Many associated primes of powers of primes (Q2001430)

It has been proven that \textit{computationally hard} ideals provide an ample source of examples and counterexamples in commutative algebra. A class of these ideals are those that have large Castelnuovo-Mumford regularity. Recently, McCullough and Peeva gave a counterexample to the Eisenbud-Goto conjecture. In particular, they constructed a family of counterexamples via new constructions of step-by-step homogenizations and Rees-like algebras that have large Castelnuovo-Mumford regularity. Another class of computationally hard ideals is the class of ideals with large numbers of associated primes. Let \(K\) be an arbitrary field. The main goal of the paper under review is to construct a prime ideal \(\mathfrak{p}\) in a polynomial ring \(R\) over \(K\) such that the number of embedded associated primes of \(\mathfrak{p}^2\) is exponential in the number of variables of \(R\). More precisely, for every two positive integers \(m\geq 3\) and \(v\), the authors show that there exists a polynomial ring \(R\) in \(mv\) variables over \(K\) with an \(m\)-generated prime ideal \(\mathfrak{p}\) of height \(m-1\) such that \(\mathfrak{p}^e\) has \(v^m(=(\sqrt[v]{v})^{mv})\) embedded associated primes for all \(e\geq 2\). For proving their results, the authors introduce two new methods for generating ideals with large numbers of primary components. They call them \textit{splitting}, which is a generalization of the step-by-step homogenization, and \textit{spreading}.