Galois extensions, plus closure, and maps on local cohomology (Q664760)

In what follows \((R,\mathfrak{m})\) will always be a commutative Noetherian local integral domain of prime characteristic \(p\). A Frobenius action on an \(R\)-module \(M\) is an additive map \(F:M\rightarrow M\) with \(F(rm)=r^pF(m)\) for every \(r\in R\) and \(m\in M\). The union of the ascending chain of submodules \(\ker F\subseteq\ker F^2\subseteq\ker F^3\subseteq\dots\) of \(M\) is the \(F\)-\textit{nilpotent} submodule of \(M\), denoted by \(M_{\text{nil}}\). The main result of the paper under review has two parts. In the first part the authors prove that if \(\mathfrak{a}\) is an ideal of \(R\) and \([\eta]\) is an element of \(H^i_{\mathfrak{a}}(R)_{\text{nil}}\), then there exists a module-finite extension domain \(S\) such that (a) the resulting extension of fraction fields is a Galois extension (such \(R\hookrightarrow S\) is called \textit{generically} Galois) with a solvable Galois group, and (b) the element \([\eta]\) maps to \(0\) under the induced map \(H^i_{\mathfrak{a}}(R)\rightarrow H^i_{\mathfrak{a}}(S)\). In the second part of their main result the authors prove that if in addition \(R\) is a homomorphic image of a Gorenstein ring, then there exists a module-finite generically Galois extension domain \(S\) such that the induced map on local cohomology \(H^i_{\mathfrak{m}}(R)\rightarrow H^i_{\mathfrak{m}}(S)\) is zero for each \(i<\dim R\). This last statement enhances a result of \textit{C. Huneke} and \textit{G. Lyubeznik} in [Adv. Math. 210, No. 2, 498--504 (2007; Zbl 1109.13003)] stating that if \(R\) is a homomorphic image of a Gorenstein ring, then there exists a module-finite extension domain \(S\) such that the induced map on local cohomology \(H^i_{\mathfrak{m}}(R)\rightarrow H^i_{\mathfrak{m}}(S)\) is zero for each \(i<\dim R\). The proof of the first part of the main result is achieved by reducing to the case where \(F([\eta])=0\), using induction, and then calculating with the \(\check{\text{C}}\)eck complexes \(\check{C}^{\bullet}(\mathbf{x};R)\) and \(\check{C}^{\bullet}(\mathbf{x};S)\) on a set of nonzero generators \(\mathbf{x}=x_0,\dots,x_n\) of \(\mathfrak{a}\). To prove the second part of their main result, the authors first note that it suffices to construct a module-finite generically \textit{separable} extension domain \(S\) such that \(H^i_{\mathfrak{m}}(R)\rightarrow H^i_{\mathfrak{m}}(S)\) is zero for each \(i<\dim R\). The proof is then achieved by induction on \(\dim R\). The above-mentioned result of Huneke and Lyubeznik, as well as the first part of the main result are used to conclude the proof. Let \(R^+\) denote the integral closure of \(R\) in an algebraic closure of its fraction field and let \(\mathfrak{a}\) be an ideal of \(R\). Define the \textit{plus closure} \(\mathfrak{a}^+\) of \(\mathfrak{a}\) as \(\mathfrak{a}R^+\cap R\) and its \textit{Frobenius closure} \(\mathfrak{a}^F\) as \(\{r\in R\mid r^{p^e}\in\mathfrak{a}^{[p^e]}\;\text{for some }e\in\mathbb{N}\}\). Another result proved in the paper under review has to do with these closure operations. It is proved: (1) given an element \(z\in\mathfrak{a}^F\), there exists a module-finite generically Galois extension domain \(S\) with a solvable Galois group, such that \(z\in\mathfrak{a}S\); (2) given an element \(z\in\mathfrak{a}^+\), there exists a module-finite generically Galois extension domain \(S\) such that \(z\in\mathfrak{a}S\). To complement statement (2) the authors provide an example, in which \(z\in\mathfrak{a}^+\), but conjecturally \(z\not\in\mathfrak{a}S\) for any module-finite generically Galois extension domain \(S\) with solvable Galois group.