On quadratic generation of ideals defining projective toric varieties (Q1433494)

Let \(X\) be an \(n\)-dimensional, projective toric variety with an ample line bundle \(\mathcal L\). It is well-known that, for \(k\geq n-1\), the sheaf \(\mathcal L^{\otimes k}\) is very ample and defines a projectively normal embedding. Moreover, \textit{K.~Nakagawa} and \textit{S. Ogata} [Manuscr. Math. 108, 33--42 (2002; Zbl 0997.14014)] showed that the defining ideal is generated by quadrics if \(k\geq n\). In the present paper, the author lowers the latter bound to \(k\geq n-1\), provided that the toric variety \(X\) is at least 3-dimensional. This assumption is indeed necessary, since, in case of a surface, \textit{R.J. Koelman}'s papers [Beitr. Algebra Geom. 34, No. 1, 57--62 (1993; Zbl 0781.14025); Tohoku Math. J., II. Ser. 45, No. 3, 385--392 (1993; Zbl 0809.14042)] show that the case \(k=n-1\) \((=1)\) fails. The appearence of the assumption ``\(\dim X\geq 3\)'' becomes quite natural when looking at the second result of Ogata. It states that if the surjectivity of \(\Gamma(\mathcal L)\otimes\Gamma(\mathcal L^k)\to\Gamma(\mathcal L^{k+1})\) is granted from a certain degree, then this implies the quadric property of the corresponding tensor power of \(\mathcal L\) if this power exeeds \(n/2\). The main technical tool for studying these questions is to consider the kernel of a map of type \(\Gamma(\mathcal L_1)\otimes\Gamma(\mathcal L_2)\otimes\Gamma(\mathcal L_3)\to \Gamma(\mathcal L_1\otimes\mathcal L_2\otimes\mathcal L_3)\).