Affine Noetherian PI-rings have enough clans (Q1065116): Difference between revisions

The past few years have seen a lot of work on the problem of obtaining a workable localisation (in Ore's sense) at a prime ideal P of a Noetherian ring R. A detailed account of much of this research (but not including the work under review here) may be found in [\textit{A. V. Jategaonkar}, Localization in Noetherian Rings, Lond. Math. Soc. Lect. Note Ser. 98 (1986)]; and [\textit{K. A. Brown}, Lect. Notes Math. 1146, 355-366 (1985; Zbl 0572.16002)] is a short survey of the same topic. Prime ideals P and Q of R are linked if there are ideals \(A\subset B\) of R with B/A a torsion free left R/Q-module and a torsion free right R/P- module. If Q and P are linked primes and \(\ell\) is an Ore subset of \(\ell (P)\) then it is not hard to see that \(\ell \subset \ell (Q)\). One of the basic problems in this area is to determine those rings R for which \(\cap_{Q}\ell (Q)\) is an Ore set, where the intersection is taken over all those primes Q connected to P by a chain of links. In the present paper the author confirms that this is so when R is an affine Noetherian PI-ring. If R contains an uncountable central subfield, this is an earlier result of the author, an account of which is given in the book cited above; considerable technical ingenuity is needed to remove this extra hypothesis.