Volume estimates for Kähler-Einstein metrics and rigidity of complex structures (Q360068)

The aim of this paper is to obtain a uniform majoration for the volume of the \(r\)-neighborhood \(Z_r\) of the locus where \(|\mathrm{Riem}| \geqslant r^{-2}\). More precisely, the authors show that any \(n\)-dimensional Kähler-Einstein manifold \((M,\omega)\) of positive scalar curvature with \(\mathrm{diam}(M,\omega) \leqslant D\) and such that the cohomology class of \(\omega /2\pi\) is integral satisfies the inequality \[ \mathrm{vol}(Z_r) \leqslant C(n,D) E(M) r^4 \] where \[ E(M)= \left(\int_M |\mathrm{Riem}|^2 dV_{\omega}\right)^2, \] a topological invariant of \(M\) depending only on \(c_1(M)\) and \(c_2(M)\). This kind of estimates is of one the main inputs in a strategy to construct Kähler-Einstein metrics on \(K\)-stable Fano manifolds, as is explained by the second author in the informal note [``Discussion of the Kähler-Einstein problem'', \url{http://www2.imperial.ac.uk/~skdona/KENOTES.PDF}]. It was previously proved in dimension three by the authors in [J. Differ. Geom. 93, No. 2, 175--189 (2013; Zbl 1279.32020)] and has been independently obtained by \textit{J. Cheeger} and \textit{A. Naber} [Invent. Math. 191, No. 2, 321--339 (2013; Zbl 1268.53053)]. The above estimate is obtained as a corollary of a ``small energy'' estimate, which states that there exist \(\varepsilon,K>0\) such that whenever the renormalized energy \[ E(x,r)=r^{4-2n}\int_{B(x,r)} |\mathrm{Riem}|^2 dV_{\omega} \] is less than \(\varepsilon\), then the inequality \(|\mathrm{Riem}| \leqslant K r^{-2}\) holds pointwise on \(B(x,r/2)\). This result (referred to as Theorem 1 in the paper) is an (indirect) consequence of a rigidity result (Theorem 3) for the complex structure of quotients \[ V_{\Gamma}(a)=\big\{(z,w) \in \mathbb{C}^{n-q} \times \mathbb{C}^q /\Gamma\,;\, |z|<a, a^{-1}<|w|<a\big\} \] where \(q \geqslant 3\), \(\Gamma \subset U(q)\) acts freely on \(S^{2q-1}\) and \(a>1\) is a real number. The ideas involved in the proof of Theorem 3 are borrowed from complex geometry and use in a crucial way recent results of \textit{D. Chakrabarti} and \textit{M-C. Shaw} [Math. Ann. 349, No. 4, 977--998 (2011; Zbl 1223.32023)]. In that way the methods are very different from the ones Hamilton or Tian used to prove similar results whenever \(n=q\). Based on this rigidity result, the authors prove (Theorem 2) that if \(\varepsilon_i\to 0\), then a based Gromov-Hausdorff limit of \((M_i, \varepsilon_i^{-2} \omega_i,x_i)\) cannot be a product \(\mathbb{C}^{n-q} \times \mathbb{C}^q /\Gamma\) with \(q \geqslant 3\) and \(\Gamma \subset U(q)\) acting freely on \(S^{2q-1}\). Theorem 1 is finally derived from Theorem 2 with the important input of the paper by \textit{J. Cheeger} et al. [Geom. Funct. Anal. 12, No. 5, 873--914 (2002; Zbl 1030.53046)].