Weak solutions to degenerate complex Monge-Ampère flows. II (Q261164)

This is a continuation of [the authors, Math. Ann. 362, No. 3--4, 931--963 (2015; Zbl 1322.35094)] but this time, focusing on Kähler manifolds instead of pseudo-convex subsets of \(\mathbb{C}^n\). The authors are interested in parabolic-type degenerate Monge-Ampère-type equations of the form \[ e^{\frac{\partial \phi_t}{\partial t} +F(t,x,\phi_t)}\mu(t,x) = (\omega_t + dd^c \phi_t)^n. \] Here \(\omega_t\) is a continuous family of semi-positive \((1,1)\)-forms on \(X\), a compact Kähler manifold of complex dimension \(n\), \(F\) is continuous and non-decreasing in the last variable, \(\mu\) is a bounded continuous nonnegative volume form in \(X\) and \(\phi\) is the unknown function. The authors propose a viscosity theory in this context, providing definitions of sub/super solution and connecting these to weak solutions of the Kähler-Ricci flow by choosing \(F(t,x,\phi)=c\phi\) for a constant \(c\). This is particularly interesting when one considers the Kähler-Ricci flow in a singular context. There are several results stated in this paper. Firstly, given a continuous initial potential \(\phi_0\), the Cauchy problem stated above admits a unique viscosity solution, which is the upper envelope of all subsolutions. Applications to various geometric problems in relation with the Song-Tian program are presented. Over a \(\mathbb{Q}\)-Calabi-Yau variety, the (normalized) Kähler-Ricci flow can be uniquely run from a closed positive current with continuous potentials towards the unique (Ricci-flat) Kähler-Einstein current. A similar statement is given for a canonical model, more precisely, to a general-type projective algebraic variety with canonical bundle big and nef and with only canonical singularities. Some ideas are presented to overcome the difficulty of understanding the behavior of the Kähler-Ricci flow over the flips. One needs to mention that the proofs are based on global comparison principles at the levels of potentials inspired from the local case of \(\mathbb{C}^n\) studied in [loc. cit.] There is a flaw in the extension to manifolds and consequently the results stated above only hold on the ample locus. See the erratum from the authors [Commun. Pure Appl. Math. 70, No. 5, 815--821 (2017; Zbl 1472.32020)]. Nevertheless, it is likely that the results concerning the Kähler-Ricci flow hold as announced, without any restriction.