Nilpotence of Frobenius actions on local cohomology and Frobenius closure of ideals (Q2415310)

In this review all rings are commutative Noetherian local rings of prime characteristic and of Krull dimension \(d\). The main result (in the reviewer's point of view) of the present paper is to give a, charming nice, necessary and sufficient condition for an equidimensional excellent ring, to be \(F\)-nilpotent. This necessary and sufficient condition is in terms of the equality of the Frobenius closure and tight closure of ideals generated by an arbitrary part of a system of parameters (or equivalently, in terms of the equality of the Frobenius closure and tight closure of any ideal generated by a full system of parameters). This is also proved to be equivalent to say that there exists a filter regular sequence \(x_1,\ldots,x_d\) so that \((x_1^{p^e},\ldots,x_t^{p^e})^F=(x_1^{p^e},\ldots,x_t^{p^e})^*\) for all \(t\le d\) and for all \(e\ge 0\). One other interesting result of the paper, achieved by applying the previously mentioned result, is that the property of being \(F\)-nilpotent localizes for (equi-dimensional) excellent local rings, which this result relaxes the superfluous \(F\)-finite condition of Proposition 2.4 of [\textit{V. Srinivas} and \textit{S. Takagi}, Adv. Math. 305, 456--478 (2017; Zbl Zbl 1352.13004)]. We remind that a \(d\)-dimensional commutative local ring \((R,\mathfrak{m})\) is said to be an \(F\)-nilpotent ring provided the iteration of the Frobenius action on non-top-local cohomologies is eventually zero and if the \(F\)-nilpotent submodule of \(H^d_{\mathfrak{m}}(R)\) is the largest possible submodule, i.e.\(0^*_{H^d_{\mathfrak{m}}(R)}\). For related geometric concepts (to the \(F\)-nilpotent property), see, [\textit{V. Srinivas} and \textit{S. Takagi}, Adv. Math. 305, 456--478 (2017; Zbl 1352.13004)] and [\textit{M. Blickle} and \textit{R. Bondu}, Ann. Inst. Fourier 55, No. 7, 2239--2256 (2005; Zbl Zbl 1092.14005)]. In [\textit{K. Shimomoto}, J. Algebra 338, No. 1, 24--34 (2011; Zbl 1241.13006)] the author introduces and investigates \(F\)-coherent rings and by one the his results in conjunction with the main result of the paper under review one can see that reduced \(F\)-coherent rings (e.g. \(R=k[x^4,x^3y,xy^3,y^4]\)) are \(F\)-nilpotent. A related numerical invariant to the \(F\)-nilpotence is the \(F\)-depth introduced and studied in [\textit{G. Lyubeznik}, Compos. Math. 142, No. 1, 207--221, (2006; Zbl 1100.13022)]. To prove the aforementioned characterization of \(F\)-nilpotent rings, the authors of the paper introduce the concept of Frobenius action relative to \(R\), and Frobenius nilpotence with respect to \(R\). The Frobenius action with respect to \(R\) is the map \(H^i_{K}(R/I)\rightarrow H^i_{K}(R/I^{[p^e]}) \) induced by the Frobenius-like map \(R/I\rightarrow R/I^{[p^e]}\) (\(r+I \mapsto r^{p^e}+I^{[p^e]}\)) (here \(I\) is an ideal of \(R\) contained in \(K\)). Having this relative Frobenius action in hand, one can define the notion of, being \(F\)-nilpotent with respect to \(R\), for the quotient ring \(R/I\). Then to achieve the aforementioned characterization of \(F\)-nilpotent rings, the authors of the paper prove the following nice two deformation/specialization type theorems which are of intrinsic interest. Theorem 5.10. Let \((R,\mathfrak{m})\) be an excellent equidimensional local ring of dimension \(d\) and prime characteristic \(p>0\). The followings are equivalent: (1) The ring \(R\) is \(F\)-nilpotent. (2) For each \(\mathfrak{m}\)-filter regular element \(x\) of \(R\), the ring \(R/(x)\) is \(F\)-nilpotent with respect to \(R\). (3) There exists an \(\mathfrak{m}\)-filter regular element \(x\) such that for each \(e\in \mathbb{N}\) the ring \(R/(x^{p^e})\) is \(F\)-nilpotent with respect to \(R\). Theorem 5.9. Let \((R,\mathfrak{m})\) be an excellent equidimensional local ring of dimension \(d\) and prime characteristic \(p>0\). Let \(x_1,\ldots,x_i\), \(i\le d\), be an \(\mathfrak{m}\)-filter regular sequence of \(R\) and let \(I=(x_1,\ldots,x_{i-1})\). The followings are equivalent. (1) \(R/I^{[p^e]}\) is \(F\)-nilpotent with respect to \(R\) for all \(e\ge 0\). (2) \(R/(I^{[p^e]},x_i^{p^{e'}})\) is \(F\)-nilpotent with respect to \(R\) for all \(e,e'\ge 0\). In view of the above two Theorems the property of the (equidimensional) excellent ring \(R\) to be \(F\)-nilpotent reduces equivalently to the Artinian quotient ring \(R/\mathfrak{q}\), to be \(F\)-nilpotent with respect to \(R\), wherein \(\mathfrak{q}\) is an ideal generated by a system of parameters (which can be chosen to be an \(\mathfrak{m}\)-filter regular sequence as well), that is to say the tight closure of the zero submodule of \(H^0_\mathfrak{m}(R/\mathfrak{q})=R/\mathfrak{q}\) being \(F\)-nilpotent with respect to \(R\), in other words \(\mathfrak{q}^*/\mathfrak{q}\) being \(F\)-nilpotent with respect to \(R\), i.e. \(\mathfrak{q}^F=\mathfrak{q}^*\). One of the ingredients of the proof of the above two Theorems, 5.9 and 5.10, is the following Theorem, proved by the authors of the present paper, which is also of intrinsic interest. Theorem 3.4. For a ring \(R\) let \(x_1,\ldots,x_t\) be an \(\mathfrak{m}\)-filter regular sequence of \(R\) and \(I=(x_1,\ldots,x_d)\). Then there exists a positive integer \(C\) such that, \(\mathfrak{m}^{Cp^{e}} H^0_\mathfrak{m}(R/I^{[p^e]})=0\), for all \(e\ge 0\). Theorem 3.4., besides its application in proving Theorems 5.9 and 5.10, shows that the (LC) Conjecture holds true for ideals generated by \(\mathfrak{m}\)-filter regular sequences. The (LC)-Conjecture is a conjecture whose affirmative answer implies another favorite conjecture on the equivalence of weakly \(F\)-regularity and \(F\)-regularity. The authors also show that \(F\)-nilpotent rings are always equidimensional, \(R\) is \(F\)-nilpotent if and only if \(R/\sqrt{0}\) is \(F\)-nilpotent, and if \(R\) is excellent then \(R\) is \(F\)-nilpotent if and only if \(\widehat{R}\) is so. We end this review by pointing out that by exploiting the notion of Frobenius nilpotence with respect to \(R\) introduced in this paper, the second named author of this paper in [\textit{P. H. Quy}, J. Algebra 518, 119--128 (2019; Zbl 1427.13005)] proves that the (uniform bound for) Frobenius test exponent for parameter ideals exists for \(F\)-nilpotent rings, and this last-mentioned intriguing result itself, is generalized in [\textit{K. Maddox}, Proc. Am. Math. Soc. 147, No. 12, 5083--5092 (2019; Zbl 1423.13050)].