Automorphisms and the canonical ideal (Q2243973): Difference between revisions

Let \({\mathcal X}\) be a genus \(g\geq 2\), non-singular, complete algebraic curve defined over an algebraically closed field of characteristic \(p \geq 0\). The automorphism group \(G = \mbox{Aut}({\mathcal X})\) of \({\mathcal X}\) is a finite group which acts naturally on the space \({\Omega_{\ell}}={\Omega_{\ell}}({\mathcal X})\) of holomorphic \(\ell\)-differentials on \({\mathcal X}\) for any \(\ell \geq 1\). Recall that the dimension of \({\Omega_1}\) is \(g\) and the dimension of \({\Omega_2}({\mathcal X})\) is \(3g-3\), for \(g\geq 2\). A theorem of Max Noether says that: if \({\mathcal X}\) is non-hyperelliptic of genus \(g \geq 3\), then the product map \(\mbox{Sym}^{\ell} ({\Omega_1}) \, \, \to \, \, {\Omega_{\ell}}\) is surjective for \(\ell \geq 2\). Thus for \(\ell=2\) the kernel \(\Lambda_2=\Lambda_2(X)\) of this map has dimension \(\frac {(g-2)(g-3)} 2\). Here \(\mbox{Sym}^{\ell}\) denotes the \(\ell\)-th symmetric power of a vector space (i.e., the space of homogeneous polynomials of degree \(\ell\) in the coordinates of a basis of the dual of the vector space). {Petri's Theorem:} Assume that \(g\geq 4\) and \({\mathcal X}\) is neither hyperelliptic, trigonal or isomorphic to a smooth quintic plane curve. Then the ideal \(I\) of the algebra \(\mbox{Sym}({\Omega_1} ) \) generated by \(\Lambda_2({\mathcal X})\) is the ideal corresponding to a \(G\)-invariant curve \({\mathcal X}'\) in \({\mathbb P} ({{\Omega_1}}^{\ast})\) that is isomorphic to \({\mathcal X}\) as a \(G\)-curve. Further, \(I\cap \mbox{Sym}^2({\Omega_1})=\Lambda_2({\mathcal X})\). For curves satisfying Petri's theorem, the authors give a criterion for identifying the automorphism group as an algebraic subgroup the general linear group. Furthermore, the action of the automorphism group is extended to a linear action on the generators of the minimal free resolution of the canonical ring of the curve \({\mathcal X}\). They prove that the automorphism group of a curve \({\mathcal X}\), as a finite set, can be seen as a subset of the \(g^2 (g + 1)^2-1\)-dimensional projective space and can be described by explicit quadratic equations. At the end of the paper they illustrate some of their results with Fermat curves.