Degenerating Kähler-Einstein cones, locally symmetric cusps, and the Tian-Yau metric (Q2097034)

Let \(X\) be a complex projective manifold, and let \(D\subset X\) be a smooth codimension one submanifold (a smooth divisor). A Kähler form with conic singularities along \(D\) is a closed positive \((1,1)\)-current \(\omega\) on \(X\) which is a \textit{bona fide} Kähler form on \(X\setminus D\), and which, locally near \(D\), is quasi-isometric to \[ \frac{i dz_1\wedge d\bar{z}_1}{\lvert z_1\rvert^{2(1-\beta)}} + \sum_{j\geq 2} idz_j\wedge d\bar{z}_j \] where \(D=(z_1=0)\), for some \(\beta\in (0,1)\). It is then said to have cone angle \(2\pi \beta\). It is Kähler-Einstein if it furthermore satisfies the equation \[ \operatorname{Ric}(\omega) = \sigma \omega + (1-\beta)[D] \] as currents, for some \(\sigma\in \mathbb{R}\). The authors study the limits of Kähler-Einstein forms with conic singularities along \(D\) as \(\beta\to 0\) in two different situations: \begin{itemize} \item when \(X\setminus D\) is a locally symmetric space and \(\sigma=-1\), \item when \(X\) is Fano, \(D\) is an anticanonical divisor and \(\sigma=1\). \end{itemize} Before saying more about the results, let us highlight one motivation. In Kähler geometry, Kähler forms with conic singularities attracted a lot of attention thanks to the strategy developped by Donaldson to solve the Yau-Tian-Donaldson conjecture for Fano manifolds [\textit{S. K. Donaldson}, in: Essays in mathematics and its applications. In honor of Stephen Smale's 80th birthday. Berlin: Springer. 49--79 (2012; Zbl 1326.32039)]. The idea was to consider Kähler-Einstein forms with conic singularities along an anticanonical divisor and varying cone angles, show that there exists such metrics for small angles, and show that K-stability obstructions were what prevented to reach \(\beta=1\) (in which case a conic Kähler form is actually a smooth Kähler form). Donaldson conjectured that, as \(\beta \to 0\), such metrics converge to the Ricci-flat metric constructed by Tian-Yau [\textit{G. Tian} and \textit{S. T. Yau}, J. Am. Math. Soc. 3, No. 3, 579--609 (1990; Zbl 0719.53041)], and that an existence result for \(\beta\) small should follow from the existence of Tian-Yau's metric. Ultimately, the existence of conic Kähler-Einstein forms for small angles was obtained by Berman using a different proof [\textit{R. J. Berman}, Adv. Math. 248, 1254--1297 (2013; Zbl 1286.58010)], and a variant of the original strategy yielded the proof by Chen, Donaldson and Sun of the Yau-Tian-Donaldson conjecture [\textit{X. Chen} et al., J. Am. Math. Soc. 28, No. 1, 183--278 (2015; Zbl 1312.53096, Zbl 1312.53097 and Zbl 1311.53059)]. In the current paper, the authors prove that the original picture was right, by showing that (rescaled) conic Kähler-Einstein metrics do converge to the Tian-Yau metric. Along the way, they provide numerous geometrical insights. For example, Tian-Yau's metrics are canonical: they are limits of canonical conic Kähler-Einstein metrics on \(X\) (for each angle, such a metric is unique). Furthermore, the existence of conic Kähler-Einstein metrics for small angles is reproved, using Tian-Yau's metric as a key ingredient, and their geometry is fully described as the angle goes to zero, with a precise description of what collapses, at which scale, and the possible Gromov-Hausdorf limits depending on the rescaling. The other situation studied by the authors is that of locally symmetric spaces. In this case, they show that conic Kähler-Einstein metrics converge to the complex hyperbolic metric on \(X\setminus D\). In the case of a ball quotient \(X\setminus D=\mathbb{B}^n/\Gamma\), precise asymptotics are furthermore obtained, and if \(\Gamma\) is arithmetic, this allows to show that \(\mathbb{B}^n/\Gamma\) equipped with the hyperbolic metric is the limit of smooth, compact Kähler-Einstein manifolds up to the action of a (larger and larger) group of isometries. This answers a question of Misha Kapovich about a complex hyperbolic analogue of Dehn filling of real hyperbolic cusps by Einstein manifolds [\textit{M. T. Anderson}, J. Differ. Geom. 73, No. 2, 219--261 (2006; Zbl 1100.53039)]. The main geometrical insight in the proof is the determination of a model conic metric, whose geometry and collapsing as \(\beta\to 0\) is then fully described. The model near \(D\) is obtained by Calabi's ansatz [\textit{E. Calabi}, Ann. Sci. Éc. Norm. Supér. (4) 12, 269--294 (1978; Zbl 0431.53056)] on the normal bundle of \(D\). In the ball quotient case, this is enough to determine the asymptotics, and the convergence is proven more generally for locally symmetric spaces using pluripotential theory. In the positive case, to prove convergence the authors need a global model, constructed by gluing the previous ansatz near \(D\) to the Tian-Yau metric on \(X\setminus D\). The main technical part of the proof is to prove Schauder estimates in order to deform model metrics to Kähler-Einstein metrics. There are two obstacles with respect to previously known results: there are collapsing at different scales, and the estimates need to be uniform in \(\beta\). Both obstacles are tackled using the precise understanding of the model geometry.