The Douady space of a complex surface (Q1570235): Difference between revisions

Let \(X\) be a complex surface (i.e. a smooth, connected complex-analytic surface) and let denote by \(X^{[n]}\), for any non-negative integer \(n\), the Douady space of zero-dimensional analytic subspaces of \(X\) of length \(n\) (if \(X\) is algebraic then \(X^{[n]}\) is the usual Hilbert scheme). In this paper the authors give: i) First, a self-contained explicit construction of \(X^{[n]}\) for \(X={\mathbb C}^2\) and \(X=\Delta:=\{(z_1,z_2)\in {\mathbb C}^2\) s.t. \(|z_i|<1\), \(i=1,2\}\) the unit bi-disk in \({\mathbb C}^2\), as well as an explicit description, following some results of Nakayima, of the Douady-Barlet morphism \(\pi : X^{[n]} \rightarrow X^{(n)}\) for \(X={\mathbb C}^2,\Delta\). Here `self-contained' means independent of the general existence result of Douady. ii) Then, the construction of the connected \(2n\)-dimensional complex manifold \(X^{[n]}\) and the projective morphism \(\pi : X^{[n]} \rightarrow X^{(n)}\) for any complex surface \(X\), by patching \(\Delta^{[m]}\) and \(\pi : \Delta^{[m]} \rightarrow \Delta^{(m)}\) for \(m\leq n\). A fundamental result of the paper is the one realizing (for \(X={\mathbb C}^2,\Delta\)) \(H(X):=\bigoplus_{n\geq 0}H^*(X^{[n]})\) as an irreducible highest weight representation of the (standard) Heisenberg algebra \({\mathcal H}(X)\), computing the Betti numbers of \(X^{[n]}\) and giving an explicit basis for the rational cohomology \(H^*(X^{[n]})=H^*(X^{[n]},{\mathbb Q})\). By the way they give a new proof of the irreducibility (proved first by Briancon) of the punctual Hilbert scheme \(\Delta_0^{[n]}\). The main result of the present paper is a Decomposition Theorem (DT) for the Douady-Barlet morphism \(\pi : X^{[n]} \rightarrow X^{(n)}\), which states (in its weak version) that the complex \(R\pi_*{\mathbb C}_{X^{[n]}}\) is quasi-isomorphic (in the corresponding derived category) to a complex with trivial differentials. This DT deeply uses the described natural stratification of \(\pi\) and can be viewed as a particular case of the Decomposition Theorems by Beilinson-Bernstein-Deligne-Gabber and by M. Saito, but here there is no use either of perverse sheaves nor Mixed Hodge Structures for \(D\)-modules. The authors also give important consequences of the DT: (i) The Leray spectral sequence for \((\pi,{\mathbb Q}_{X^{[n]}})\) is \(E_2\)-degenerate. (ii) The Göttsche formula for any complex surface \(X\). (iii) The mixed Hodge structure of \(X^{[n]}\) in the Kähler case. (iv) A connection with the Equivariant \(K\)-theory. The paper is very well written and structured. In reference 4 `J. N. Deligne' should be J.N. Bernstein.