Integral closures, local cohomology and ideal topologies (Q2477999): Difference between revisions

Let \(R\) be a commutative ring, \(\mathfrak a\subseteq \mathfrak b\) ideals of \(R\) and \(A\) an Artinian \(R\)-module. The ideal \(\mathfrak a\) is said to be a reduction of the ideal \(\mathfrak b\) relative to \(A\) if there exists an integer \(s\in \mathbb{N}\) such that \((0:_A\mathfrak a \mathfrak b^s)=(0:_A\mathfrak b^{s+1})\). An element \(x\) of \(R\) is said to be integrally dependent on \(\mathfrak a\) relative to \(A\) if \(\mathfrak a\) is a reduction of \(\mathfrak a+(x)\) relative to \(A\). The set of all such elements of \(R\) is called the integral closure of \(\mathfrak a\) relative to \(A\) and is denoted by \(\mathfrak a^{*(A)}\). Theses notions were introduced by \textit{R. Y. Sharp} and \textit{A. J. Taherizadeh} [J. Lond. Math. Soc., II. Ser. 37, No. 2, 203--218 (1988; Zbl 0656.13001)]. Next, assume that \((R,\mathfrak m)\) is local and noetherian and let \(\Phi\) be a system of non-zero ideals of \(R\), i.e. for any two ideals \(\mathfrak a,\mathfrak b\in \Phi\), there exists an ideal \(\mathfrak c\in \Phi\) such that \(\mathfrak c\subseteq \mathfrak a \mathfrak b\). Recall that for each \(i\geq 0\), the \(i\)th local cohomology module of \(R\) with respect to \(\Phi\) is defined by \(H^i_{\Phi}(R):={\varinjlim_{\mathfrak a \in \Phi}}\text{Ext}^{i}_{R}(R/\mathfrak a,R)\). It is known that \(H^{\dim R}_{\Phi}(R)\) is an Artinian \(R\)-module and \(H^i_{\Phi}(R)\cong {\varinjlim_{\mathfrak a \in \Phi}}H^i_{\mathfrak a}(R)\) for all \(i\geq 0\). Assume that \(R\) is formally equidimensional (i.e. \(\dim \widehat{R}/\mathfrak p= \dim \widehat{R}\) for all minimal prime ideals \(\mathfrak p\) of \(\widehat{R}\)) and that \(\dim R/\mathfrak a+\mathfrak p=0\) for all \(\mathfrak a\in \Phi\) and all minimal prime ideal \(\mathfrak p\) of \(R\). The main result of the paper under review asserts that for any ideal \(\mathfrak b\) of \(R\), the ideal \(\mathfrak b^{*(H^{\dim R}_{\Phi}(R))}\) is equal to \(\overline{\mathfrak b}\), the classical Northcott-Rees integral closure of \(\mathfrak b\). It is worth to mention that the imposed assumption on \(\Phi\) clearly implies that each element \(\mathfrak a\in\Phi\) is \(\mathfrak m\)-primary and so \(H^{\dim R}_{\Phi}(R)\cong {\varinjlim_{\mathfrak a \in \Phi}}H^{\dim R}_{\mathfrak a}(R)\cong H^{\dim R}_{\mathfrak m}(R)\).