F-regularity does not deform

From MaRDI portal



Abstract: We show that the property of F-regularity does not deform, and thereby settle this longstanding open question in the theory of tight closure. Specifically, we construct a three dimensional domain R which is not F-regular (or even F-pure), but has a quotient R/tR which is F-regular. Similar examples are also constructed over fields of characteristic zero. Our work has an immediate application to questions related to flat base change: Hochster and Huneke showed that if (A,m) -> (R,n) is a flat map with regular generic and closed fibers, R is excellent, and A is weakly F-regular, then R is weakly F-regular. They asked if the condition on the fibers may be relaxed by requiring instead that the two fibers are F-regular. Our result shows that this condition is not enough, even if the base ring A is a discrete valuation ring: taking A=Kt as a subring of the ring R above, the closed fiber R/tR and the generic fiber R[1/t] are F-regular, whereas R is not.


The author constructs three-dimensional \(\mathbb N\)-graded integral domains which are not F-regular but have quotients by a principal ideal which are. In other words, F-regularity does not deform, thus settling an open question from the theory of tight closure. Earlier work by Aberbach-Katzman-MacCrimmon, Hochster-Huneke, Singh and Smith has shown that in special cases \(F\)-regularity does deform [see \textit{I. Aberbach, M. Katzman} and \textit{B. MacCrimmon}, J. Algebra 204, No. 1, 281-285 (1998; Zbl 0915.13003), \textit{M. Hochster} and \textit{C. Huneke}, Trans. Am. Math. Soc. 346, No. 1, 1-62 (1994; Zbl 0844.13002), \textit{A. K. Singh}, J. Pure Appl. Algebra 140, No. 2, 137-148 (1999; see the preceding review Zbl 0946.13001) and \textit{K. E. Smith}, ``\(F\)-regularity deforms for \(\mathbb Q\)-Gorenstein rings (preprint)]. This paper also answers another open question in the negative: Does there exist an integral domain which is not F-pure but one of its quotients by a principal regular ideal is an F-pure integral domain? This is much stronger than saying that F-purity does not deform, the latter fact having been established first by \textit{R. Fedder} [Trans. Am. Math. Soc. 278, 461-480 (1983; Zbl 0519.13017)] and later by Singh in the paper quoted above. The author shows also the corresponding results for characteristic 0.




Cited in
(28)








This page was built for publication: F-regularity does not deform

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q4264548)