I have been reading the famous paper of Alves, Bonatti, and Viana where they proved that there is an SRB measure for partially hyperbolic systems. Since I am new to this field, I have some basic questions.

1)All results were proved for $C^2$ diff, but I saw at some talks that they mentioned their results for $C^{1+\varepsilon}.$ Is it true that their results work for $C^{1+\varepsilon}.$ I am aware that the limsup condition of their result can be replaced by the liminf condition (Alves, Luzzatto. etc), but I want to know whether all their results work for $C^{1+\varepsilon}$ or not.

2)Can one give an example that a Gibbs-u state is not a hyperbolic SRB measure?


1 Answer 1


The results are true for $C^{1+}$. This paper that you mentioned uses $C^{1+}$ regularity. Actually, in this business the main thing that usually comes from $C^{1+}$ condition is control of distortion along an unstable disc.

For you second question, consider the following. Let $f$ be a $C^{1+}$ Anosov diffeomorphism of $\mathbb{T}^2$. We know that such system has only one $u$-Gibbs measure that is the SRB measure $\mu_f$.

Now, let $g$ be the north/South Pole on $S^1$ and suppose that the hyperbolicity of $g$ is weaker then the hyperbolicity of $f$, so that $f\times g$ is a partially hyperbolic diffeomorphism of $\mathbb{T}^3$. Let $p_1$ be the North Pole (repeller for $g$) and $p_2$ be the South Pole (attractor). The measure $\mu_f \times \delta_{p_1}$ is $u$-Gibbs but it is not SRB. However the measure $\mu_f \times \delta_{p_2}$ is $u$-Gibbs and SRB.

  • $\begingroup$ Thank you very much for the answer. A quick question: what do you mean by hyperbolicity of $g$ is weaker then the hyperbolicity of $f$? $\endgroup$
    – Adam
    Commented Mar 3, 2022 at 10:20
  • 1
    $\begingroup$ To talk about u-Gibbs you need to have a strong unstable direction, so you can have a splitting of the form E^s + E^c + E^u for example. In the example I mentioned, you want g to give the dynamics in the center direction. For the system f x g to really be partially hyperbolic, you want ||Df|_E^s|| < m(Dg) \leq ||Dg|| <|| Df|_E^u||, where m(Dg) is the conorm. That is what I meant by Hyperbolicity of g be weaker than Hyperbolicity of f. $\endgroup$ Commented Mar 4, 2022 at 22:10
  • $\begingroup$ Thank you very much. $\endgroup$
    – Adam
    Commented Mar 6, 2022 at 16:56

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.