Essential dimension of moduli of curves and other algebraic stacks (Q550566)

The object of the paper under review is to compute the essential dimension of various stacks of curves and abelian varieties over fields. Let \(k\) be a field, and let \(\mathcal X\) be an algebraic stack over \(k\). Given a field extension \(L/k\) and an object \(a\) in \(\mathcal X(L)\), the \textit{essential dimension \(\text{ed}(a)\)} of \(a\) is the minimal transcendence degree \(\text{trdeg}_k(K)\) of an intermediate field extension \(k \to K \to L\) such that \(a\) is in the essential image of the functor \(F(K) \to F(L)\). The \textit{essential dimension \(\text{ed}(\mathcal X)\)} of \(\mathcal X\) is the supremum of the values \(\text{ed}(a)\) taken over all field extensions \(L/k\) and all objects \(a\) in \(F(\mathcal X)\). When \(\mathcal X\) is a scheme (or more generally an algebraic space) of finite type, the essential dimension of \(\mathcal X\) recovers the usual dimension of \(\mathcal X\). The main result of the paper is to compute the essential dimension of the stack \(\mathcal M_{g,n}\) of smooth \(n\)-pointed curves of genus \(g\) over a field \(k\) of characteristic \(0\). The authors show that \(\text{ed} (\mathcal M_{g,n})\) equals \(2\) if \((g,n) = (0,0)\) or \((1,1)\); \(0\) if \((g,n) = (0,1)\) or \((0,2)\); \(+\infty\) if \((g,n) = (1,0)\); \(5\) if \((g,n) = (2,0)\); and \(3g-3+n\) for all other pairs \((g,n)\). They furthermore show that the stack \(\overline{\mathcal M}_{g,n}\) of stable \(n\)-pointed curves of genus \(g\) has essential dimension \(\text{ed}(\overline{\mathcal M}_{g,n}) = \text{ed}(\mathcal M_{g,n})\) for \(2g-2+n > 0\). Finally, they show that the stack of hyperelliptic curves has essential dimension \(2g\) if \(g\) is odd, and \(2g+1\) if \(g\) is even. Stacks of abelian varieties are treated in the appendix. Let \(\mathcal A_g\) denote the stack of principally polarized \(g\)-dimensional abelian varieties over a field \(k\), and let \(\mathcal B_g\) denote the stack of all \(g\)-dimensional abelian varieties over \(k\). Let \(2^a\) be the largest power of \(2\) dividing \(g\). Then the main result is that \(\text{ed}(\mathcal A_g) = g(g+1)/2 + 2^a\) if \(\text{char}(k) = 0\), or if \(\text{char}(k) = p > 0\) and \(p\nmid \#Sp_{2g}(\mathbb Z/\ell\mathbb Z)\) for some prime \(\ell > 2\); and that \(\text{ed}(\mathcal B_g) = \text{ed}(\mathcal A_g)\) if \(\text{char}(k) = 0\). For \(g\) odd this result is attributed to Miles Reid. It is not known whether the restriction on \(\text{char}(k)\) is necessary. A key ingredient in the proofs of the main results in both the main body and the appendix is a genericity result asserting that the essential dimension of a smooth integral tame Deligne-Mumford stack is the sum of its dimension and the essential dimension of its generic gerbe. A consequence of this is that the essential dimension of any such stack is equal to the essential dimension of any nonempty open substack.