Typical dynamics of plane rational maps with equal degrees (Q317003)

From MaRDI portal
scientific article
Language Label Description Also known as
English
Typical dynamics of plane rational maps with equal degrees
scientific article

    Statements

    Typical dynamics of plane rational maps with equal degrees (English)
    0 references
    0 references
    0 references
    0 references
    30 September 2016
    0 references
    Consider a dominant rational map of the complex projective plane \(\mathbb P^2\), denoted by \(f=[f_1:f_2:f_3]\) in homogeneous coordinates, where \(f_i\) are homogeneous polynomials with the same degree, and have no common factors of positive degree. The algebraic degree is the common degree of the \(f_i\), denoted by \(d(f)\). The first dynamical degree is the limit \[ \lambda_1(f):=\lim_{n\to\infty}d(f^n)^{1/n}. \] For any generic point \(p\in\mathbb P^2\), the number of elements in \(\{f^{-1}(p)\}\) is said to be the topological degree, denoted by \(\lambda_2(f)\). Since \(\lambda_2(f^n)=\lambda_2(f)^n\), the second dynamical degree of \(f\) is \(\lambda_2(f)\). Denote by \(\mathcal I=\mathcal I(f)\subset\mathbb P^2\) as the common zeros of the components \(f_i\), where \(f\) is not well defined. For any point \(p\in\mathcal I\), let \(f(p)\) consist of all the possible limits \(\lim_{p_k\to p}f(p_k)\). If \(f(\mathbb P^2\setminus\mathcal I)\) is not contained in an algebraic curve, then \(f\) is called dominant. If \(V\) is an irreducible curve with \(f(V)\) a point, \(V\) is said to be exceptional. The authors show that if \(d(f)=\lambda_2(f)\geq 2\), then there is an open subset of linear maps \(T\in\mathrm{Aut}(\mathbb P^2)\) such that \(d(f)=\lambda_1(f_T)=\lambda_2(f_T)\), the map \(f_T:=T\circ f\) has two invariant and ergodic measures \(\mu\) and \(\nu\) of maximal entropy \(\log d(f)\), where \(f_T\) is uniformly hyperbolic of saddle type on \(\mathrm{supp}(\nu)\), \(f_T\) is uniformly expanding on \(\mathrm{supp}(\mu)\), neither measure is supported in an algebraic curve, where the saddle type means that one Lyapunov exponent is positive and the other is negative. Some other properties are the consequences of the proof of the main results. The tricky part in the proofs of the main results is based on the choice the linear map \(T\in\mathrm{Aut}(P^2)\). Take a line \(L_0\subset\mathbb P^2\) and \(p_0\in\mathbb P^2\) with \(p_0\not\in L_0\) such that {\parindent=0.7cm\begin{itemize}\item[(A)] \(p_0\not\in f(\mathcal I)\cup f(L_0)\); \item[(B)] \(L_0\cap\mathcal I=\emptyset\); \item[(C)] \(p_0\) is a regular value of \(f\) with \(\lambda_2\) distinct preimages; \item[(D)] \(\deg f(L_0)\geq 2\), and \(L_0\) is not exceptional. \end{itemize}} And, consider a surjective linear map \(T_0:\mathbb P^2\setminus\{p_0\}\to L_0\), and for any \(\delta>0\), set \[ \mathcal T(T_0,\delta):=\{T\in\mathrm{Aut}(\mathbb P^2):\mathrm{dist}(T(p),T_0(p))<\delta,\,\forall p\not\in B_{\delta}(p_0)\}. \] For any given sufficiently small \(\epsilon>0\), there is \(\delta>0\) such that for any \(T\in\mathcal T(T_0,\delta)\), the existence of the measure \(\nu\) of saddle type for \(f_T\) is based on (A), (B), and (D), and the choice of the map \(T_0\), where the measure \(\nu\) is on \(B_{\epsilon}(L_0)\) with entropy \(\log d(f)\), and the induced one-dimensional map \(f_0=T_0\circ f\) is hyperbolic. On the other hand, for any given sufficiently small \(\epsilon>0\), there exists \(\delta>0\) such that for any \(T\in\mathcal T(T_0,\delta)\), \(f_T\) is uniformly expanding on \(\mathcal A=\cap_{n\geq 0}f^{-n}_T(B_{\epsilon}(f^{-1}(p_0)))\), and the support of \(\mu\) is contained in \(B_{\epsilon}(f^{-1}(p_0))\) by using (A), (B), and (C). By contradiction and some elegant arguments, the authors prove that neither measure is supported in an algebraic curve. In the last section, the authors study a family of rational maps \(f_t=T_t\circ f\), \(t\in[0,1]\), where \(f=g\circ h\), \[ g[x_1:x_2:x_3]=[x^2_1:x^2_2:x^2_3], \] \[ h[x_1:x_2:x_3]=[x_1(-x_1+x_2+x_3):x_2(x_1-x_2+x_3):x_3(x_1+x_2-x_3)], \] and \[ T_t[x_1:x_2:x_3]=[x_1:x_2:x_3]-\frac{1-t}{3}(x_1+x_2+x_3)[1:1:1],\;t\in[0,1]. \] By symmetry of the map \(f\), the authors show the main results more directly for \(t\in(0,1]\) close enough to \(0\). And, the support of the measure \(\mu\) is contained in \(\mathbb{RP}^2\subset\mathbb{P}^2\), whereas for the measure \(\nu\) it is not the case.
    0 references
    0 references
    0 references
    algebraic curve
    0 references
    rational maps
    0 references
    ergodic properties
    0 references
    dynamical degree
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references