Constructing chains of primes in power series rings (Q652176)

For an integral domain \(D\) of dimension \(n\), the dimension of the polynomial ring \(D[x]\) is known to be bounded by \(n+1\) and \(2n+1\). While \(n+1\) is a lower bound for the dimension of the power series ring \(D[[x]]\), it often happens that \(D[[x]]\) has infinite chains of primes. For example, such chains exist if \(D\) is either an almost Dedekind domain that is not Dedekind or a rank-one nondiscrete valuation domain. \textit{J. T. Arnold} proved [Trans. Am. Math. Soc. 177, 299--304 (1973; Zbl 0262.13007)] that for a particular large class of non-Noetherian domains the ring of formal power series is always infinite-dimensional. This holds true even when the domain \(D\) has dimension \(1\). The property in question in Arnold's paper is the SFT (strong finite type) property. An ideal \(I\) of a domain \(D\) is said to have the SFT-property provided there exists a finitely generated ideal \(J\) with \(J\subseteq I\) and a positive integer \(n\) such that \(d^n\in J\) for every \(d\in I\). A domain \(D\) is said to have the SFT property provided every ideal of \(D\) has the SFT property. Arnold's main result is that if \(D\) is an \(n\)-dimensional domain which does not have the SFT property, then \(D[[x]]\) is infinite dimensional. The key to Arnold's result is the existence of a prime ideal \(P\) in a non-SFT-domain \(D\) such that \(P[[x]]\), the ideal of power series with coefficients in \(P\), is properly larger than the ideal \(PD[[x]]\), the ideal in \(D[[x]]\) generated by elements of \(P\). The infinite chains of prime ideals lie between \(PD[[x]]\) and \(P[[x]]\). The conditions that Arnold places on power series in order to prove his results are quite technical and nonintuitive, and the proof is an existence proof which does not indicate the structure of the prime ideals which are proven to exist. The primary goal of the paper under review is to construct infinite chains of prime ideals in \(D[[x]]\) for a non-Noetherian almost Dedekind domain \(D\). The principal tool is to construct an appropriate function which provides a measure of how ``large'' a given power series is. The key is that once such a function has been defined, the authors show that the collection of power series which are ``larger'' than a given power series constitutes a prime ideal. They then show that one can construct power series which are arbitrarily large or arbitrarily small relative to a given power series (of a certain type). A consequence is that the chains of prime ideals are actually doubly infinite (infinite ascending and infinite descending). In fact, the authors prove some very precise results concerning the cardinality of chains of primes (of a certain type) and the existence of large chains of primes in between two given (nonadjacent) primes. This paper consists of three sections in addition to the introduction. The results in Section \(1\) are quite general and not restricted to almost Dedekind domains. In Section \(2\), the authors use the general results from Section \(1\) and some fairly specific algorithms to construct chains of primes in \(D[[x]]\) for a very restricted type of non-Noetherian almost Dedekind domain. In Section \(3\), they extend these methods to arbitrary non-Noetherian almost Dedekind domains and include examples to illustrate some differences.