The global dimension of a \(q\)-skew polynomial ring (Q1569819): Difference between revisions

Let \(S=R[x;\sigma,\delta]\) be a skew polynomial ring (Ore extension), with \(\sigma\) an automorphism of \(R\). When \(\sigma\) and \(\delta\) are both trivial, the Hilbert Syzygy Theorem says that the global homological dimension of \(S\) is given by \(\text{gl.dim}(S)=\text{gl.dim}(R)+1\), and this conclusion also holds if only \(\delta\) is trivial [\textit{K. L. Fields}, J. Algebra 13, 1-4 (1969; Zbl 0181.04803)]. In general, however, \(\text{gl.dim}(S)\) is not determined by \(\text{gl.dim}(R)\). In particular, when \(\text{gl.dim}(R)=n<\infty\) it is known that \(\text{gl.dim}(S)\) is either \(n\) or \(n+1\) (see [\textit{J. C. McConnell} and \textit{J. C. Robson}, Noncommutative Noetherian rings, Wiley-Interscience, New York (1987; Zbl 0644.16008)]), but effective criteria to distinguish between these two possibilities have not been determined except in special cases. For instance, if \(R\) is semiprime Noetherian, then \(\text{gl.dim}(S)=n+1\) if and only if there exists a simple right \(S\)-module \(M\) such that \(M_R\) is finitely generated with projective dimension \(n\) [ibid.]. In case \(R\) is commutative Noetherian and \(\sigma\) is trivial, \(\text{gl.dim}(S)=n+1\) if and only if \(R\) contains a maximal ideal \(M\) of height \(n\) such that either \(\text{char}(R/M)>0\) or \(\delta(M)\subseteq M\) [the reviewer, Trans. Am. Math. Soc. 209, 65-85 (1975; Zbl 0306.16018)]. Here the author analyzes the \(q\)-skew case, where \(\delta\sigma=q\sigma\delta\) for some central, invertible \((\sigma,\delta)\)-constant \(q\in R\), again assuming that \(\text{gl.dim}(R)=n<\infty\). The majority of his work allows \(R\) to be noncommutative Noetherian, but the main theorem requires commutativity. For \(R\) commutative Noetherian, it is proved that \(\text{gl.dim}(S)=n+1\) if and only if \(R\) contains a maximal ideal \(M\) of height \(n\) which satisfies a suitable stability condition relative to \(\sigma\) and \(\delta\). At one extreme, this stability is automatically satisfied when \(\text{char}(R/M)>0\) and \(q\) is a root of unity modulo \(M\). On the other hand, if \(q\) is not a root of unity modulo \(M\), the required condition is that \(M\) be stable under \((a\delta+b)\sigma^{-1}\) for some \(a\in R\setminus P\) and \(b\in R\). Finally, when \(\text{char}(R/M)=0\) and \(q\) is a primitive \(t\)-th root of unity modulo \(M\), the condition above must be modified to include a \(\delta^t\) term. Examples of the various possible outcomes are given.