Pith. sign in

REVIEW 4 major objections 5 minor 25 references

Entropy Geometry and Normalized Means on Infinite-Dimensional Hamiltonian Manifolds

T0 review · 4 major / 5 minor · reviewed 2026-08-03 · deepseek-v4-flash

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

Pith's one-line read Replacing probability measures by normalized means, the paper extends equilibrium statistical mechanics to infinite-dimensional Hamiltonian systems and proves existence and uniqueness of exponential-family equilibrium states.

desk verdict A careful conditional framework for Souriau thermodynamics via normalized means, but its central theorem has no verified infinite-dimensional instance—the Gaussian examples make the admissibility domain empty for β>0. read the letter →

arxiv 2607.28660 v1 pith:U3MEX5HP submitted 2026-07-21 math-ph math.DGmath.FAmath.MP

Jean-Pierre Magnot This is my paper · ORCID
classification math-phmath.DGmath.FAmath.MP MSC 37K0553D2082B1046L5558D0537L5082C3146A03
keywords normalizedmeansequilibriumstatisticalmechanicsinfinite-dimensionalHamiltoniansystemsweaksymplecticFréchetmanifoldsexponentialfamiliesrelativeentropygeometryhydrodynamicmodels
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's central claim is that equilibrium statistical mechanics can be carried over to infinite-dimensional Hamiltonian systems even when no probability measure exists to play the role of a reference state. The device is the normalized mean, a positive normalized linear functional on an algebra of observables that generalizes probability measures and invariant means. Given such a reference, the paper defines relative entropy and free energy, and shows that the free energy has a unique minimizer given by an explicit exponential tilt of the reference. Under a local exponential regularity condition, the log-partition function is smooth and convex, its Hessian is the covariance of the observables, and its Legendre–Fenchel dual is a concave entropy. If correct, this gives a rigorous equilibrium formalism for infinite-dimensional Hamiltonian PDEs and hydrodynamic models.

What carries the argument

The central mechanism is the exponential tilt of a normalized mean: for an admissible potential F, the tilted mean m_F(f) = m0(f e^F)/m0(e^F). The argument rests on a change-of-reference identity H(n∥m_F) = H(n∥m0) − n(F) + log m0(e^F), which is valid precisely when the admissible classes satisfy translation compatibility, i.e. E(m_F)+F = E(m_0). This identity converts the variational problem into a relative-entropy minimization and yields the unique minimizer; a local exponential regularity condition then justifies differentiating the partition function and identifies the Hessian with the covariance form.

What would settle it

Exhibit an infinite-dimensional Hamiltonian system where, after tilting the reference mean by the Hamiltonian, the admissible potentials for the tilted mean are not exactly the original admissible potentials shifted by that Hamiltonian (translation compatibility fails), or where the finite-dimensional cut-off partition functions converge to zero; either would invalidate the uniqueness theorem.

Watch

Extended reading notes

Core claim

On a weak symplectic Fréchet manifold equipped with a normalized mean m0, the paper defines the free energy G_{β,λ}(n) = H(n∥m0) + β n(H) + Σ λ_a n(Φ_a) and proves (Theorem 2.10) that, under translation compatibility and separation hypotheses, it has the unique minimizer n*_{β,λ}(f) = m0(f e^{−βH−Σλ_aΦ_a}) / m0(e^{−βH−Σλ_aΦ_a}). The logarithmic partition function ψ = log m0(e^{−βH−Σλ_aΦ_a}) is then smooth, with Hessian equal to the covariance form of the Hamiltonian and constraints; the Legendre–Fenchel dual of ψ is a concave entropy on extensive variables. The construction is presented as the infinite-dimensional analogue of the classical geometric thermodynamics of coadjoint orbits, with n

Load-bearing premise

The construction collapses if the class of admissible exponential weights does not shift exactly by the tilt potential after reweighting (translation compatibility); the paper assumes this in every infinite-dimensional example without verifying it.

Editorial extensions

If this is right

  • Equilibrium states exist and are unique for infinite-dimensional Hamiltonian systems without assuming a σ-additive reference measure, whenever the admissibility and translation-compatibility conditions hold.
  • The log-partition function is smooth and convex, and its Hessian is the covariance form; this yields strict convexity modulo thermodynamically null directions and a metric structure on the reduced parameter space.
  • The Legendre–Fenchel dual of the log-partition is a concave entropy that agrees with the constrained variational entropy on the equilibrium image, so intensive/extensive duality survives in infinite dimension.
  • Equilibrium means are stationary under Hamiltonian flows that preserve the reference mean and the constraints, and satisfy a classical Poisson–KMS identity when the reference mean is Poisson invariant.
  • The framework covers geodesic equations on current groups and diffeomorphism groups, including hydrodynamic and field-theoretic examples, with exponential-family equilibria given by explicit tilts.

Reading between the lines

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

  • If translation compatibility can be verified for concrete Hamiltonian PDEs, the normalized-mean equilibrium states would supply rigorous stationary statistical states for systems where no invariant probability measure is currently known.
  • The covariance Hessian defines a metric on the parameter space, suggesting an information-geometric notion of distance between equilibria that could be used to study thermodynamic processes in infinite dimension.
  • The classical Poisson–KMS identity could serve as a selection criterion identifying equilibrium states among all stationary normalized means, paralleling the role of KMS states in algebraic quantum statistical mechanics.
  • The appendix's renormalized determinants suggest a concrete recipe for constructing reference means: relative normalization or second-order determinant renormalization of cut-off partition functions may yield admissible normalized means even when the covariance perturbation is only Hilbert–Schmidt.
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

4 major / 5 minor

Summary. The paper develops a geometric-analytic framework for equilibrium statistical mechanics on infinite-dimensional Hamiltonian systems, replacing σ-additive probability measures with normalized means (positive normalized linear functionals on a vector lattice algebra L). It defines the relative entropy H(n∥m0) via a Donsker–Varadhan-type supremum over exponentially admissible potentials, introduces exponential tilts, and, under admissibility, translation compatibility, and separation hypotheses, proves a Gibbs variational principle with explicit exponential-family minimizer (Theorem 2.10). It also proves equivariance properties (Theorem 1.17), stationarity under Hamiltonian flows preserving the reference mean and constraints (Theorem 2.12), a classical Poisson–KMS identity (Theorem 2.17), smoothness/convexity of the log-partition functional under local exponential regularity (Proposition 1.20), and a Legendre–Fenchel dual entropy (Propositions 1.27–1.28). Applications are sketched for H^s geodesic equations on current groups and diffeomorphism groups, including Camassa–Holm/EPDiff and 2D Euler relations, with the repeated caveat that the abstract hypotheses are assumed to hold.

Significance. If the framework were instantiated in a genuinely infinite-dimensional, non-σ-additive setting, it would extend Souriau's Gibbs formalism beyond Gaussian or σ-additive reference measures and provide a common language for equilibrium states of Hamiltonian PDEs. The algebraic core of the paper is clean: the change-of-reference identity, the covariance Hessian computation, and the Poisson–KMS identity are proved by explicit finite-order estimates, with no uncontrolled interchange of limits. The paper is also honest in flagging its conditional assumptions. The obstacle is that no such infinite-dimensional instantiation is actually provided: the only explicit Gaussian computations, in Appendix A, show that the unrenormalized partition function for the energy tilt vanishes in the infinite-dimensional limit, so the central admissibility hypothesis 0<m0(e^{-βH})<∞ fails. Consequently the main theorem is not known to have any non-σ-additive infinite-dimensional instance, and the claimed existence/uniqueness result is at present a conditional algebraic statement.

major comments (4)
  1. [Appendix A (Camassa–Holm/EPDiff Gaussian computation)] The only explicit infinite-dimensional Gaussian calculation violates the central admissibility condition. The cut-off log-partition for the energy tilt is log Z_N(β,0)=-(2N+1)/2 log(1+β), so Z_N(β,0)→0 for every β>0; hence m0(e^{-βH})=0 and F_{β,0}∉E(m0). Thus D in Definition 2.3 is empty for the energy tilt in this model, and Theorem 2.10 has no instance there. The appendix's statement that 'only the resulting relative or renormalized free energy... enter the corresponding renormalized variational principle' acknowledges the difficulty, but no renormalized variational principle is stated or proved. A revision must either supply a genuinely admissible example or develop a renormalized version of Theorem 2.10.
  2. [Definition 2.6 / Theorem 2.10 / Sections 3–4] Translation compatibility E(m_F)+F=E(m_0) is never verified in any example. The applications repeatedly say 'whenever F belongs to E(m0) and satisfies the translation-compatibility and separation hypotheses' without checking them. Since this condition is used in the change-of-reference identity (Theorem 2.7) to conclude that the exponential tilt is the global minimizer, the main theorem remains conditional in every concrete case. At least one nontrivial class of means and potentials satisfying translation compatibility is needed, or the theorem should be presented purely as an implication with no claimed instances.
  3. [Proposition 1.13 / Theorem 2.10] The separation and local-perturbation hypotheses are ad hoc and unverified. The proof requires a subspace T separating normalized means and differentiability of t↦log m0(e^{t h}) at t=0 with derivative m0(h). For finitely additive means this differentiation is not automatic, and the manuscript does not give a criterion by which to verify it. Since uniqueness in Proposition 2.9 and Theorem 2.10 relies on H(n∥m*)=0 ⇒ n=m*, the uniqueness assertion is also conditional on an unverified analytic assumption.
  4. [Definition 2.3 / Theorem 2.10 / Abstract] The admissible domain D is defined by requiring 0<m0(e^{F_{β,λ}})<∞, so part of the 'existence' conclusion is assumed rather than derived. The theorem constructs the tilt from an assumed finite positive partition function; it does not prove that such a partition function exists for any given system. The paper is often careful about this, but the abstract and the statement of Theorem 2.10 say 'existence and uniqueness' without sufficiently emphasizing that existence is contingent on admissibility. The central claim should be restated as a conditional: if F_{β,λ}∈E(m0) and translation compatibility holds, then the tilt is the unique minimizer.
minor comments (5)
  1. [General typography] Several LaTeX artifacts remain: 'F r´echet', 'smallers', 'ordersof', and missing spaces in the references. A careful proofreading pass is needed.
  2. [Definition 1.19 vs Definition 2.3] Definition 1.19 refers to the parameter domain D before D is formally defined in Definition 2.3; add a forward pointer or move the definition earlier.
  3. [Proof of Proposition 1.13] The proof says 'Since the admissible potentials separate normalized means on L', but the separation assumption is stated only for a subspace T⊂L. Clarify that T⊂E(m0), or extend the separation assumption explicitly to all admissible potentials.
  4. [Appendix D, normalizability summary] The sentence 'all β>0 are admissible at that level' could mislead, since the infinite-dimensional limit requires relative normalization. Please distinguish finite-cut-off admissibility from infinite-dimensional admissibility in D.
  5. [References] Reference [19] appears incomplete ('A classification of coadjoint orbits carrying Gibbs ensembles' lacks a period/journal) and [14] is a self-citation to a 2026 Annals of Physics article; verify bibliographic data.

Circularity Check

0 steps flagged · score 0.0 of 10

Conditional infinite-dimensional Gibbs formalism; no circular step identified

full rationale

The derivation chain is conditional but self-contained. The lower bound G_{β,λ}(n) ≥ −log Z(β,λ) is literally the definition of relative entropy as a supremum; the candidate n*_{β,λ} is by definition the exponential tilt of m0 by F_{β,λ}. Equality at n* and uniqueness do not follow from that definition alone: they use the explicitly stated translation-compatibility hypothesis E(m_F)+F=E(m0), which is proved to imply the change-of-reference identity (Theorem 2.7), and the explicitly stated separation/local-perturbation hypothesis of Proposition 1.13. Smoothness, convexity, covariance-Hessian, and Legendre duality are derived from local exponential regularity and the covariance formula, not assumed as conclusions. No parameter is fitted to data, and the infinite-dimensional examples are explicitly conditional ('Whenever F belongs to E(m0) and satisfies the translation-compatibility and separation hypotheses...'), so they do not smuggle the theorem into its assumptions. The author's self-citations, [13] for normalized means from cut-offs and [14] for the programme, are not load-bearing: the abstract definition of normalized means and the compactness/positivity arguments stand independently. The appendix honestly records that the unrenormalized Gaussian partition function vanishes as N→∞ and that only a relative or renormalized free energy would enter a renormalized variational principle; since no such renormalized theorem is proved, this is a gap in the examples, not a circular derivation. I therefore find no circular step that reduces a claimed result to its own inputs.

Assumptions & free parameters 0 free parameters · 7 assumptions · 1 invented entities

The framework's central theorems rest on a stack of assumptions (A1–A4 plus Poisson invariance and flow preservation) that are stated explicitly but not verified in the infinite-dimensional examples. The paper's own text repeatedly conditions the applications on these hypotheses (Section 3, Section 4.7). No numerical free parameters are fitted to data; the reference mean m0 is a chosen input.

assumptions (7)
  • domain assumption There exists a unital vector lattice algebra L of observables and a normalized mean m0 on L such that e^F ∈ L and 0 < m0(e^F) < ∞ for admissible F.
    Definition 1.10 (exponentially admissible potentials) and Definition 2.3 (admissible thermodynamic domain) presuppose such objects; the paper does not prove their existence in concrete infinite-dimensional settings.
  • ad hoc to paper Translation compatibility: for the tilt F, E(m_F)+F = E(m_0) (Definition 2.6).
    Introduced to make the change-of-reference identity (Theorem 2.7) and thus the Gibbs variational principle (Theorem 2.10) hold. The paper states it as an assumption and never verifies it in the Map(M,G) or Diff(M) examples.
  • ad hoc to paper Separation: the subspace T ⊂ L separates normalized means and the derivative d/dt log m0(e^{th}) at t=0 equals m0(h) (Proposition 1.13).
    Required for uniqueness in Theorem 2.10 and Proposition 2.9; it is an explicit assumption on the observable algebra.
  • domain assumption Local exponential regularity: for each θ ∈ D there is ε>0 with m0(exp(F_θ + ε Σ|O_j|)) < ∞ (Definition 1.19).
    Used to prove smoothness of the partition function (Proposition 1.20) and the Hessian formula (Proposition 1.23). It is assumed to hold on D.
  • domain assumption Poisson invariance of the reference mean: m0({f,g}) = 0 for f,g ∈ P (Definition 2.15).
    Necessary for the classical Poisson-KMS identity (Theorem 2.17); the paper notes this is the algebraic counterpart of integration by parts and does not prove it for the examples.
  • domain assumption Weak symplectic Fréchet manifold structure: smooth closed weak symplectic form ω and well-defined Hamiltonian vector fields (Section 1.1).
    The entire setting; such structures are not automatic on Fréchet manifolds.
  • domain assumption The Hamiltonian flow Φ_t^H is global and preserves m0 and L, and H and constraints are invariant under it (Theorem 2.12).
    Stationarity of equilibrium means depends on this; global existence for the H^s geodesic flows is not proven in the paper.
invented entities (1)
  • Normalized means on vector lattice algebras L
    purpose: Replace σ-additive probability measures to define entropy, partition functions, and equilibrium states on infinite-dimensional phase spaces.
    Defined in Definition 1.2; these are formal algebraic objects whose existence and relevance are taken as axioms. No external testable consequence is provided outside the paper's own theorems.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Entropy Geometry and Normalized Means on Infinite-Dimensional Hamiltonian Manifolds." pith.science (2026). https://pith.science/paper/U3MEX5HP

@misc{pith2026260728660,
  author       = {Pith},
  title        = {Pith review of: Entropy Geometry and Normalized Means on Infinite-Dimensional Hamiltonian Manifolds},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/U3MEX5HP}},
  note         = {Machine review of arXiv:2607.28660}
}
abstract

We propose a geometric--analytic framework for equilibrium statistical mechanics on infinite-dimensional Hamiltonian systems. In situations where no suitable $\sigma$-additive invariant measure is available, we use \emph{normalized means}, which generalize probability measures and normalized integrals. This construction yields entropy and free-energy functionals on weak symplectic Fr\'echet manifolds and gives existence and uniqueness of exponential-family equilibrium states under explicit admissibility and separation assumptions. These states are stationary under Hamiltonian flows preserving both the reference mean and the equilibrium weight, and satisfy a classical Poisson--KMS identity when the reference mean is Poisson invariant. Under a local exponential regularity assumption, the logarithmic partition functional is smooth and convex, with Hessian given by the covariance form. It is strictly convex modulo thermodynamically null directions and, through Legendre--Fenchel duality, induces a concave entropy on the domain of extensive variables. We illustrate the framework with $H^s$-geodesic equations on current groups $\operatorname{Map}(M,G)$ and diffeomorphism groups $\operatorname{Diff}(M)$, including hydrodynamic and field-theoretic examples.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

25 extracted references

  1. [1]

    Abraham and J

    R. Abraham and J. E. Marsden,Foundations of Mechanics, 2nd ed., Benjamin/Cummings Publishing Co., Reading, MA, 1978

    1978

  2. [2]

    Ammari and V

    Z. Ammari and V. Sohinger,Gibbs measures as unique KMS equilibrium states of nonlinear Hamil- tonian PDEs, Rev. Mat. Iberoam.39(2023), no. 6, 2035–2073

    2023

  3. [3]

    Barbaresco,Information geometry of Souriau Lie groups thermodynamics and Koszul–Vey equa- tions, Entropy11(2009), 329–364

    F. Barbaresco,Information geometry of Souriau Lie groups thermodynamics and Koszul–Vey equa- tions, Entropy11(2009), 329–364

    2009

  4. [4]

    Barbaresco, Jean-Marie Souriau’s Symplectic Foliation Model of Sadi Carnot’s Thermodynamics

    F. Barbaresco, Jean-Marie Souriau’s Symplectic Foliation Model of Sadi Carnot’s Thermodynamics. Entropy 27 (2025), 509

    2025

  5. [5]

    Bourgain,Periodic nonlinear Schr¨ odinger equation and invariant measures, Comm

    J. Bourgain,Periodic nonlinear Schr¨ odinger equation and invariant measures, Comm. Math. Phys. 166(1994), no. 1, 1–26

    1994

  6. [6]

    Bratteli and D

    O. Bratteli and D. W. Robinson,Operator Algebras and Quantum Statistical Mechanics 2: Equilibrium States. Models in Quantum Statistical Mechanics, 2nd ed., Texts and Monographs in Physics, Springer, Berlin, 1997

    1997

  7. [7]

    M. M. Day,Amenable semigroups, Illinois J. Math.1(1957), 509–544

    1957

  8. [8]

    M. D. Donsker and S. R. S. Varadhan,Asymptotic evaluation of certain Markov process expectations for large time, Comm. Pure Appl. Math.28(1975), 1–47

    1975

Show all 25 references
  1. [9]

    D. G. Ebin and J. Marsden,Groups of diffeomorphisms and the motion of an incompressible fluid, Ann. of Math. (2)92(1970), 102–163

    1970

  2. [10]

    R. S. Hamilton,The inverse function theorem of Nash and Moser, Bull. Amer. Math. Soc. (N.S.)7 (1982), no. 1, 65–222

    1982

  3. [11]

    Iglesias,Sym´ etries et moment, Hermann, Paris, 2000

    P. Iglesias,Sym´ etries et moment, Hermann, Paris, 2000

    2000

  4. [12]

    Kriegl and P

    A. Kriegl and P. W. Michor,The Convenient Setting of Global Analysis, Mathematical Surveys and Monographs, vol. 53, American Mathematical Society, Providence, RI, 1997

    1997

  5. [13]

    Magnot,The mean value for infinite volume measures, infinite products, and heuristic infinite dimensional Lebesgue measures, J

    J.-P. Magnot,The mean value for infinite volume measures, infinite products, and heuristic infinite dimensional Lebesgue measures, J. Math.2017(2017), Article ID 9853672, 14 p

    2017

  6. [14]

    Magnot, Geometric thermodynamics and entropy functionals for infinite-dimensional Hamilton- ian systems

    J.-P. Magnot, Geometric thermodynamics and entropy functionals for infinite-dimensional Hamilton- ian systems. Annals of Physics,493(2026), pp.170604

    2026

  7. [15]

    Marle,From tools in symplectic and Poisson geometry to J.-M

    C.-M. Marle,From tools in symplectic and Poisson geometry to J.-M. Souriau’s theories of statistical mechanics and thermodynamics, Entropy18(2016), no. 10, 370

    2016

  8. [16]

    J. E. Marsden and T. S. Ratiu,Introduction to Mechanics and Symmetry, 2nd ed., Texts in Applied Mathematics, vol. 17, Springer, New York, 1999

    1999

  9. [17]

    Robert and J

    R. Robert and J. Sommeria,Statistical equilibrium states for two-dimensional flows, J. Fluid Mech. 229(1991), 291–310

    1991

  10. [18]

    Neeb,Towards a Lie theory of locally convex groups, Japanese J

    K.-H. Neeb,Towards a Lie theory of locally convex groups, Japanese J. Math.1(2006), no. 2, 291–468

    2006

  11. [19]

    K. H. Neeb A classification of coadjoint orbits carrying Gibbs ensemblesarXiv:2601.04934

  12. [20]

    Oh and N

    T. Oh and N. Tzvetkov,Quasi-invariant Gaussian measures for the cubic fourth order nonlinear Schr¨ odinger equation, Probab. Theory Related Fields169(2017), no. 3–4, 1121–1168

    2017

  13. [21]

    Omori,Infinite-Dimensional Lie Groups, Translations of Mathematical Monographs, vol

    H. Omori,Infinite-Dimensional Lie Groups, Translations of Mathematical Monographs, vol. 158, American Mathematical Society, Providence, RI, 1997

    1997

  14. [22]

    A. L. T. Paterson,Amenability, Mathematical Surveys and Monographs, vol. 29, American Mathe- matical Society, Providence, RI, 1988

    1988

  15. [23]

    Souriau,Structure des syst` emes dynamiques, Dunod, Paris, 1970

    J.-M. Souriau,Structure des syst` emes dynamiques, Dunod, Paris, 1970

    1970

  16. [24]

    Tzvetkov,Invariant measures for the defocusing nonlinear Schr¨ odinger equation, Ann

    N. Tzvetkov,Invariant measures for the defocusing nonlinear Schr¨ odinger equation, Ann. Inst. Fourier (Grenoble)58(2008), no. 7, 2543–2604

    2008

  17. [25]

    von Neumann,Zur allgemeinen Theorie des Masses, Fund

    J. von Neumann,Zur allgemeinen Theorie des Masses, Fund. Math.13(1929), 73–116. SFR MATHSTIC, LAREMA, Universit ´e d’Angers, 2 Bd Lavoisier, 49045 Angers cedex 1, France; Lyc ´ee Jeanne d’Arc, 40 avenue de Grande Bretagne, 63000 Clermont-Ferrand, France; Lepage Research Instit...

    1929

Pith tools

Reviewed August 3, 2026 · model on record in the stance chip above.