← All problems

Typical Periodic Optimization (TPO) conjecture

Let T:X→XT:X\to X be a dynamical system in the class considered by the conjecture, such as a uniformly expanding map or an Axiom A system, and let F\mathcal{F} be an admissible regularity class of potentials on XX. The Typical Periodic Optimization conjecture asserts that, for a generic f∈Ff\in\mathcal{F}, there is a unique maximizing invariant probability measure μf∈MT(X)\mu_f\in\mathcal{M}_T(X), meaning ∫Xf dμf=sup⁡μ∈MT(X)∫Xf dμ\int_X f\,d\mu_f=\sup_{\mu\in\mathcal{M}_T(X)}\int_X f\,d\mu, and this measure is supported on a periodic orbit: there exist x∈Xx\in X and p≥1p\geq 1 such that Tpx=xT^p x=x and μf=1p∑j=0p−1δTjx\mu_f=\frac{1}{p}\sum_{j=0}^{p-1}\delta_{T^j x}.

References

Source: arXiv

Progress summary

Refreshed
Open · progress reported

The conjecture is confirmed in several important settings, but new examples claim that its broad regularity version fails, so the overall picture is unsettled.

The Typical Periodic Optimization conjecture, proposed by Yuan and Hunt in 1999, predicts that generic optimization potentials have a unique maximizing measure supported on a periodic orbit. Its validity depends strongly on the dynamical system and the regularity class.

Known results

  • Contreras (2013, 2016): TPO holds generically for Lipschitz expanding open maps.
  • Yuan–Hunt conjecture: generic Hölder, Lipschitz, and suitable C1C^1 observables have periodic maximizing measures for Axiom A and uniformly expanding systems.
  • Contreras (2020): for expanding maps, an open dense subset of CrC^r observables has a unique periodic maximizing measure.
  • Joint TPO (2025): an open dense set of map–potential pairs has a unique periodic maximizing measure, including expanding maps, Anosov diffeomorphisms, and beta-transformations.

October 2026 regularity claim

Jiang, Li, and Sun report a construction showing failure at Dini regularity and identify weak log-log Dini as the threshold for freezing phase transitions, while retaining Dini regularity for a quantitative freezing analogue. This is a claimed advance, not independently verified here; it concerns regularity thresholds rather than failure in every smoother setting. Separately, a Gauss-map paper gives an open set of C0,αC^{0,\alpha} potentials without TPO for every α∈(0,1]\alpha\in(0,1].

Current status (as of October 2026): TPO is established in several expanding and hyperbolic settings, while the reported Dini-regularity failure and broader counterexamples leave the unrestricted conjectural picture open.

Sources

Solutions 0

No solutions have been posted yet.