Localization and ideal theory in iterated differential operator rings (Q1086331)

This paper contains a wealth of ideas about the ideal theory of a skew enveloping algebra R{\#}(U(L) and related rings. Here R is usually an algebra over a field k and L is a finite-dimensional Lie k-algebra acting on R via k-linear derivations. The main results concern the second layer condition and its consequences for clique localization in R{\#}U(L). For instance, if L is solvable and \(char(k)=0\), then R{\#}U(L) is proved to satisfy the second layer condition in case R is a noetherian P.I. ring, or a simple noetherian ring, or an artinian ring. (An analogous result is proved when L is nilpotent and \(char(k)>0.)\) Consequences include the result that if L is solvable, k is uncountable with \(char(k)=0\), and R is a noetherian P.I. algebra, then all cliques of prime ideals in R{\#}U(L) are classically localizable. (Other results provide cases in which single prime ideals must be classically localizable.) Combining the previous result with work of Brown and Warfield, the author concludes that the classical Krull dimension of R{\#}U(L) must be bounded by its global dimension, if finite. In preparation for the localization work, primeness, semiprimeness, Jacobsonianism, hypercentrality, and simplicity of R{\#}U(L) are investigated. To set up inductive arguments, many results are worked out in the more general context of iterated differential operator rings. Much of the research in this paper is analogous to the author's work on noetherian strongly group-graded rings [J. Algebra 105, 76-115 (1987; Zbl 0607.16013)], although different techniques are required.