\(F\)-regular and \(F\)-pure normal graded rings (Q1175539)

Let \(R\) be a commutative Noetherian ring of characteristic \(p>0\), and \(F:R\to R\) the Frobenius map defined by \(F(a)=a^ p\). If we denote by \(^ 1R\) the ring \(R\) viewed as an \(R\)-module via \(F\), then \(R\) is said to be \(F\)-pure if for every \(R\)-module \(M\) the canonical morphism \(M\to M\otimes_ R{^ 1R}\) is injective. Using the notion of tight closure of an ideal introduced by \textit{M. Hochster} and \textit{C. Huneke} in J. Am. Math. Soc. 3, No. 1, 31-116 (1990; Zbl 0701.13002) it can be defined the notion of \(F\)-regular ring: \(R\) is said to be \(F\)-regular if every ideal of \(R\) is tightly closed (in the terminology of Hochster and Huneke this notion is called ``weakly \(F\)-regular''). The purpose of the paper under review is to give a criterion to find whether \(R\) is \(F\)-regular (respectively \(F\)-pure) or not via the action of \(F\) on the highest local cohomology group of the canonical module. This criterion is then applied to normal graded rings. A complete classification of normal \(F\)-regular (respectively \(F\)-pure) graded rings of dimension 2 is also given.