Fiber cones and the integral closure of ideals (Q5935767)

In general the product of two integrally closed ideals \(I\) and \(J\) in a regular local ring \((R,{\mathfrak m})\) is not integrally closed again, and it is an interesting problem to find conditions when it is. In this note we look at this question in the special situation that \(J={\mathfrak m}\). Our motivation came from the fact, that if \((R,{\mathfrak m})\) is a regular local ring, essentially of finite type and residually finite over a field \(k\) of characteristic 0, and if \(I\subseteq R\) is a prime ideal such that \({\mathfrak m} \cdot I\) is integrally closed, then \(R/I\) is evolutionarily stable as a \(k\)-algebra, thus answering a question of Mazur in this situation. -- In case \(I \subseteq R\) is a normal unmixed ideal in a normal local Cohen-Macaulay domain \((R,{\mathfrak m})\), we find two quite general conditions for the completeness of \({\mathfrak m}\cdot I\): If the ideal defining the Rees algebra of \(I\) contains no monic cubics, and if \({\mathfrak m}\) is an embedded component of \(I^2\), then \({\mathfrak m}\cdot I\) is integrally closed. Similarly, if \(I\) is a syzygetic ideal of height \(\dim(R)-1\) and analytic spread \(\dim(R)\), if \(I\) is generically a complete intersection, and if it has reduction number 2 (i.e. if the ideal defining the Rees algebra of \(I\) contains as many monic cubics as possible), then \({\mathfrak m}\cdot i\) is integrally closed (and so is \({\mathfrak m} \cdot I^n\) for all integers \(n>0)\). The main technical tool used in this paper is the fibre cone \(F(I)=\bigoplus I^n/{\mathfrak m}\cdot I^n\) of \(I\) and its relations to the Rees-algebra \(R[It]\) of \(I\).