A note on the projective normality of special line bundles on abelian varieties (Q1823283)

This paper contains not only an amusing mixture of misprints and interesting grammatical constructions, but also a proof of the following theorem: Given an abelian variety A over an algebraically closed field k of characteristic \(\neq 2.\) Let L be an ample line bundle on A. Suppose that \([-1]^*L=L\) and also that the space of global sections of L has odd dimension. Then the obvious maps \(H^ 0(A,L^{\otimes 2})^{\otimes n}\quad \to \quad H^ 0(A,L^{\otimes 2n})\) are surjective for all n if and only if the base locus of L does not contain a point in the 2-torsion subgroup of A(K). Over the complex numbers this can be proven using theta functions. It boils down to proving that a theta function \(\theta\left[\begin{matrix} \xi \\ \sigma \end{matrix} \right](\tau,2z)\) can be expressed in terms of functions \(\theta\left[\begin{matrix} \eta \\ 0\end{matrix} \right](2\tau,2z)\theta \left[ \begin{matrix} \eta '\\ 0\end{matrix} \right](2\tau,2z)\). By multiplying our theta function by a suitable theta constant, the Riemann theta relations provide such a formula. In order that this theta constant can be chosen to be non-zero, the assumption `odd dimension' is used. This classical proof is then translated to the more general case by means of Mumford's theory of theta functions.