Decay of volumes under iteration of meromorphic mappings. (Q1774099)

Let \(X\) be a compact Kähler manifold of dimension \(N\) with a Kähler form \(\omega\) normalized so that \(\int_X\omega^N=1\). Let \(f:X\to X\) be a meromorphic self-map. For any \(l\in\{1,\ldots, N\}\), one can define the current \(f^\ast\omega^l\) and the degree \(\delta_l(f):=\int_X f^\ast \omega^l\wedge \omega^{N-l}\). The main result of this paper is a precise lower estimate on the volume of \(f^j(\Omega)\) for all \(j\in{\mathbb N}\) and open sets \(\Omega\subset X\). More detailed, let \(\lambda>1\) and assume that \(\delta_1(f^j)\lambda^{-j}\) is bounded from above. Then there exists \(C>0\) such that for all \(j\in \mathbb N\) and open sets \(\Omega\subset X\) the volume \(\text{Vol}_\omega(f^j\Omega)\) is bounded from below from \(\exp(-C\lambda^j/\text{Vol}_\omega(\Omega))\). This result is then used to construct invariant Green's functions for algebraically stable maps whose associated data satisfying certain growth estimates, generalizing previous results by Sibony, Cantat and Favre and the author himself. More precisely, let \(f:X\to X\) be algebraically stable, (that is, \((f^{n+1})^\ast=(f^{n})^\ast\circ f^\ast\) in cohomology). Let \(\lambda_1(f):=\limsup_{j\to \infty}[\delta_1(f^j)]^{1/j}\) be the first dynamical degree of \(f\). Then there exists a class \(\alpha\in H^{1,1}_{psef}(X,\mathbb R)\) (the cohomology group of classes that can be represented by a positive closed current) such that \(f^\ast\alpha=\lambda_1(f)\alpha\). To any \(\theta\) smooth closed real \((1,1)\)-form such that \(\{\theta\}=\alpha\) one can associate an upper semi-continuous function \(v^\theta_{\min}\). Under the hypotheses that \(\text{Vol}_\omega(v^\theta_{\min}<-t)\) has a sub-exponential type growth as \(t\to\infty\), that \(\lambda_1(f)>1\) and \(\delta_1(f^j)\lambda_1(f)^{-j}\) is bounded from above, the author proves that there exists a positive closed \((1,1)\)-current \(T_\alpha\) with minimal singularities such that \(f^\ast T_\alpha=\lambda_1(f)T_\alpha\) and \(\{T_\alpha\}=\alpha\). The paper concludes with a careful discussion of the hypotheses assumed and some open questions about them. In particular, the hypothesis \(\lambda_1(f)>1\) is dynamically significant, for otherwise \(f\) is bimeromorphic with zero topological entropy. The request that \(\delta_1(f^j)\lambda_1(f)^{-j}\) is bounded from above is connected to non-existence of \(f\)-invariant fibrations. Finally, the estimate on the growth of the volume of \(v^\theta_{\min}\) is satisfied in all cases considered so far in the literature, e.g., [\textit{N. Sibony}, Panor. Synth. 8, 97--185 (1999; Zbl 1020.37026)]. Thus it is an open question whether such an hypothesis is always satisfied.