3, 541-558 (1982; Zbl 0519.58003), Proc. Symp. Pure Math. 40, Part 1, 127-136 (1983; Zbl 0519.58004), Ann. Inst. Fourier 37, No. 1, 187-239 (1987; Zbl 0611.32002), and Ann. Inst. Fourier 37, No. 2, 49-77 (1987; Zbl 0611.32003)] and introduced the notion of the mapping \(\varphi\) being regular and \(x\) if \(r_ 1(x)=r_ 3(x)\). They used this notion to prove the composition hypothesis of Glaeser. The author proves that the set \(nR\varphi=\{x\in X:\varphi\) is not regular at \(x\}\) is nowhere dense and analytic in \(X\) (Theorem 1). He also constructs a local filtration by analytic subsets \(U=X_ 0\supset X_ 1\supset\cdots\supset X_{s+1}=nR(\varphi)\cap U\) such that, for each \(j=0,\dots,s\), the sheaf of ideals \(\text{Ker} \varphi^*\) is finitely generated over \(X_ j-X_{j+1}\) (Theorem 2). Both results are difficult and useful.