A smoothness test for higher codimensions (Q1680162)

Let \(K\) be a field of characteristic zero, \(x=(x_1, \ldots, x_n)\) and \(X\subseteq W\) affine algebraic varieties, \(W\) smooth, given by the radical ideals \(I_W\subseteq I_X\subseteq K[x]\). For \(p\in W\) let \(\text{ord}_p(I_X):=\max\{m\;|\;I_X\subseteq \mathfrak{m}^m_{W,p}\}\). Let \(I_X=\langle f_1, \ldots, f_s\rangle\) and \(\Delta(I_X)\) the ideal sheaf locally generated by \(f_1, \ldots, f_s\) and all partial derivatives thereof with respect to a regular system of parameters of \(W\). The order of \(I_X\) is at most \(1\) everywhere iff \(1\in \Delta(I_X)_w\) for all \(w\in W\). The aim of the paper is to give a smoothness test for \(X\) withour using the Jacobian criterion. The idea is based on the algorithmic version of Hironaka's proof of resolution on singularities given by \textit{A. M. Bravo} et al. [Rev. Mat. Iberoam. 21, No. 2, 349--458 (2005; Zbl 1086.14012)]. The algorithm is more efficient than the standard method for testing smoothness using the Jacobian criterion if the number of minors is very large. The test assumes additionally that \(W\) is a complete intersection on the open set \(D(g)\) for some \(g\in K[x]\). This is no restriction since non-singular varieties are locally complete intersections. The algorithm for the smoothness test starts checking if \(I_W=I_X\) on \(D(g)\). Then checking the necessary condition for smoothness: \(1\in \Delta(I_X)\) on \(D(g)\). The main step is a descent of the embedding similar to the algorithm of resolution of singularities. A list of triples \((I_{Z_i}, I_{X|Z_i}, g_i)\) is computed such that \(Z_i\) contains the support of \(I_X\) in \(D(g_i)\) and is a non-singular complete intersection and \(I_X\) is the restriction of \(I_X\) to \(Z_i\cap G(g_i)\). Then the algorithm is recursively applied to the \((I_{Z_i}, I_{X|Z_i}, g_i)\). Parallel and hybrid (i.e. using the Jacobian criterion at a certain moment in the descent) versions of the algorithm are discussed. The algorithm is implemented in the computer algebra \textsc{Singular}. Comparisons and timings are given.