Pith. sign in

REVIEW 1 major objections 3 minor 1 cited by

Typical Uniqueness in Ergodic Optimization

T0 review · 1 major / 3 minor · reviewed 2026-08-07 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read The paper proves that in any topological dynamical system, potentials in a separable Banach space with non-unique maximizing measure are confined to a countable union of hypersurfaces.

desk verdict Theorem 2 is correct and a genuine strengthening, but the abstract overstates it by dropping the dense-embedding hypothesis that is essential. read the letter →

arxiv 2506.01518 v1 pith:WCXEJC4C submitted 2025-06-02 math.DS

Oliver Jenkinson , Xiaoran Li , Yuexin Liao , Yiwei Zhang This is my paper
classification math.DS MSC 37A9937B0237D2037F05
keywords ergodicoptimizationmaximizingmeasuretypicaluniquenessseparableBanachspacehypersurfaceGateauxdifferentiabilityconvexfunctionprevalence
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

The paper proves that in any topological dynamical system -- a continuous map, a group or semigroup action, or an upper semi-continuous multivalued map -- and for any separable Banach space $V$ of real-valued potentials that is densely and continuously embedded in the space of continuous functions, the potentials whose maximizing invariant measure is not unique form an extremely small set: a countable union of hypersurfaces. Each hypersurface is the graph, over a codimension-one subspace, of a function that is the difference of two Lipschitz convex functions. This is a strictly stronger notion of 'typical' than the residual and prevalent uniqueness results previously known for the same setting. The proof works by showing that a potential $f$ has a unique maximizing measure exactly when the maximum ergodic average $\beta(f)$ is directionally differentiable at $f$, so a classical theorem about differentiability of convex functions places all exceptional potentials on countably many hypersurfaces.

What carries the argument

The object carrying the argument is the maximum ergodic average $\beta(f)=\sup_{\mu\in M}\int f\,d\mu$, regarded as a function on the potential space $V$. It is convex and Lipschitz continuous, and the set of $f$-maximizing measures is exactly the subdifferential of $\beta$ at $f$. The proof's key identity is that $\beta$ is Gateaux differentiable (differentiable in every direction) at $f$ if and only if this subdifferential is a singleton, i.e. $f\in V^!$. This reduces the ergodic-optimization question to the differentiability theory of convex functions, where the theorem that a continuous convex function on a separable Banach space is differentiable off a countable union of hypersurfaces applies directly.

What would settle it

Find a topological dynamical system and a separable Banach potential space $V$ (densely and continuously embedded in $C(X)$) for which the set of potentials with at least two maximizing measures has nonempty interior. Because every hypersurface has empty interior, such an open set cannot be covered by countably many hypersurfaces, so this would directly contradict Theorem 2.

Watch

Extended reading notes

Core claim

The central claim is Theorem 2: if $V$ is a separable Banach space densely and continuously embedded in $C(X)$, then the complement in $V$ of the set $V^!$ of potentials with a unique maximizing measure is contained in a countable union of hypersurfaces. A hypersurface here means the graph over a hyperplane of a function that is the difference of two Lipschitz convex functions, sometimes called a delta-convex hypersurface. The result upgrades earlier theorems that $V^!$ is residual or prevalent, because the class of countable unions of hypersurfaces is a finer class of negligible sets than either the meager sets or the Haar null sets. The proof identifies the exact mechanism: $\beta(f)=\sup_{\mu\in M}\int f\,d\mu$ is a continuous convex function on $V$, and $f\in V^!$ if and only if $\beta$ is Gateaux differentiable (all directional derivatives exist) at $f$; the differentiability theorem for convex functions on separable Banach spaces then covers the exceptional set by countably many hypersurfaces.

Load-bearing premise

The result leans on a deep theorem saying that a continuous convex function on a separable Banach space can fail to be differentiable only on a countable union of hypersurfaces, and on the potential space being dense in the continuous functions so that distinct maximizing measures can be separated by integrals of admissible potentials; if either fails, the conclusion need not hold.

Editorial extensions

If this is right

  • For every topological dynamical system and every separable Banach potential space $V$ densely embedded in $C(X)$, the complement of $V^!$ is contained in countably many hypersurfaces, so uniqueness is typical in a sense strictly finer than both residual and prevalent.
  • The result covers not just single continuous maps but also group and semigroup actions and upper semi-continuous multivalued maps, whenever invariant measures exist.
  • If the set of invariant measures has only countably many extreme points, the exceptional set is contained in countably many hyperplanes -- proper closed codimension-one subspaces -- rather than merely Lipschitz-convex graphs.
  • The proof supplies a shorter route to the previously known generic-uniqueness and prevalent-uniqueness theorems, as a by-product of the differentiability equivalence.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • If the central claim is right, 'almost every potential' in ergodic optimization can be sharpened from prevalence to a finer notion of negligibility: outside a countable union of hypersurfaces, maximizing measures are unique, which may be the natural setting for asking whether the unique maximizing measure is also periodic.
  • The same convex-differentiability mechanism should apply to other optimization problems over compact convex sets of measures, such as Lyapunov-exponent optimization, whenever the relevant differentiability theorem for convex functions is available; the paper does not develop these applications.
  • A testable extension is to compute, for a concrete finite-dimensional family of potentials (for example on a full shift or an expanding circle map), the hypersurfaces carrying non-unique maximizing measures and check numerically that the exceptional set is exactly their union.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

1 major / 3 minor

Summary. The paper studies the set V^! of potentials with a unique maximizing measure in ergodic optimization. Its main new result, Theorem 2, states that if V is a separable Banach space densely and continuously embedded in C(X), then V \ V^! is contained in a countable union of delta-convex hypersurfaces. The proof centers on the function beta(f) = sup_{mu in M} int f d mu, showing that beta is Gateaux differentiable at f exactly when f has a unique maximizing measure, and then applying Zajicek's theorem on differentiability of continuous convex functions. The paper also gives short proofs of the known residual uniqueness theorem (Theorem 1) and prevalent uniqueness theorem (Theorem 4) via the same equivalence.

Significance. If the main theorem is correct, it is a genuine strengthening of earlier generic and prevalent uniqueness results: the exceptional set for non-uniqueness is not merely meager or Haar null but is contained in countably many codimension-one graphs. The proof is mostly self-contained, the key inequality (2.1) is elegant, and the external theorems of Fort, Zajicek, and Christensen are used appropriately. The central mathematical argument is sound. However, the paper as written advertises a false statement in the abstract, so the presentation must be corrected before acceptance.

major comments (1)
  1. [Abstract] The abstract's first sentence overclaims: the uniqueness result is stated for 'any separable Banach space B of continuous functions', but Theorem 2 requires V to be densely and continuously embedded in C(X). The density condition is essential, not a technical convenience: it is used in the last paragraph of the proof of Theorem 2 (and in the proof of Theorem 1) to find g in V separating two distinct maximizing measures. Without density the advertised statement is false. For example, take X={1,2}, T=id, M all probability measures, and V the one-dimensional space of constant functions. Then every f in V is maximized by every mu in M, so V^! is empty, while beta(f)=f is linear and hence Gateaux differentiable everywhere; moreover, in a one-dimensional space every hypersurface in the paper's definition is a singleton, so V is not coverable by countably many hypersurfaces. The abstract should be revised to state the dense-continuous-embedding hypothesis explicitly, and the standing assumptions of compact metrizable X and nonempty M should also be reflected.
minor comments (3)
  1. [Abstract and Section 2] The abstract says 'any topological dynamical system' without qualification, but the body assumes that X is compact metrizable and that the set M of invariant measures is nonempty; please align the abstract with these standing assumptions.
  2. [Section 2, proof of Theorem 1] In the lower semi-continuity argument, the open set U defined by int g d mu < int g d mu_1 contains mu_2 rather than mu_1; a sentence making this explicit would avoid confusion for the reader.
  3. [Section 2, Proposition 3] The proof states that extreme points of M_max(f) are extreme in M; this is standard (M_max(f) is a face of M), but a brief reference here would make the proof more self-contained.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: the hypersurface-covering conclusion follows from an independently proved differentiability-uniqueness equivalence plus Zajicek's external theorem; self-citations are not load-bearing.

full rationale

The paper's central derivation is self-contained. Starting from beta(f)=sup integral f dmu over invariant measures, it proves that beta is Lipschitz and convex, then proves—using the dense embedding of V in C(X)—that f has a unique maximizing measure if and only if beta is Gateaux differentiable at f. The exceptional-set conclusion then follows by applying Zajicek's theorem on Gateaux differentiability of continuous convex functions on separable Banach spaces, which is an external result not derived from or assumed by the paper. The only self-citations are [26] (whose generic-uniqueness result is reproved as Theorem 1), [28] (a companion paper on multi-valued systems, used only for context on non-emptiness of M), and [20] in the introduction; none is used to justify the hypersurface theorem. The abstract's wording omits the dense-embedding hypothesis, but that is an accuracy/overclaim issue about the theorem's hypotheses, not a case of the conclusion being assumed or fitted. No fitted parameter is renamed as a prediction, and no uniqueness theorem is imported from the authors' previous work as a forced choice. Hence no circular step can be exhibited.

Assumptions & free parameters 0 free parameters · 5 assumptions · 0 invented entities

No free parameters or invented entities are used. The central claim rests on the compact-metrizable structure of invariant measures, the dense-continuous embedding hypothesis for V, and three standard differentiability theorems from functional analysis (Fort, Zajicek, Christensen). These are external results, not assumptions of the conclusion, so the paper itself is not circular.

assumptions (5)
  • standard math Fort's theorem: for an upper semi-continuous set-valued map taking non-empty compact values in a metric space, the set of continuity points is residual.
    Invoked in the proof of Theorem 1 to pass from upper semi-continuity of the maximizing-measure map to a residual set of continuity points; cited to [17].
  • standard math Zajicek's theorem: a continuous convex function on a separable Banach space is Gateaux differentiable except on a set that can be covered by countably many hypersurfaces.
    This is the core tool in Theorem 2. It is cited as [40] and [3, Thm. 4.20], not proved in the paper.
  • standard math Christensen's theorem: a Lipschitz real-valued function on a separable Frechet space is Gateaux differentiable outside a Haar null set.
    Used in Theorem 4 to obtain prevalence; cited to [11, Thm. 2].
  • domain assumption The set M of invariant Borel probability measures is non-empty, compact, and metrizable in the weak* topology for the classes of systems considered.
    Stated in Section 2. For continuous maps and amenable actions this follows from Krylov-Bogolioubov-type results; for multi-valued maps from existence of infinite orbits. If M were empty or non-compact, maximizing measures and the arguments using weak* compactness would fail.
  • domain assumption V is a separable Banach space densely and continuously embedded in C(X).
    This is the standing hypothesis of Theorems 1, 2, and 4. Denseness lets integrals separate distinct measures; continuous embedding makes beta Lipschitz; separability makes Zajicek's theorem applicable.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Typical Uniqueness in Ergodic Optimization." pith.science (2026). https://pith.science/paper/WCXEJC4C

@misc{pith2026250601518,
  author       = {Pith},
  title        = {Pith review of: Typical Uniqueness in Ergodic Optimization},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/WCXEJC4C}},
  note         = {Machine review of arXiv:2506.01518}
}
abstract

For ergodic optimization on any topological dynamical system, with real-valued potential function $f$ belonging to any separable Banach space $B$ of continuous functions, we show that the $f$-maximizing measure is typically unique, in the strong sense that a countable collection of hypersurfaces contains the exceptional set of those $f\in B$ with non-unique maximizing measure. This strengthens previous results asserting that the uniqueness set is both residual and prevalent.

Discussion (0). Sign in to comment.

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. Ergodic Optimization with Linear Constraints

    math.DS 2026-08 conditional novelty 6.0 of 10

    Under linear constraints on invariant measures, the constrained ergodic optimization problem still admits optimizers, has a unique optimizer for generic and prevalent objective functions, and satisfies a duality formu...

Reference graph

Works this paper leans on

41 extracted references · 40 canonical work pages · cited by 1 Pith paper

  1. [26]

    Jenkinson, Ergodic optimization,Discrete & Cont

    O. Jenkinson, Ergodic optimization,Discrete & Cont. Dyn. Sys.,15(2006), 197–224

    work page 2006

  2. [34]

    I. D. Morris, Prevalent uniqueness in ergodic optimisation,Proc. Amer. Math. Soc.,149 (2021), 1631–1639

    work page 2021

  3. [1]

    Akin,The general topology of dynamical systems, Amer

    E. Akin,The general topology of dynamical systems, Amer. Math. Soc., Providence, 1993

    work page 1993

  4. [2]

    C. D. Aliprantis & K. C. Border,Infinite-dimensional analysis. A hitchhiker’s guide, Second edition, Springer-Verlag, Berlin, 1999

    work page 1999

  5. [3]

    Benyamini & J

    Y. Benyamini & J. Lindenstrauss,Geometric nonlinear functional analysis. Vol. 1, Amer. Math. Soc. Colloq. Publ., 48 American Mathematical Society, Providence, RI, 2000

    work page 2000

  6. [4]

    Bochi, Ergodic optimization of Birkhoff averages and Lyapunov exponents,Proceedings of the International Congress of Mathematicians, Rio de Janeiro 2018

    J. Bochi, Ergodic optimization of Birkhoff averages and Lyapunov exponents,Proceedings of the International Congress of Mathematicians, Rio de Janeiro 2018. Vol. III. Invited lectures, 1825–1846, World Sci. Publ., Hackensack, NJ, 2018

    work page 2018

  7. [5]

    Bochi, Genericity of periodic maximization: Proof of Contreras’ theorem following Huang, Lian, Ma, Xu, and Zhang, 2019, available athttps://personal.science.psu

    J. Bochi, Genericity of periodic maximization: Proof of Contreras’ theorem following Huang, Lian, Ma, Xu, and Zhang, 2019, available athttps://personal.science.psu. edu/jzd5895/docs/Contreras_dapres_HLMXZ.pdf

    work page 2019

  8. [6]

    Bochi & Y

    J. Bochi & Y. Zhang, Ergodic optimization of prevalent super-continuous functions.Int. Math. Res. Not.,19(2016), 5988–6017

    work page 2016

Show all 41 references
  1. [7]

    Bousch, Le poisson n’a pas d’arˆ etes.Ann

    T. Bousch, Le poisson n’a pas d’arˆ etes.Ann. Inst. Henri Poincar´ e Probab. Stat.,36(2000), 489–508

    2000

  2. [8]

    Bousch, La condition de Walters,Ann

    T. Bousch, La condition de Walters,Ann. Sci. ´Ecole Norm. Sup.,34(2001), 287–311

    2001

  3. [9]

    Bousch, Nouvelle preuve d’un th´ eor` eme de Yuan et Hunt,Bull

    T. Bousch, Nouvelle preuve d’un th´ eor` eme de Yuan et Hunt,Bull. Soc. Math. France,136 (2008), 227–242

    2008

  4. [10]

    J. P. R. Christensen, On sets of Haar measure zero in abelian Polish groups,Israel J. Math.,13(1972), 255–260

    1972

  5. [11]

    J. P. R. Christensen, Measure theoretic zero-sets in infinite dimensional spaces and ap- plications to differentiability of Lipschitz mappings, 2-i` eme Colloq. Analyse Fonctionnel (Bordeaux, 1973),Publications du D´ epartement de Math´ ematiques de Lyon,1973, tome 10, fascicu...

    1973

  6. [12]

    Contreras, Ground states are generically a periodic orbit,Invent

    G. Contreras, Ground states are generically a periodic orbit,Invent. Math.,205(2016), 383–412

    2016

  7. [13]

    Contreras, A

    G. Contreras, A. O. Lopes & Ph. Thieullen, Lyapunov minimizing measures for expanding maps of the circle.Ergodic Theory Dynam. Systems,21(2001), 1379–1409

    2001

  8. [14]

    J. Ding, Z. Li & Y. Zhang, On the prevalence of the periodicity of maximizing measures, Adv. Math.438(2024), 109485

    2024

  9. [15]

    Elekes & D

    M. Elekes & D. Nagy, Haar null and Haar meager sets: a survey and new results,Bull. Lond. Math. Soc.,52(2020), 561–619. TYPICAL UNIQUENESS IN ERGODIC OPTIMIZATION 7

    2020

  10. [16]

    Engelking,General topology, Second edition, Sigma Ser

    R. Engelking,General topology, Second edition, Sigma Ser. Pure Math., 6, Heldermann Verlag, Berlin, 1989

    1989

  11. [17]

    M. K. Fort, Jr., Points of continuity of semi-continuous functions,Publ. Math. Debrecen, 2(1951), 100–102

    1951

  12. [18]

    R. Gao, W. Shen & R. Zhang, Typicality of periodic optimization over an expanding circle map,Preprint, (2025), (arXiv:2501.10949v1)

    2025 arXiv

  13. [19]

    Glasner,Ergodic theory via joinings, Mathematical surveys & monographs 101, Amer

    E. Glasner,Ergodic theory via joinings, Mathematical surveys & monographs 101, Amer. Math. Soc., 2003

    2003

  14. [20]

    Huang, Z

    W. Huang, Z. Lian, X. Ma, L. Xu and Y. Zhang, Ergodic optimization theory for a class of typical maps,J. Eur. Math. Soc., to appear

  15. [21]

    B. R. Hunt & E. Ott, Optimal periodic orbits of chaotic systems,Phys. Rev. Lett.,76 (1996), 2254–2257

    1996

  16. [22]

    B. R. Hunt & E. Ott, Optimal periodic orbits of chaotic systems occur at low period,Phys. Rev. E,54(1996), 328–337

    1996

  17. [23]

    B. R. Hunt, T. Sauer & J. A. Yorke, Prevalence: a translation-invariant ‘almost every’ on infinite-dimensional spaces,Bull. Amer. Math. Soc.,27(1992), 217–238

    1992

  18. [24]

    Prevalence: a translation-invariant ‘almost every’ on infinite-dimensional spaces

    B. R. Hunt, T. Sauer & J. A. Yorke, Prevalence. An addendum to: “Prevalence: a translation-invariant ‘almost every’ on infinite-dimensional spaces”,Bull. Amer. Math. Soc.,28(1993), 306–307

    1993

  19. [25]

    Jenkinson, Frequency locking on the boundary of the barycentre set,Exp

    O. Jenkinson, Frequency locking on the boundary of the barycentre set,Exp. Math.,9 (2000), 309–317

    2000

  20. [27]

    Jenkinson, Ergodic optimization in dynamical systems.Ergodic Theory Dynam

    O. Jenkinson, Ergodic optimization in dynamical systems.Ergodic Theory Dynam. Sys- tems,39(2019), 2593—2618

    2019

  21. [28]

    Jenkinson, X

    O. Jenkinson, X. Li, Y. Liao & Y. Zhang, Ergodic optimization for multi-valued topological dynamical systems,Preprint, (2025), arXiv:2503.18092

    2025 arXiv

  22. [29]

    Krylov & N

    N. Krylov & N. Bogolioubov, La th´ eorie g´ en´ erale de la mesure dans son application ` a l’´ etude des syst` emes dynamiques de la m´ ecanique non lin´ eaire,Ann. Math.,38(1937), 65–113

    1937

  23. [30]

    Z. Li & Y. Zhang, Ground states and periodic orbits for expanding Thurston maps,Math. Ann.,391(2025), 3913–3985

    2025

  24. [31]

    Miller & E

    W. Miller & E. Akin, Invariant measures for set-valued dynamical systems,Trans. Amer. Math. Soc.,351(1999), 1203–1225

    1999

  25. [32]

    W. B. Moors, A note on Fort’s theorem,Topology Appl.,160(2013), 305–308

    2013

  26. [33]

    I. D. Morris, Maximizing measures of generic H¨ older functions have zero entropy,Nonlin- earity,21(2008), 993–1000

    2008

  27. [35]

    Pavlica, On the points of non-differentiability of convex functions,Comment

    D. Pavlica, On the points of non-differentiability of convex functions,Comment. Math. Univ. Carolin.,45(2004), 727–734

    2004

  28. [36]

    Preiss & L

    D. Preiss & L. Zaj ´ ıˇ cek, Directional derivatives of Lipschitz functions,Israel J. Math.,125 (2001), 1–27

    2001

  29. [37]

    Quas & J

    A. Quas & J. Siefken, Ergodic optimization of supercontinuous functions on shift spaces, Ergodic Theory Dynam. Systems,32(2012), 2071–2082

    2012

  30. [38]

    Vesely & L

    L. Vesely & L. Zaj ´ ıˇ cek, On differentiability of convex operators,J. Math. Anal. & Appl. 402(2013), 12–22

    2013

  31. [39]

    Walters,An introduction to ergodic theory, Graduate texts in mathematics 79, Springer, 1981

    P. Walters,An introduction to ergodic theory, Graduate texts in mathematics 79, Springer, 1981. 8 OLIVER JENKINSON, XIAORAN LI, YUEXIN LIAO, AND YIWEI ZHANG

    1981

  32. [40]

    Zaj ´ ıˇ cek, On the differentiation of convex functions in finite and infinite dimensional spaces,Czech

    L. Zaj ´ ıˇ cek, On the differentiation of convex functions in finite and infinite dimensional spaces,Czech. Math. J.,29(1979), 340–348

    1979

  33. [41]

    Zaj ´ ıˇ cek, Onσ-porous sets in abstract spaces,Abstr

    L. Zaj ´ ıˇ cek, Onσ-porous sets in abstract spaces,Abstr. Appl. Anal.,5(2005), 509–534. (O. Jenkinson)School of Mathematical Sciences, Queen Mary University of Lon- don, Mile End Road, London, E1 4NS, UK Email address:o.jenkinson@qmul.ac.uk (X. Li)The Division of Physics, Mat...

    2005

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.