Real topological entropy versus metric entropy for birational measure-preserving transformations

Abstract

We consider a family of birational measure-preserving transformations of two complex variables, depending on one parameter for which simple rational expressions for the dynamical zeta function have been conjectured, together with an equality between the topological entropy and the logarithm of the Arnold complexity (divided by the number of iterations). Similar results have been obtained for the adaptation of these two concepts to dynamical systems of real variables, yielding to introduce a “real topological entropy” and a “real Arnold complexity”. We try to compare, here, the Kolmogorov–Sinai metric entropy and this real Arnold complexity, or real topological entropy, on this particular example of a one-parameter dependent birational transformation of two variables. More precisely, we analyze, using an infinite precision calculation, the Lyapunov characteristic exponents for various values of the parameter of the birational transformation, in order to compare these results with the ones for the real Arnold complexity. We find a quite surprising result: for this very birational example, and, in fact, for a large set of birational measure-preserving mappings generated by involutions, the Lyapunov characteristic exponents seem to be equal to zero or, at least, extremely small, for all the orbits we have considered, and for all values of the parameter. Birational measure-preserving transformations, generated by involutions, could thus allow to better understand the difference between the topological description and the probabilistic description of discrete dynamical systems. Many birational measure-preserving transformations, generated by involutions, seem to provide examples of discrete dynamical systems which can be topologically chaotic while they are metrically almost quasi-periodic. Heuristically, this can be understood as a consequence of the fact that their orbits seem to form some kind of “transcendental foliation” of the two-dimensional space of variables.