On two notions of semistability (Q953048)

The notion of semistability for sheaves is well-established in the literature and allows the construction of moduli spaces of sheaves with fixed Hilbert polynomial on a projective variety. The theory of monads provides a useful approach to study moduli spaces of sheaves. When the group acting on the space of monads is reductive, then geometric invariant theory provides a notion of semistability of a monad. \textit{W. Barth} [Invent. Math. 42, 63--91 (1977; Zbl 0386.14005)] and \textit{K. Hulek} [Math. Ann. 242, 241--266 (1979; Zbl 0407.32013)] described the moduli space of rank 2 semistable torsion-free sheaves on \(\mathbb{P}^2\) as a good quotient of the space of semistable monads modulo a reductive group. More recently, \textit{J.-M. Drézet} [Math. Ann. 290, No. 4, 727--770 (1991; Zbl 0755.14005)] and \textit{J.-M. Drézet} and \textit{G. Trautmann} [Ann. Inst. Fourier 53, No. 1, 107--192 (2003; Zbl 1034.14023)] proposed a notion of semistability for complexes of morphisms of sheaves modulo non-reductive groups. \textit{H. G. Freiermuth} and \textit{G. Trautmann} [Am. J. Math. 126, No. 2, 363--393 (2004; Zbl 1069.14012)] investigated this notion in the case of sheaves on \(\mathbb{P}^3\) with certain fixed invariants. In the paper under review, the author investigates the moduli spaces \(M(r,\chi)\) of semistable sheaves on \(\mathbb P^2\) with linear Hilbert polynomial \(P(m)=rm+\chi\). For certain fixed multiplicity \(r\) and Euler characteristic \(\chi\), and under suitable cohomological conditions, these sheaves are proved to be cokernels of semistable morphisms \(\varphi: \mathcal{E}_1\to \mathcal{E}_2\), where \(\mathcal{E}_1, \mathcal{E}_2\) are direct sums of line bundles. Since the group \(G=\Aut(\mathcal{E}_1)\times \Aut(\mathcal{E}_2)\) acting on the space of morphisms is not reductive, the notion of semistability of Drézet-Trautmann is needed. The locally closed subvarieties (or open dense subsets) defined in the moduli space by the cohomological conditions are studied and in some cases they are proved to be good or geometric quotients of the set of morphisms modulo \(G\). Finally a duality result is proved, which implies that certain moduli spaces considered by the author are birational. This investigation leads to a good description of the cases of low multiplicity. On the other hand, for large multiplicity the picture is far from complete and it seems that the notion of semistability of Drézet-Trautmann is not enough for a complete classification.