Recovering the good component of the Hilbert scheme (Q2453586)

The Hilbert scheme \({\text{Hilb}}^n_{X/S}\) parametrizing length \(n\) closed subschemes of \(X\) over \(S\) continues to draw great interest from algebraic geometers. The case \(X = \mathbb P^N\) is already interesting, as \({\text{Hilb}}^n_{\mathbb P^N/k}\) is smooth and irreducible for \(N=2\), but reducible for \(N = 3\) and \(n\) large [\textit{A. Iarrobino}, Invent. Math. 15, 72--77 (1972; Zbl 0227.14006)] and hence singular by \textit{R. Hartshorne}'s connectedness theorem [Publ. Math. Inst. Hautes Études Sci. 29, 5--48 (1966; Zbl 0171.41502)]. Motivated by \textit{Haiman's} construction of \({\text{Hilb}}^n_{\mathbb A^2/\mathbb C}\) as the blow-up of \(\text{Sym}^n (\mathbb A^2)\) at a concrete ideal [Discrete Math. 193, No. 1--3, 201--224 (1998; Zbl 1061.05509)], the authors construct the \textit{good component} \(G^n_{X/S} \subset {\text{Hilb}}^n_{X/S}\), the closure of subschemes consisting of \(n\) distinct points, as a concrete blow-up of a symmetric product for separated morphisms \(f:X \to S\) of algebraic spaces. Working at this level of generality, the authors need to show existence of the Hilbert scheme as an algebraic space, extending \textit{M. Artin's} result [in: Global Analysis, Papers in Honor of K. Kodaira 21--71 (1969; Zbl 0205.50402)] for \(f\) locally of finite presentation. In this context, the \(n\)th symmetric product need not exist, so instead they use the \(n\)th divided power product \(\Gamma^n_{X/S}\) (the affine model is due to \textit{N. Roby} [C. R. Acad. Sci., Paris, Sér. A 290, 869--871 (1980; Zbl 0471.13008)]), which is homeomorphic to \(\text{Sym}^n X\) in general and isomorphic when \(f\) is flat [\textit{D. Rydh}, ``Families of zero cycles and divided powers: I. Representability'', \url{arXiv:0803.0618}]. When \(X\) and \(S\) are affine, the authors define the \textit{ideal of norms} \(I\) in the ring corresponding to \(\Gamma^n_{X/S}\) and show that these patch together to define a closed subscheme \(\Delta_X \subset \Gamma^n_{X/S}\). With this machinery in place, the authors prove that if \(f: X \to S\) is a separated morphism of algebraic spaces, then \(G^n_{X/S} \subset {\text{Hilb}}^n_{X/S}\) is isomorphic to the blow-up of \(\Gamma^n_{X/S}\) along the closed subspace \(\Delta_X\). In the important case that \(f\) is flat, \(G^n_{X/S}\) is obtained by blowing up the geometric quotient \(X^n_S/S_n\). As a byproduct of their method, they show that \(G^n_{X/S} = {\text{Hilb}}^n_{X/S}\) is smooth for \(f\) smooth and separated of relative dimension two, extending the result of \textit{J. Fogarty} [Am. J. Math. 90, 511--521 (1968; Zbl 0176.18401)].