Problem 8

Primary tabs

Non Axiom A examples. Newhouse, Abraham-Smale, Simon, Lorenz, billiards.

  1. Axiomatic description
  2. Statistical properties
  3. For all $\epsilon$, there exists a horseshoe $X_\epsilon$ inside with $ h(f|X_\epsilon) \geq h(f) - \epsilon$
  4. Statistical properties of Lorenz in particular
  5. Any specification type property



c. proved by Katok ([1]) in dimension 2 and more generally when there is no zero Lyapunov exponents for measures with entropies arbitrarily close to $h(f)$.


  1. [katok1980lyapunov] Katok A.  1980.  Lyapunov exponents, entropy and periodic orbits for diffeomorphisms. Publications Mathématiques de l'IHÉS. 51:137–173.

(For part e.) While perhaps not related so much to the kinds of examples Rufus had in mind here, various kinds of specification properties have been developed in connection with uniqueness of measures of maximal entropy, e.g., Climenhaga-Thompson [1], Kulczycki, Kwietniak, and Oprocha [2], and Pavlov [3]; see also problem (32)


Regarding (weaker) specification properties: The approach in [C.-Thompson 2012] has recently been extended (by C.--Fisher--Thompson) to some smooth systems, including examples due to Mañé [arXiv:1703.05722], Bonatti-Viana [arXiv:1505.06371], as well as (by Burns--C.--Fisher--Thompson) to geodesic flows in nonpositive curvature [].  To my eye, the Mañé and Bonatti-Viana examples share some significant features with the Abraham-Smale and Simon examples.  The main thing driving the approach in the preprints cited above is that there is some region B such that the specification property holds for orbit segments that spend "enough" time outside of B, so B in some sense represents the "obstructions" to specification.  Then if entropy of B is smaller than total entropy, one gets a unique MME (modulo some other technicalities), and a similar result holds for pressure and equilibrium states.  It seems reasonable to me to expect that a similar game could be played with the Abraham-Smale and Simon examples, but I don't think anyone's looked at this yet.

There have been a lot of activities around all these questions, that would deserve a survey by themselves.

Related question:

SRB measures for Henon maps which are not small perturbations of 1-d maps.

In particular computer assisted estimates of the measure of SRB parameters near to Henon values a=1.4, b=0.3.

Related question.

Computer assisted estimates of the measure of SRB parameters in the quadratic family x->ax(1-x) for parameters

not too close to a=4.

Add a new comment

Log in or register to post comments