When is a power series ring \(n\)-root closed? (Q674460)

Let \(A\subseteq B\) be two commutative rings with the same identity. For any positive integer \(n\), \(A\) is said to be \(n\)-root closed in \(B\) if \(b^n\) belongs to \(A\) for any \(b\) in \(B\) implies that \(b\) is already in \(A\). If \(A\) is \(n\)-root closed in \(B\) for every \(n\), then \(A\) is said to be root closed. In theorem 1.18 the authors characterize the \(n\)-root closure of \(A[[X]]\) in the ring \(B[[X]]\) in terms of the pair of rings \(A\subseteq B\) and a property \({\mathcal P}_p(A,B)\) for each prime factor \(p\) of \(n\). Here \({\mathcal P}_n (A,B)\) says that if \(b\) in \(B\), \(a\) in \(A\) and \(n.a.b\) is in \(A\) then \(n.a.b^2\) is also in \(A\). In particular \(A[[X]]\) is root closed in \(B[[X]]\) if and only if \(A\) is root closed in \(B\) and \(a.b\) in \(A\) implies \(a.b^2\) is also in \(A\) where \(a\) is in \(A\) and \(b\) in \(B\). Denoting by \({\mathcal C}(A,B)\) the set of all integers \(n>1\) for which \(A\) is \(n\)-root closed in \(B\), the authors show that for any projective ring \(A\) and any extension \(B\) of \(A\), \({\mathcal C}(A,B)= {\mathcal C}(A[[X]], B[[X]])\). For a domain \(A\), let \(A^*\) denote the complete integral closure of \(A\) in its field of fractions. Set \({\mathcal P}_n(A)= {\mathcal P}_n(A,A^*)\). In theorem 2.4, a characterization of the \(n\)-root closure of \(A\) is given in terms of \({\mathcal P}_p(A)\) for each prime factor of \(n\) (where \(n\geq 2)\). When \(A\) is a Mori domain, then \(A[[X]]\) is root closed if and only if \(A\) is root closed. If \(A\) is a Prüfer domain then \(A[[X]]\) is root closed if and only if \(A\) is completely integrally closed. Theorem 3.3 characterizes the complete integral closure of \(A[[X]]\) when \(A\) is a seminormal domain.