Moduli spaces of semistable sheaves of dimension \(1\) on \(\mathbb{P}^2\) (Q2344036)

Let \(M(d,\chi)\) be the moduli space of semistable sheaves on \({\mathbb P}^2\) of rank \(0\), first Chern class \(dH\) (\(d>0\)) and Euler characteristic \(\chi\). These spaces were first studied by \textit{J. Le Potier} [Rev. Roum. Math. Pures Appl. 38, No. 7-8, 635--678 (1993; Zbl 0815.14029)] and more recently by \textit{J.-M. Drézet} and \textit{M. Maican} [Geom. Dedicata 152, 17--49 (2011; Zbl 1236.14012)]. The object of the present paper is to describe a big open subset \(W^d\) of \(M(d,1)\) and to use this to determine \(M(d,1)\) for \(d\leq3\) and to calculate the classes in the Grothendieck ring of varieties of \(M(4,1)\), \(M(5,1)\) and \(M(5,2)\). (Note here that \(M(d,\chi)=M(d,\chi')\) if \(\chi\equiv\pm\chi'\;(\text{mod }d)\).) The paper is based on building a 1-1 coorespondence between pure sheaves of dimension \(1\) on \({\mathbb P}^2\) and pairs \((E,f)\) with \(E\) a direct sum of line bundles on \({\mathbb P}^2\) and \(f:E\otimes {\mathcal O}_{{\mathbb P}^2}(-1)\to E\) an injective map. One can then view \(M(d,\chi)\) as a moduli space for semistable pairs (with a suitable definition of stability). This allows the author to generalise a result of Drézet and Maican [loc. cit.] and describe a large open subset \(W^d\) of \(M(d,\chi)\) (Proposition 1.1/4.5). More precisely, \(W^d\simeq{\mathbb P}({\mathcal V}^d)\), where \({\mathcal V}^d\) is a vector bundle of rank \(3d\) over \(\displaystyle N_0^d:=\mathrm{Hilb}^{[\frac{(d-1)(d-2)}2]}({\mathbb P}^2)-\Omega^{[\frac{(d-1)(d-2)}2]}_{d-3}\), where \(\Omega^{[n]}_k\) is the closed subscheme of \( \mathrm{Hilb}^{[n]}({\mathbb P}^2)\) parametrizing \(n\) points lying on a curve of class \(kH\). Using this, it follows that \(M(d,1)=W^d\) for \(d\leq3\) (Theorem 1.2/5.1). Finally the class \([M(d,\chi)]\) of \(M(d,\chi)\) in the Grothendieck ring is obtained for \((d,\chi)=(4,1), (5,1), (5,2)\). In all cases, \([M(d,\chi)]\) is polynomial-count, that is it is of the form \([M(d,\chi)]=\sum b_{2i}{\mathbb L}^i\) for some integers \(b_{2i}\), where \({\mathbb L}:=[{\mathbb A}^1]\). The values of \(b_{2i}\) are calculated for \(M(4,1)\) (Theorem 1.3/5.2) and for \(M(5,1)\) and \(M(5,2)\) (Theorem 1.4/6.1). In fact, it turns out that \([M(5,1)]=[M(5,2)]\). As a consequence of these results, the Euler number \(e(M(4,1))=192\), while \(e(M(5,1))=e(M(5,2))=1695\). The Euler numbers of all \(M(d,\chi)\) with \(\gcd(d,\chi)=1\) have been calculated by physical arguments (BPS states) and depend only on \(d\). The cases calculated here agree with these predictions.