Normalization of modules (Q2509248): Difference between revisions

Let \(R\) be a noetherian ring, let \(E\) be a finitely generated torsion free \(R-\)module having a rank, and choose an embedding \(\varphi :E\hookrightarrow R^e\). The Rees algebra \(R(E)\) is the \(R\) subalgebra of the polynomial ring \(R[t_1,\dots,t_e]\), generated by all linear forms \(a_1t_1+\dots+a_et_e\), with \((a_1,\dots,a_e)\in \varphi (E)\). \(E\) is said to be integral over the submodule \(U\subset E\) if \(R(E)\) is integrally closed over \(R(U)\subset R(E)\), in this case \(U\) is said to be a reduction of \(E\). In an analogous way the integral closure \(\overline E\) of \(E\) in \(R^e\) is defined, and we speak about integrally closed modules. If \(R(E)\) is integrally closed in \(R[t_1,\dots,t_e]\) then \(E\) is said to be normal. The aim of this paper is to introduce techniques to derive estimates for the degrees of the generators of the integral closure of several classes of Rees algebras of modules, and to bound the length of the normalization processes. If \(R\) is regular the bounds are expressed in terms of Buchsbaum-Rim multiplicities and a module version of Briancon-Skoda numbers. The paper is clear and well written, but the subject needs a lot of definitions.