Thresholds, valuations, and K-stability (Q2308315)

K-stability is an algebro-geometric notion of stability, intimately related with the existence of Kähler-Einstein metrics on Fano manifolds. Originally defined by \textit{G. Tian} [Invent. Math. 89, 225--246 (1987; Zbl 0599.53046)] and \textit{S. K. Donaldson} [J. Differ. Geom. 62, No. 2, 289--349 (2002; Zbl 1074.53059)] using \(\mathbb{C}^*\)-degenerations of the Fano variety \(X\). \textit{K. Fujita} [J. Reine Angew. Math. 751, 309--338 (2019; Zbl 1435.14039)] and \textit{C. Li} [Duke Math. J. 166, No. 16, 3147--3218 (2017; Zbl 1409.14008)] have given a much more concrete reinterpretation of K-stability in terms of valuations on \(X\). The numerical quantities involved in this valuative criterion are then the volume and the log discrepancy. \textit{K. Fujita} and \textit{Y. Odaka} have introduced another numerical invariant, relevant to the study of K-stability, which they call the delta invariant, which involves measuring singularities of so-called ``basis-type divisors'' [Tohoku Math. J. (2) 70, No. 4, 511--521 (2018; Zbl 1422.14047)]. This is a very novel invariant, which seems quite mysterious at first. They then go on to prove that a Fano variety which satisfies \(\delta(X)\geq 1\) is K-semistable, and uniformly K-stable if \(\delta(X)>0\). The paper under review deeply develops the theory of the delta invariant, by showing that it equals \[ \delta(X) = \inf_F \frac{A_X(F)\mathrm{Vol}(L)}{\int_0^{\infty} \mathrm{Vol}}(L-xF)dx. \] Thus they show that the condition \(\delta(X)\geq 1\) is actually equivalent to K-semistability, and that \(\delta(X)>0\) is equivalent to uniform K-stability. There are many other foundational results the authors prove concerning these quantities. These are important results, which have been completely essential in many of the recent developments in the theory of K-stability of Fano varieties. The proofs are similar to those of the first author's work, which can be viewed as local analogues of the results of the paper under review. The paper is clearly written.