On equiresolution and a question of Zariski (Q5926360)

The Zariski stratification of a hypersurface \(V\) in a smooth complex variety \(W\) is a partition of \(V\) into smooth locally closed subvarieties \(Y_i\) such that, locally at \(x\in Y_i\), \(V\) is equisingular and equimultiple along \(Y_i\), and the \(Y_i\) are maximal with respect to these properties. The question addressed in this paper is stated in the article of \textit{J. Lipman} [In: Prog. Math. 181, 485-505 (2000; Zbl 0970.14011)]: given a Zariski stratum \(Y\subset V\) and a point \(x\in Y\), is \(V\) equiresolvable along \(Y\) locally at~\(x\)? Equisingularity in the sense of Zariski is defined inductively: two smooth points are equisingular, and \(x\in V\) and \(y\in V\) are equisingular if \(\beta(x)\) and \(\beta(y)\) are equisingular as points of the reduced discriminant locus \(\Sigma_\beta\), where \(\beta\) is a generic smooth projection of relative dimension~\(1\). The meaning of the term ``equiresolution'' is less settled. Following Lipman [loc. cit.] we say that \(\text{ER}(x,V\subset W, Y)\) holds at \(x\in Y\) if there is a smooth morphism \(\pi:W\to Z\) such that \(\pi|_V\) is a family of reduced hypersurfaces and an embedded resolution \(f:W'\to W\) of \(V\) defining a family of embedded resolutions of the fibres of~\(\pi\), and \(f^{-1}(Y)\) is a union of exceptional divisors. This last condition amounts in practice to requiring that \(f^{-1}(Y)\) be of pure codimension~\(1\). Note that we do not require \(f\) to be a sequence of blowups. Indeed, an example of \textit{I.~Luengo} [Math. Ann. 267, 487-494 (1984; Zbl 0539.14007)] shows that we cannot expect to achieve equiresolution under such restrictions. The main technical theorem of the paper is that if \(\beta:W\to W_1\) is a suitably general smooth morphism of relative dimension~\(1\) and \(\Sigma_\beta\) is the reduced discriminant then \(\text{ER}(x,\Sigma_\beta\subset W_1, \beta(Y))\) implies \(\text{ER}(x,V\subset W, Y)\). Applying this to a Zariski stratum~\(Y\) gives an affirmative answer to the question above. In view of Luengo's example a Hironaka-style approach to constructing~\(f\), by blowing up, cannot succeed directly. Instead the author turns to procedures similar to those of Jung and Abhyankar, among others, using an abelian Galois cover to separate \(V\to W_1\) into a union of sections (locally) and then carefully blowing up the locus of intersection of these sections. There remains a problem of resolving toric quotient singularities in a compatible way, which is solved by explicit calculations. This procedure is in its essentials not very different from the method described e.g. by \textit{K.~Paranjape} [In: Lond. Math. Soc. Lect. Note Ser. 264, 347-358 (1999; Zbl 0958.14007)]; but it is far from clear a priori how to make such an approach succeed. By doing so the paper greatly clarifies both how equiresolution ought to be thought of and when it can be expected to be attainable.