The first two Betti numbers of the moduli spaces of vector bundles on surfaces (Q1389073): Difference between revisions

Let \(M(I,d)^0\) be the moduli space of rank two Mumford-stable vector bundles \(E\) on a polarized surface \((X,H)\) where \(c_1(E)=I\in \text{Pic}(X)\) and \(c_2(E)=d\). Let \(b_i=\dim H_i(X)\). The author proves that there is an integer \(N\) depending on \((X,I,H)\) such that for \(d\geq N\) the following holds: \(\dim H_1(M(I,d)^0)=b_1\), \(\dim H_2(M(I,d)^0)=b_2+{1\over 2}b_1(b_1-1)\) . If \(M(I,d)\) is the quasi-projective moduli space of stable sheaves, containing \(M(I,d)^0\) as Zariski open subset, the author proves that there is an integer \(N\) depending on \((X,I,H)\) such that for \(d\geq N\) the following holds: \(\dim H_1(M(I,d))=b_1\), \(\dim H_2(M(I,d))=b_2+{1\over 2}b_1(b_1-1)+1\) . The author announces a paper [cf. \textit{J. Li}, in: Moduli of vector bundles, Pap. 35th Taniguchi sympos. 1994, Lect. Notes Pure Appl. Math. 179, 129-146 (1996; Zbl 0930.14005)] where he computes the Picard group of the above moduli spaces from the results gained in this paper. Many results about \(M(I,d)^0\) are known for \(d\gg 0\) [see for example \textit{J. Li}, Invent. Math. 115, No.~1, 1-40 (1994; Zbl 0799.14015) and \textit{D. Gieseker} and \textit{J. Li}, J. Differ. Geom. 40, No.~1, 23-104 (1994; Zbl 0827.14008)]. The author compares (according to Taubes) the cohomology groups of the moduli spaces under study with those of the space \({\mathcal B}(P_d)^*\) of gauge equivalent classes of irreducible connections on the corresponding principal bundle \(P_d\) over \(X\). It is constructed a homomorphism \(\tau(d)_i\colon H_i(M(I,d)^0)\to H_i(M(I,d+1)^0)\) and it is shown that it is surjective for \(i\leq 2\). A main technical point in order to show that \(\tau(d)_i\) is surjective is to apply the Lefschetz hyperplane theorem to the pair \((M(I,d), M(I,d)\setminus M(I,d)^0)\). This cannot be done directly because \(M(I,d)\setminus M(I,d)^0\) is not ample. So the author finds an ample subvariety \({\mathcal Y}\) of \(M(I,d)\), defined by those \(E\in M(I,d)\) such that \(E_{|C}\) is not semistable where \(C\in |nH|\) is a fixed smooth divisor for some \(n>0\). The author succeeds in applying the Lefschetz hyperplane theorem to the pair \((M(I,d), {\mathcal Y})\) by using the stratified Morse theory of Goresky and MacPherson and deformation techniques. This paper is well written, contains many ideas and also some expository part. It is recommended to understand several important results and techniques about moduli spaces.