No semistability at infinity for Calabi-Yau metrics asymptotic to cones (Q6101154)

This paper studies complete Calabi-Yau metrics with maximal volume growth, whose asymptotic tangent cone has smooth link. These metrics are the object of active research in the past decade. Previously, \textit{R. J. Conlon} and \textit{H.-J. Hein} [Duke Math. J. 162, No. 15, 2855--2902 (2013; Zbl 1283.53045)]; ``Classification of asymptotically conical Calabi-Yau manifolds'', Preprint, \url{arXiv:2201.00870}] achieved the classification in a series of papers under the extra assumption of asymptotic conicality. The main gap in the literature is the following (reviewed in detail in the paper): \begin{itemize} \item From Riemannian geometry, by the celebrated work of Colding-Minicozzi on Ricci-flat metrics, the smoothness of the asymptotic tangent cone implies the uniqueness of the tangent cone, with a logarithmic decay rate. This paper improves the decay rate to be polynomial, in the Calabi-Yau setting. \item The noncompact version of Donaldson-Sun theory identifies the tangent cone at infinity through an algebro-geometric two step degeneration procedure. This involves first producing a K-semistable cone \(W\) from the growth rate filtration of polynomial functions, and then degenerating \(W\) to the asymptotic tangent cone \(C\) using GIT. This theory depends on \(L^2\)-techniques to produce holomorphic sections, and works symmetrically for local tangent cones and asymptotic tangent cones. In the local setting, it is known that \(W\) and \(C\) can actually disagree, which makes the theory rather intricate. The main technical contribution of this paper is that \(W=C\) in the asymptotic tangent cone case. This reveals an interesting asymmetry between the asymptotic case and the local case, which does not seem to have been anticipated in the literature. \end{itemize} The upshot is that smooth asymptotic cone in fact implies the Calabi-Yau metric is asymptotically conical, so the Conlon-Hein result applies. This is an important step in the classification problem. The rough strategy of the proof is to produce a Calabi-Yau metric on \(W\) by a non-traditional application of Tian-Yau's existence proof. This metric interpolates between \(C\) and \(W\). Using Bishop-Gromov monotonicity, and the algebro-geometric literature on the volume of tangent cones, one shows the metric on \(W\) must in fact be conical, which leads to the conclusion \(W=C\). A technical issue is that \(W\) has local singularity; the way to deal with this problem is similar to the previous work of Collins et al, based on uniform estimates for metrics on Crepant resolutions with shrinking exceptional locus. Near the infinity of \(W\), some technical work shows how to graft the Kähler potential from \(X\) to \(W\). The last part of the paper contains some quite ambitious speculations on Yau's compactification conjecture, and the case of singular asymptotic tangent cones.