Divisorial components of the Petri locus for pencils (Q2477650): Difference between revisions

Let \(C\) be a non-singular projective curve of genus \(g\) and \(L\) a line bundle on \(C\). The Petri map is the map \(\mu_0(L): H^0(C,L)\otimes H^0(C,K\otimes L^{-1})\rightarrow H^0(C,K)\) defined by the cup product. According to the Gieseker--Petri theorem [see e.g. \textit{D. Eisenbud} and \textit{J. Harris}, Invent. Math. 74, 269--280 (1983; Zbl 0533.14012)], the Petri map is injective if the curve \(C\) is general. This gives rise to the problem of describing the Petri locus, i.e. the locus of curves \(C\) for which the Gieseker--Petri theorem fails, as a subvariety of the moduli space \(\mathcal{M}_g\) of non-singular genus \(g\) curves. Let \(d\geq1\) be an integer such that Brill--Noether number \(\rho(g,1,d)=2d-g-2\) is non-negative. \textit{G. Farkas} [J. Reine Angew. Math. 581, 151--173 (2005; Zbl 1076.14035)] proved that the Petri locus contains a component of codimension \(1\) in \(\mathcal{M}_g\), corresponding to curves \(C\) such that \(\mu_0(L)\) is not injective for a line bundle \(L\) of degree \(d\) with \(\dim H^0(C,L)\geq 2\). In this paper, the authors give a different proof of this result, using degeneration to a certain class of singular curves with components of genus \(0\) and \(1\). Note that proving the existence of a divisorial component of the Petri locus is equivalent to proving that the Petri locus contains a curve with only a finite number of degree \(d\) line bundles \(L\) with \(\mu_0(L)\) not injective and \(H^0(C,L)\) of the prescribed dimension.