The Gieseker-Petri divisor in \({\mathcal{M}_g}\) for \(g \leq 13\) (Q431062): Difference between revisions

Let \(C\) be a smooth irreducible projective curve of genus \(g\) and \([C] \in {\mathcal M}_g\) its isomorphism class in the coarse moduli space. For any line bundle \(L\) on \(C\), the Petri map \[ \mu_{0, L}: \; H^0(C,L) \otimes H^0(C,K_C \otimes L^{-1}) \longrightarrow H^0(C,K_C) \] has been widely studied in the recent literature. In particular, this map plays a crucial role in the description of the varieties \(G_d^r(C):= \{(L,V): \; L \in \mathrm{Pic}^d(C), V \in G(r+1, H^0(L))\}\) consisting of the linear series \(g_d^r\) on \(C\) of degree \(d\) and dimension \(r\). An important result in this subject is the Gieseker-Petri Theorem [\textit{D. Gieseker}, ``Stable curves and special divisors: Petri's conjecture'', Invent. Math. 66, 251--275 (1982; Zbl 0522.14015)]: if \([C] \in {\mathcal M}_g\) is general, then for any line bundle \(L\) the map \(\mu_{0, L}\) is injective. As a consequence, if moreover the Brill-Noether number \(\rho(g,r,d)\) is non negative then \(G_d^r(C)\) is smooth of the expected dimension \(\rho\). A well-known conjecture concernes the locus of \({\mathcal M}_g\) where the Petri map fails to be injective: is the so-called Gieseker-Petri locus \[ GP_g := \{[C] \in {\mathcal M}_g \; | \; \text{ker} (\mu_{0, L}) \neq 0\} \] of pure codimension \(1\) in \({\mathcal M}_g\)? An affirmative answer has been given if \(g \leq 8\) in [\textit{A. Castorena}, ``Curves of genus seven that do not satisfy the Gieseker-Petri theorem'', Boll. Unione Mat. Ital., Sez. B, Artic. Ric. Mat. (8) 8, No. 3, 697--706 (2005; Zbl 1178.14024)] and [\textit{A. Castorena}, ``Remarks on the Gieseker-Petri divisor in genus eight'', Rend. Circ. Mat. Palermo (2) 59, No. 1, 143--150 (2010; Zbl 1195.14043)]. In the present paper the author proves the Conjecture when \(9 \leq g \leq 13\). The strategy consists in examining the different components of \(GP_g\), a sort of ``stratification'' given by the degree and the dimension of the linear series involved. In order to do this the \textit{GP locus of type \((r,d)\)} is defined as \(GP_{g,d}^r : = \{[C] \in {\mathcal M}_g \; | \; \text{there exists a base-point-free} \; (L,V) \in G_d^r(C) \; \text{such that} \; \text{ker} (\mu_{0, V}) \neq 0\}\) where \(\mu_{0, V}\) is the obvious restriction of \(\mu_{0, L}\) to \(V \otimes H^0(C,K_C \otimes L^{-1})\). The proof of the main theorem is divided in two parts: in the first one the cases \(g=9, 10, 11\) are examined, with the aid of some inclusions between the involved components \(GP_{g,d}^r \) and comparing them with the Brill-Noether varieties \({\mathcal M}_{g,d}^r \). The second part considers the cases \(g=12,13\) with some ``ad hoc'' arguments, quite tricky and carefully explained. This last part suggests that the study of the Conjecture for higher values of \(g\) could be rather intricate.