Irreducible components of the equivariant punctual Hilbert schemes (Q596610)

Fixed an algebraically closed field \(k\), the Hilbert scheme \({\mathbf H}\) parametrizing the zero-dimensional subschemes of the affine plane Spec \(k[x,y]\) is a disjoint union of its components \({\mathbf H}^l\) parametrizing the subschemes of length \(l\). The linear action of the torus \(k^*\times k^*\) on the polynomial ring \(k[x,y]\), defined by \((t_1,t_2) \cdot x^\alpha y^\beta=(t_1\cdot x)^\alpha (t_2\cdot y)^\beta\), induces an action on the Hilbert schemes \({\mathbf H}\) and \({\mathbf H}^l\). Given two relatively prime integers \(a\) and \(b\), one may consider the closed subschemes \({\mathbf H}_{ab}\subset{\mathbf H}\) and \({\mathbf H}_{ab}^l \subset{\mathbf H}^l\), parametrizing the zero-dimensional subschemes invariant under the action of the subtorus \(T_{ab}=\{(t^{-b}, t^a),\;t\in k^*\}\). Sometimes, information on \({\mathbf H}_{ab}\) may be lifted to \({\mathbf H}\). In fact \textit{G. Ellingsrud} and \textit{S. A. Strømme} [Invent. Math. 87, 343--352 (1987; Zbl 0625.14002)] have computed the Chow group of \({\mathbf H}^l\) examining the embedding \({\mathbf H}^l_{ab}\subset {\mathbf H}^l\) for general \((a,b)\), and \textit{M. Brion} [Transform. Groups 2, No. 3, 225--267 (1997; Zbl 0916.14003)] has described the equivariant cohomology of \({\mathcal H}^l\) in terms of the equivariant cohomology of all the \({\mathbf H}^l_{ab}\). This accounts for the interest in studying the schemes \({\mathbf H}_{ab}\). In the paper under review, the author determines the irreducible components of the schemes \({\mathbf H}_{ab}\) (by \textit{R. Hartshorne} [Publ. Math., Inst. Hautes Étud. Sci. 29, 5--48 (1966; Zbl 0171.14502)] and \textit{J. Fogarty} [Am. J. Math. 70, 511--521 (1968; Zbl 0176.18401)] one already knows that the smooth subschemes \({\mathbf H}^l\) are the irreducible components of \({\mathbf H})\). The basic and natural idea is that one may separate \({\mathbf H}_{ab}^l\) into disjoint subschemes by fixing a Hilbert function. More precisely, define the degree \(d\) of a monomial by the formula \(d(x^\alpha y^\beta)=-b\alpha +a\beta\). Then a subscheme \(Z\subset\text{Spec}\,k[x,y]\) is in \({\mathbf H}_{ab}\) if and only if its ideal \(I(Z)\) is quasi-homogeneous with respect to \(d\), i.e. \(I(Z)= \bigoplus_{n\in\mathbb{Z}}I(Z)_n\), where \(I(Z)_n=I(Z)\cap k[x,y]_n\) and \(k[x,y]_n\) denotes the vector space generated by the monomials \(m\) of degree \(d(m)=n\). One defines the numerical sequence \(H=(\text{codim} (I(Z)_n\), \(k[x,y]_n))_{n\in \mathbb{Z}}\) as the Hilbert function of \(Z\). Now, fix any sequence of integers \(H=(h_n)_{n\in \mathbb{Z}}\), and denote by \({\mathbf H}_{ab}(H)\) the (possibly empty) closed subscheme of \({\mathbf H}_{ab}\) parametrizing the subschemes \(Z\) whose Hilbert function is \(H\). It turns out that \({\mathbf H}_{ab}\) is the disjoint union of the subschemes \({\mathbf H}_{ab}(H)\), and the main result of the paper consists in proving that, when it is not empty, then \({\mathbf H}_{ab}(H)\) is smooth and irreducible. The methods involved in the proof allow the author to recover some of the results appearing in previous papers of \textit{G. Ellingsrud} and \textit{S. A. Strømme} [loc. cit.; Invent. Math. 91, No. 2, 365--370 (1988; Zbl 1064.14500)].