\(G\)-uniform stability and Kähler-Einstein metrics on Fano varieties (Q2073773)

On a compact Kähler manifold \(X\), the search for Kähler-Einstein metrics, namely Kähler metrics \(\omega\) satisfying \[ \mathrm{Ric}\, \omega = \lambda \omega \] for \(\lambda \in \mathbb{R}\), separates into three cases. When \(\lambda\) is negative, this forces the canonical class \(K_X\) of \(X\) to be ample and in particular \(X\) is of general type and Kähler-Einstein metrics always exist [\textit{T. Aubin}, Bull. Sci. Math., II. Sér. 102, 63--95 (1978; Zbl 0374.53022); \textit{S.-T. Yau}, Commun. Pure Appl. Math. 31, 339--411 (1978; Zbl 0369.53059)]. When \(\lambda=0\), this forces \(K_X\) to be numerically trivial and hence \(X\) is Calabi-Yau, and again Kähler-Einstein metrics always exist [\textit{S.-T. Yau}, Commun. Pure Appl. Math. 31, 339--411 (1978; Zbl 0369.53059)]. The final case is that of Fano manifolds, where \(-K_X\) is ample, and here the obstruction is actually obstructed. A famous result of Chen-Donaldson-Sun characterises the existence of Kähler-Einstein metrics in this case through the algebro-geometric notion of K-polystability [\textit{X. Chen} et al., J. Am. Math. Soc. 28, No. 1, 235--278 (2015; Zbl 1311.53059)], proving what is known as the Yau-Tian-Donaldson conjecture. The Kähler-Einstein problem is also interesting in the case where one considers singular varieties, rather than (smooth) Kähler manifolds, especially since the notion of K-stability is nowadays in itself an interesting algebro-geometric condition, primarily due to its links with the moduli theory of algebraic varieties. In this direction, Li-Tian-Wang proved the so-called ``uniform'' version of the Yau-Tian-Donaldson conjecture, proving that uniform K-stability is equivalent to the existence of a Kähler-Einstein metric on a (possibly singular) Fano variety with \textit{discrete} automorphism group [the author et al., Commun. Pure Appl. Math. 74, No. 8, 1748--1800 (2021; Zbl 1484.32041)]. Here uniform K-stability is a stronger version of K-stability, asking for a uniform lower bound on the \textit{Donaldson-Futaki invariant of test configurations} in terms of an appropriate norm [the reviewer, Int. Math. Res. Not. 2016, No. 15, 4728--4783 (2016; Zbl 1405.32032); \textit{S. Boucksom} et al., Ann. Inst. Fourier 67, No. 2, 743--841 (2017; Zbl 1391.14090)]. The main purpose of the paper under review is to remove the discrete-automorphism-group hypothesis: setting \(G=\mathrm{Aut}(X)\), the existence of a Kähler-Einstein metric on \(X\) is equivalent to \(G\)-uniform K-stability of \(X\). Here \(G\)-uniform K-stability involves considering \(G\)-equivariant test configurations, and modifying their norm by twisting by one-parameter subgroups of \(G\), as first suggested by \textit{T. Hisamoto} [``Stability and coercivity for toric polarizations'', Preprint, \url{arXiv:1610.07998}]. In particular this algebro-geometrically characterises the existence of Kähler-Einstein metrics on \textit{any} Fano variety, in essence finishing the problem. The strategy is also new even in the case \(X\) is smooth (though the ``stability implies existence'' direction of the result follows from [\textit{V. Datar} and \textit{G. Székelyhidi}, Geom. Funct. Anal. 26, No. 4, 975--1010 (2016; Zbl 1359.32019)]. The basic strategy of the proof goes back to important work of \textit{R. J. Berman} et al. [J. Am. Math. Soc. 34, No. 3, 605--652 (2021; Zbl 1487.32141)], who (after Chen-Donaldson-Sun) gave a new proof that the existence of a Kähler-Einstein metric is equivalent to \textit{uniform} K-stability in the discrete automorphism group case. There are two main new steps. The first is to characterise in terms of valuations the notion of \(G\)-uniform K-stability; this largely follows prior work of Li along with work of \textit{K. Fujita} [J. Reine Angew. Math. 751, 309--338 (2019; Zbl 1435.14039); the author, Duke Math. J. 166, No. 16, 3147--3218 (2017; Zbl 1409.14008)], with the main new difficulties coming from the twisting action on test configurations, filtrations and valuations. The second new step is to pass to a resolution of singularities (unnecessary in the Berman-Boucksom-Jonsson) and argue ``uniformly'' on the resolution. This second step is reminiscent of the approach of Li-Tian-Wang, but serious new ideas are needed beyond this prior work. While the paper is quite long, the length is largely because it is very comprehensive and very clearly written. Experts in the field will be largely familiar with much of the first half of the paper.