F-thresholds, tight closure, integral closure, and multiplicity bounds (Q841547): Difference between revisions

Suppose that \(\mathfrak{a}, J \subseteq R\) are ideals in a \(d\)-dimensional Noetherian local domain \((R, \mathfrak{m})\) of positive characteristic where we further assume that \(\mathfrak{a} \subseteq \sqrt{J}\). \textit{M. Mustata; S. Takagi; K.-i. Watanabe} [in: Proceedings of the 4th European congress of mathematics (ECM), Stockholm, Sweden, June 27--July 2, 2004. Zürich: European Mathematical Society (EMS). 341--364 (2005; Zbl 1092.32014)] introduced an invariant of these ideals. It is denoted \(c^{J}(\mathfrak{a})\) and called the \textit{\(F\)-threshold}. Also compare with [\textit{S. Takagi; K.-i. Watanabe}, J. Algebra 282, No. 1, 278--297 (2004; Zbl 1082.13004)]. This invariant (under certain hypotheses) is closely related to the log canonical threshold in characteristic zero. Explicitly, first define \(\nu_{\mathfrak{a}}^J(p^e) = \text{max}\{r | \mathfrak{a}^r \not\subseteq J^{[p^e]} \}\). Then \[ c_+^J(\mathfrak{a}) := \limsup_{e \to \infty} {\nu_{\mathfrak{a}}^J(p^e) \over p^e} \text{ and } c_{-}^J(\mathfrak{a}) := \liminf_{e \to \infty} {\nu_{\mathfrak{a}}^J(p^e) \over p^e}. \] If \(c_+^J(\mathfrak{a}) = c_-^J(\mathfrak{a})\) then we call the common value \(c^J(\mathfrak{a})\). While \(c^{J}(\mathfrak{a})\) was previously explored in the case that \(R\) is regular, the more general theory had not been worked out. In this paper, these invariants are studied in a wider setting -- and in particular, it is shown that the invariant exists in a number of cases (see Lemma 2.3 and the surrounding discussion). In this paper, the authors explore connections between this invariant and the tight closure of ideals. Explicitly, if \(J\) is an ideal generated by a full system of parameters and \(J \subseteq I\), then \(I \subseteq J^*\) if and only if \(c^I_+(J) = d =\dim R\) (see Corollary 3.2). Other relations with integral closure of ideals are also explored in section 3. In section 4, this theory is generalized even further to the context of modules. In section 5, inspired by relations between the log canonical threshold and the multiplicity, see [\textit{T. de Fernex; L. Ein; M. Mustata}, J. Algebr. Geom. 13, No. 3, 603--615 (2004; Zbl 1068.14006)] the authors also make the following conjecture. If \(J\) is assumed to be generated by a full system of parameters then \[ e(\mathfrak{a}) \geq \left({d \over c^J_{-}(\mathfrak{a})}\right) e(J). \] This was previously shown by Takagi and Watanabe in the case that \(R\) is regular and \(J = \mathfrak{m}\). The authors are able to prove this conjecture in several special cases, see Proposition 5.5, Theorem 5.6, and Corollary 5.9.