Pseudo-random endomorphisms in projective spaces. II (Q2256828)

From MaRDI portal
scientific article
Language Label Description Also known as
English
Pseudo-random endomorphisms in projective spaces. II
scientific article

    Statements

    Pseudo-random endomorphisms in projective spaces. II (English)
    0 references
    0 references
    23 February 2015
    0 references
    The paper is the sequel to [the author, Manuscr. Math. 142, No. 3--4, 347--367 (2013; Zbl 1279.32023)]. Given a holomorphic endomorphism \(f\) of \(\mathbb{P}^k(\mathbb{C})\) of degree \(d\geq 2\), J.-E Fornæss and N. Sibony defined the \textit{Green current \(T\) associated to \(f\)}, whose support is the Julia set of \(f\), as the limit in the sense of currents of the sequence \({(f^n)^*\omega \over d^n}\), where \(\omega\) is the Fubini-Study form of \(\mathbb{P}^k(\mathbb{C})\). Such a current has a continuous potential, and it is possible to define its self-intersection \(\mu= T^k\), which is the unique maximal entropy measure and it is the limit of the probability measures \({(f^n)^*\omega^k \over d^{kn}}\). The paper under review deals with one possible generalization of the convergence of the sequences \({(f^n)^*\omega \over d^n}\) and \({(f^n)^*\omega^k \over d^{kn}}\). Let \(\mathcal{H}_d\) be the set of points corresponding to holomorphic endomorphisms of degree \(d\) of \(\mathbb{P}^k(\mathbb{C})\) in the projective space \(\mathbb{P}^N(\mathbb{C})\), where \(N=(k+1){(d+k)!\over d!k!} -1\), of rational maps of degree \(d\) of \(\mathbb{P}^k(\mathbb{C})\), and denote by \(\mathcal{M}\) the complement of \(\mathcal{H}_d\) in \(\mathbb{P}^N(\mathbb{C})\). Let \(F:\mathbb{P}^N(\mathbb{C})\to \mathbb{P}^N(\mathbb{C})\) be a measurable map, and \(\Lambda\) an ergodic invariant measure for \(F\). Given a point \(f_0\) of \(\mathbb{P}^N(\mathbb{C})\), generic with respect to \(\Lambda\), it is possible to consider the sequence \(f_n = F^n(f_0)\) of rational maps. In the previous paper the author studied the pseudo-random sequence of iterates \(f_n\circ\cdots\circ f_0\), constructing a pseudo-random Green current and a pseudo-random Green measure having mixing properties. More precisely, he proved that, if \(\int \log\text{dist}(f, \mathcal{M}) d\Lambda(f) > - \infty\), then there exists a set \(A\subset\mathbb{P}^N(\mathbb{C})\) of full measure for \(\Lambda\) such that, for all holomorphic endomorphism \(f_0\) of \(\mathbb{P}^k(\mathbb{C})\) belonging to \(A\), the sequence \({F_n^*\omega\over d^{n+1}}\) converges towards a current \(T[f_0]\), called the \textit{Green pseudo-random current associated to \(f_0\)}, where \(F_n:= f_n\circ\cdots\circ f_0\). The Green pseudo-random current associated to \(f_0\) has continuous potential, and therefore it is possible to define its self-intersection \(T[f_0]^l\) for \(1\leq l\leq k\), and the \textit{Green pseudo-random measure associated to \(f_0\)} is the maximal self-intersection \(\mu[f_0] = T[f_0]^k\). Since the set \(A\) satisfies \(F(A)\subset A\), given \(f_0\in A\), it is possible to define the currents \(T[f_i]^l\) for all \(i\in\mathbb{N}\) and such currents have invariance properties, namely \({f_i^*T[f_{i+1}]^l\over d^l} = T[f_i]^l\) and \((f_i)_*T[f_{i}]^l = {T[f_{i+1}]^l\over d^{k-l}}\), which allow the author to prove a random mixing result. On the other hand, such a setting gives rise also to a skew-product transformation \(\tau: \mathbb{P}^k(\mathbb{C})\times \mathbb{P}^k(\mathbb{C})\to \mathbb{P}^k(\mathbb{C})\times \mathbb{P}^k(\mathbb{C})\) defined by \[ \tau(f,x)=(F(f),f(x)), \] and in the present paper, the author focuses on the ergodic properties of the skew-product map \(\tau\). Starting from the measures \(\mu[f]\) and \(\Lambda\), it is possible to define an invariant measure \(\alpha\) for \(\tau\). The author proves that \(\alpha\) is ergodic, and that if \(\Lambda\) is mixing then so is \(\alpha\). Then, following [\textit{F. Ledrappier} and \textit{P. Walters}, J. Lond. Math. Soc., II. Ser. 16, 568--576 (1977; Zbl 0388.28020)], it is possible to define a notion of ``mixed entropy'' for \(\alpha\) satisfying a relative variational principle. The author proves that the mixed entropy of \(\alpha\) is maximal and equal to \(k\log(d)\) and that the measures \(\mu[f]\) are hyperbolic with Lyapunov exponents bounded from below by \(\log(d)/2\).
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    complex dynamics
    0 references
    random maps
    0 references
    entropy
    0 references
    0 references