Separating sets, metric tangent cone and applications for complex algebraic germs (Q5962307): Difference between revisions

Let \(X\subseteq \mathbb{R}^n\) be a \(k\)-dimensional semialgebraic set and \(x_0\in X\) such that the link of \(X\) at \(x_0\) is connected and the \(k\)-density of \(X\) at \(x_0\) is positive. A \((k-1)\)-dimensional closed rectificable subset \(Y\subset X\) with \(x_0\in Y\) is called a separating set of \(X\) at \(x_0\) if there is \(\varepsilon >0\) such that {\parindent=6mm\begin{itemize}\item[(1)] \((B_\varepsilon(x_0)\cap X)\smallsetminus Y\) has at least two connected components \(A\) and \(B\), \item[(2)] the superior \((k-1)\)-density of \(Y\) at \(x_0\) is zero, \item[(3)] the inferior \(k\)-densities of \(A\) and \(B\) at \(x_0\) are non-zero. \end{itemize}} Let \((X,0)\subseteq (\mathbb{C}^3,0)\) be a quasihomogeneous surface singularity and \(z\) the variable of lowest weight. If \(X\cap \{z=0\}\) has more than one branch at the origin then \((X,0)\) has a separating set. This implies especially that \(X_t=\{(x,y,z)\in\mathbb{C}^3\;|\;x^5+z^{15}+y^7z+txy^6=0\}\) has a separating set at the origin if \(t\neq 0\). It is proved that \(X_0\) does not have a separating set. This shows that the inner bi-Lipschitz type of a normal surface germ is not determined by the topological type, even in a family of constant topological type. If the tangent cone of an isolated singularity \((X,0)\) has a non-isolated singularity and the non-isolated locus is a separating set of the tangent cone then \((X,0)\) has a separating set. The converse is not true.