Compactified Picard stacks over \({\overline{\mathcal M}_g}\) (Q1038700): Difference between revisions

Let \(X\) be a connected projective nodal curve of genus \(g \geq 3\) over an algebraically closed field \(k\). If \(Z\) is a proper subcurve of \(X\), one denotes by \(k_Z\) the number of intersection points of \(Z\) with its complement \(\overline{X\setminus Z}\). Recall that \(X\) is called \textit{stable} (resp., \textit{semistable}) if, for every smooth rational component \(E\) of \(X\) one has \(k_E \geq 3\) (resp., \(k_E \geq 2\)). If \(X\) is semistable and \(k_E = 2\) then \(E\) is called an \textit{exceptional component} of \(X\). A line bundle \(L\) of degree \(d\) on a semistable curve \(X\) is called \textit{balanced} if it satisfies the following inequalities [considered by \textit{D. Gieseker}, Lectures on moduli of curves (Lectures on Mathematics and Physics. Mathematics, 69. Tata Institute of Fundamental Research, Bombay. Springer Verlag, Berlin-Heidelberg-New York) (1982; Zbl 0534.14012)]: \[ \frac{\text{deg}({\omega}_X\, | \, Z)}{2g-2}d - \frac{k_Z}{2}\;\leq \;\text{deg}(L\, | \, Z)\;\leq \;\frac{\text{deg}({\omega}_X\, | \, Z)}{2g-2}d + \frac{k_Z}{2} \] for every connected proper subcurve \(Z\) of \(X\) and if \(\text{deg}(L\, | \, E) = 1\) for every exceptional component \(E\) of \(X\). If \(X\) admits a balanced line bundle then it must be \textit{quasistable} in the sense that any two exceptional components of it never meet. \textit{L. Caporaso} [J. Am. Math. Soc. 7, No. 3, 589--660 (1994; Zbl 0827.14014)] constructed a compactification \(\overline{P}_{d,g} \rightarrow \overline{M}_g\) of the universal Picard variety \(\text{Pic}^d_g \rightarrow M^0_g\), where \(\overline{M}_g\) is the moduli scheme of stable curves of genus \(g\) and \(M^0_g \subset \overline{M}_g\) is the locus of automorphism-free nonsingular curves. More precisely, for \(d \gg 0\), \(\overline{P}_{d,g}\) is the GIT-quotient \(H_d/\text{PGL}(r+1)\), where \(H_d\) is the locus of GIT-semistable points of \(\text{Hilb}^{dt-g+1}_{{\mathbb P}^r}\), \(r = d-g\). The points of \(H_d\) correspond to nondegenerate quasistable curves in \({\mathbb P}^r\) embedded by a balanced line bundle. Let \(\overline{\mathcal G}_{d,g}\) be the category whose objects are pairs \((f:{\mathcal C} \rightarrow Y,{\mathcal L})\), where \(f\) is a family of quasistable curves of genus \(g\) and \(\mathcal L\) a balanced line bundle of relative degree \(d\) over \(Y\). A morphism between two such pairs is a cartesian diagram: \[ \begin{tikzcd} \mathcal C \rar["h"]\dar & \mathcal C^\prime\dar \\Y \rar & Y^\prime \end{tikzcd} \] toghether with an isomorphism \({\mathcal L} \simeq h^{\ast}{\mathcal L}^{\prime}\). In the paper under review, the author shows that, for \(d >> 0\), \(\overline{\mathcal G}_{d,g}\) is isomorphic to the quotient stack \([H_d/\text{GL}(r+1)]\), hence it is an algebraic (Artin) stack. She also shows that the quotient stack \(\overline{\mathcal P}_{d,g} := [H_d/\text{PGL}(r+1)]\) is a stackification of the prestack whose sections over a scheme \(Y\) are given by pairs \((f:{\mathcal C} \rightarrow Y, {\mathcal L})\) as above and whose arrows between two such pairs are cartesian diagrams as above together with an isomorphism \({\mathcal L} \simeq h^{\ast}{\mathcal L}^{\prime} \otimes f^{\ast}M\), for some \(M \in \text{Pic}\, Y\). Moreover, let \(U_d\) denote the open subset of \(H_d\) corresponding to stable curves \(X\) which are \(d\)-\textit{general} in the sense that for any balanced line bundle of degree \(d\) on \(X\) the first inequality of Gieseker is strict for all \(Z \subset X\). The author shows that the quotient stack \(\overline{\mathcal P}^{\text{Nér}}_{d,g} := [U_d/\text{PGL}(r+1)]\) is a Deligne-Mumford stack and it is strongly representable over the open set \(\overline{\mathcal M}^d_g\) of the moduli stack \(\overline{\mathcal M}_g\) corresponding to \(d\)-general stable curves. She also gives a geometric description of the set \({\Sigma}^d_g \subset \overline{M}_g\) of \(d\)-special stable curves. These last results generalize some results of \textit{L. Caporaso} [Am. J. Math. 130, No. 1, 1--47 (2008; Zbl 1155.14023)] who considered the case \(\text{gcd}(d-g+1,2g-2) = 1\) which implies \(U_d = H_d\).