Divisible modules and localization (Q2583032)

Given that \(R\) is an integral domain with quotient field \(Q,\) \(R\) is called a Matlis domain if \(\text{pd}(Q)=1.\) \textit{S. B. Lee} [Arch. Math. 53, No. 3, 259--262 (1989; Zbl 0652.13007)] proved that \(\text{pd}(Q)\leq 1\) if and only if \(Q/R\) decomposes into a direct sum of countably generated modules. This generalized the work of \textit{R. M. Hamsher} [J. Algebra 19, 416--425 (1971; Zbl 0225.13005)]. \textit{L. Fuchs} and \textit{L. Salce} [Forum Math. 4, No. 4, 383--394 (1992; Zbl 0767.13001)] generalized Lee's work by showing, among other things, that (if \(S\) is a multiplicative set of \(R\)) \(\text{pd}(R_{S})\leq 1\) if and only if \(R_{S}/R\) decomposes into a direct sum of countably generated modules. In the paper under review the authors extend the work of Fuchs and Salce mentioned above to rings of fractions of general commutative rings with 1. Matlis domains can also be characterized in terms of the properties of the category of \(R\) -modules generated by \(Q\) or, more generally, for an arbitrary commutative ring (with 1), the category of \(R\)-modules generated by the total quotient ring. This was \textit{E. Matlis}' approach [``1-dimensional Cohen-Macaulay rings'', Lect. Notes Math. 327 (1973; Zbl 0264.13012)]. The authors extend this to rings of fractions of (not necessarily commutative) rings with respect to left Ore sets of regular elements. To accomplish all this, in one go, the authors make extensive use of the theory of tilting modules and prove the following theorem, the proof of which is spread over four sections. Theorem. Let \(R\) be an associative ring with 1 and let \(S\) be a left Ore set of regular elements of \(R.\) Then the following are equivalent. (1) \(\text{pd}(S^{-1}R_{R})\leq 1,\) (2) \(S^{-1}R\oplus S^{-1}R/R\) viewed as a right \(R\)-module is a tilting module, (3) The class \(\text{Gen}(S^{-1}R_{R})\) of all right \(R\)-modules generated by \(S^{-1}R\) equals \(\ker (\text{Ext}_{R}^{1}(S^{-1}R/R,-).\) Further if \(R\) happens to be commutative then, \( S \) is just a multiplicative set generated by regular elements and, (1)--(3) are equivalent to the following three: (4) \(\text{Gen}(S^{-1}R)\) coincides with the class of all \(S\)-divisible modules, (5) \(S^{-1}R/R\) is a direct sum of countably presented \(R\)-submodules, (6) \(R\) has an \(S\)-divisible envelope. If \(R\) is commutative noetherian, then all the above are further equivalent to (7) \(S^{-1}R/R\) is a direct sum of countably generated modules \( \{M_{\alpha }:\alpha <\kappa \}\) such that for each \(\alpha <k,\) \( \bigoplus_{\beta <\alpha }M_{\beta }=T_{\alpha }^{-1}R/R\) for a submonoid \(T_{\alpha }\) of \(S.\) Reviewer's remark. The second paragraph of section 3 opens with ``Let \(R\) be a commutative ring and \(S\) a submonoid of \(R\backslash \{0\}\)'' I believe it should be ``Let \(R\) be a commutative ring and \(S\) a submonoid of \(R\) such that \(0\notin S\)''. The reason is that if \(R\) has zero divisors then \(R\backslash \{0\}\) is not closed under multiplication.