Global dynamical stability for KMS-states

Heide Narnhofer

REVIEW 3 major objections 4 minor 20 references

Global dynamical stability for KMS-states

T0 review · 3 major / 4 minor · reviewed 2026-08-05 · deepseek-v4-flash

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

Pith's one-line read This paper argues that, under a space-time multiclustering assumption, the only time-invariant states of strongly interacting lattice systems that are stable under a slow global perturbation of the dynamics are KMS states at some inverse te

desk verdict A genuinely new fixed-point route to KMS, but the load-bearing spectral step in (11) is asserted rather than derived, so it is a research program with a solid target, not a theorem. read the letter →

arxiv 2508.10728 v1 pith:TTHAOKQC submitted 2025-08-14 math-ph math.MP

Heide Narnhofer This is my paper · ORCID

classification math-phmath.MP MSC 46L5582B1046L30

keywords KMSstatesBoltzmann-HugenholzevolutiondynamicalstabilityquantumlatticesystemsLindbladmodularoperatormulticlusteringthermalequilibrium

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 seeks to show that the KMS condition, the standard characterization of thermal equilibrium in quantum statistical mechanics, follows from a single dynamical stability requirement: if a time-invariant state of a strongly interacting lattice system is stable under a global, slowly varying perturbation of the dynamics, then in the appropriate scaling limit it must be a KMS state for some inverse temperature β. The author generalizes the Hugenholz-Boltzmann evolution from weakly interacting or quasifree systems to strongly interacting lattice systems, and proves that any fixed point of this evolution is KMS, provided the state satisfies a space-time multiclustering assumption. The significance is that it reduces the equilibrium condition to a physically motivated stability demand without invoking passivity or a long list of perturbations. The price is a clustering assumption that the paper itself concedes is plausible but not proven for generic interacting systems.

What carries the argument

The central mechanism is the modular-spectrum cancellation argument applied to the Boltzmann-Hugenholz evolution. The evolution is generated by $W = \int_0^\infty dt\, \alpha_0(t) H'$, an operator on the GNS space, and one studies the vanishing of the quadratic form $\langle\Omega|WW A + AW^*W^* - 2WAW^*|\Omega\rangle$ characterizing invariant states. Since the modular operator $\Delta$ and the time evolution commute, the expression decomposes in the joint spectral representation ($\lambda$ for modular variables, $\mu$ for time); the integrals over the spectral variables have poles at $\mu'=0$ and $\mu'=\mu$, and these singularities cancel only if the measure contains a factor $\delta(\lambd

What would settle it

Check the tracial state of a finite-range interacting Fermi lattice model: if it satisfies the space-time multiclustering condition (6) and the limiting conditions (4)-(5), then the theorem's conclusion that every such invariant state is KMS would be violated, since the tracial state is not KMS for any finite β.

Watch

Extended reading notes

Core claim

The paper shows that time-invariant states of strongly interacting Fermi lattice systems that are fixed points of the generalized Boltzmann-Hugenholz evolution are KMS states. Starting from a Hamiltonian $H_\lambda = K + V + \lambda H'$ and taking the scaled limit $\tau = \lim_{t\to\infty,\, \lambda^2 t = \tau} \omega(\alpha_0(-t)\alpha_\lambda(t) A)$, the first-order terms vanish by time clustering and the second-order terms satisfy a Lindblad-type differential equation. Invariance under this evolution is expressed in the GNS representation as $\langle\Omega|WW A + AW^*W^* - 2WAW^*|\Omega\rangle = 0$, with $W = \int_0^\infty dt\, \alpha_0(t) H'$. Using that the modular operator $\Delta$ com

Load-bearing premise

The proof assumes that every time-invariant state of a strongly interacting lattice system satisfies space-time multiclustering: correlations factorize when events are far apart in space or time; if a non-KMS invariant state fails this clustering, the theorem's 'only' statement does not cover it.

Editorial extensions

If this is right

If correct, any stationary state of a quantum lattice model that is stable under slow global perturbations in the $\lambda^2 t=\tau$ scaling is automatically a thermal equilibrium state, with no need to enumerate many perturbations.
The Boltzmann-Hugenholz flow increases entropy density because it is of Lindblad type, so the approach to equilibrium in this scaling is consistent with the second law at the level of local density matrices.
The result holds with only one type of perturbation $H'$, whereas earlier dynamical-stability arguments required sufficiently many perturbation directions.
The theorem provides a possible dynamical derivation of the KMS condition that does not pass through the passivity assumption.
The proof method yields a spectral criterion, the presence of $\delta(\lambda-\beta\mu)$, that could be checked in models where clustering and commuting automorphisms are under control.

Reading between the lines

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

Editorial inference: If the multiclustering assumption (eq. (6)) is the true obstruction, then the theorem's practical value hinges on proving decay of correlations for interacting models; the paper's own examples suggest this may fail for tracial states, so the theorem may only apply to states that already look thermal in their clustering.
Editorial inference: The modular-spectrum argument is model-independent, so it may extend beyond lattice Fermi systems to any C*-dynamical system where a Lindblad-type global perturbation can be constructed and clustering holds; a testable extension would be continuous quantum systems or Bose lattices.
Editorial inference: The entropy-density increase under the Boltzmann-Hugenholz flow suggests a possible bound on thermalization times: relaxation to the KMS state in this scaling is set by the inverse square of the perturbation, and deviations from KMS should be accompanied by positive entropy production of order $\lambda^2$.
Editorial inference: The paper leaves open whether the limit (5) exists for non-KMS initial states; if it does and multiclustering holds, the theorem would imply that the long-time limit under $H_\lambda$ is always thermal, effectively proving a form of thermalization for these lattice systems.

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.

Desk Editor's Note

read the letter

Short version: this is a serious research announcement, not a proof, and its own abstract says so. The new thing is the way the author pushes Hugenholz's Boltzmann-type argument past quasifree dynamics: instead of controlling an explicit two-point function, she uses the modular operator in the GNS representation of an arbitrary time-invariant state and tries to force the KMS temperature out of a fixed-point identity. That is a genuinely different route, and it is a plausible target.

Credit where due: the construction of commuting automorphisms in Section IV (eqs. (15)–(17)) gives a concrete class of invariant states to test the mechanism on, and the paper is transparent about which premises are unproven. The closing discussion explicitly concedes that Lieb-Robinson bounds are 'promising but not sufficient.' That honesty makes the note useful to people working on the problem.

The soft spot is load-bearing. Equation (11) is the only place the inverse temperature β appears, and the text jumps from a formal integral to 'the vector vanishes only if the integral contains δ(λ−βμ)' in two sentences. There is no derivation of the 'only if', the singular integrals are not controlled (the iε prescriptions matter and are left implicit), and the W/W* adjoint placement has typos in (9)–(12), so the algebra cannot be checked. The stress-test note lands: if a non-KMS state had spectral support or phases satisfying a different cancellation, the argument would not rule it out. That is not a minor gap; it is the main step.

The other weakness is the premise. Space-time multiclustering (6) is assumed for interacting systems, and the paper notes that cluster properties can fail in interacting cases ([5], [6]). So even a correct (11) would only apply to states satisfying a condition that may rule out the very non-KMS states the conclusion wants to exclude. The author acknowledges this, but it means the 'only' in the conclusion is doubly conditional.

The citations look honest; the self-citations are to her earlier work on cluster properties and interacting systems, which is relevant. Who should read it: people invested in the Haag–Kastler–Trych-Pohlmeyer program and the passivity/stability route to KMS. It gives a clear research program — complete the spectral argument and prove (6) for a concrete model — and a completed version would be a substantial theorem. As written, it deserves a serious referee rather than a desk reject, because the target is important and the author is candid about the gaps. I would not cite it as a proof, but I would bring it to our reading group as an honest, compact research announcement.

Referee Report

3 major / 4 minor

Summary. The paper proposes to generalize the Hugenholz-Boltzmann evolution to strongly interacting lattice systems and claims that, under suitable clustering and limit assumptions, any state invariant under this effective evolution must satisfy the KMS condition. The argument proceeds by deriving a fixed-point equation (Eq. (9)) from a second-order perturbative expansion, passing to a GNS spectral representation (Eqs. (10)-(11)), and asserting that the resulting vector can vanish only if the modular spectrum contains a delta function δ(λ − βμ), forcing a KMS state. The paper closes with a discussion of the plausibility of the clustering assumptions and an entropy-monotonicity argument for the associated Lindblad form.

Significance. If the central claim were rigorously established, it would supply a dynamical stability characterization of KMS states for interacting quantum lattice systems, extending the classical and quasifree results of Hugenholz to strongly interacting systems with a relatively mild set of hypotheses. The paper also makes a useful conceptual connection to Lindblad evolution and entropy increase. However, the manuscript as written does not provide a proof of the decisive spectral argument, and it explicitly concedes that the key clustering assumptions are not verified for interacting systems. The result is therefore currently a conjecture supported by heuristic calculations rather than a theorem.

major comments (3)

[Section II, Eq. (11)] The load-bearing 'only if' step is asserted, not proved. After writing the fixed-point condition in the joint spectral representation, the paper states that the vector vanishes only if the integral contains δ(λ − βμ), because the singularity at μ′=0 gives λ′=0 and the singularity at μ′=μ cancels e^{−μ}=e^{−λ}. This is a two-sentence argument. No derivation is given that no other spectral support or phase relation can cancel the singularities. Since Eq. (11) is the only place where the inverse temperature β enters, the entire conclusion hinges on this unproved assertion. Without a rigorous treatment, the possibility remains that non-KMS states satisfy Eq. (10) via a different spectral identity.
[Section II, Eqs. (7)-(9)] The reduction from the perturbative expansion to the Lindblad-type fixed-point equation is not justified. The paper claims that factorized terms cancel and only order-λ^{2n} terms survive, but this relies on the multiclustering assumption (6) and on the summability of the resulting series. No proof is supplied that the double integral in Eq. (7) reduces to the simple W-integral form in Eq. (9). Moreover, W = ∫_0^∞ H′_t dt is an unbounded half-line integral, and the manipulations involving the modular operator and the density of A*|Ω⟩ require domain and adjoint considerations that are not addressed. The repeated ambiguity between W and W* in Eqs. (9)-(12) makes the algebra impossible to check.
[Section II and Section IV] The theorem's applicability is conditional on assumptions (4)-(6), but the paper itself provides evidence that these assumptions may fail for the very states the theorem aims to exclude. Near the end of Section IV, it is conceded that Lieb-Robinson-type bounds are 'promising but not sufficient' and that refs. [5],[6] show cluster properties can fail. If Eq. (6) fails for non-KMS time-invariant states, then the argument never applies to them, and the claimed 'only KMS states are stable' conclusion becomes vacuous. The paper needs either a proof of the clustering assumptions in a nontrivial model or a clear statement that the result is conditional and the 'only' applies only within the assumed class.

minor comments (4)

[Throughout] There are multiple typos and notational inconsistencies involving W and W*, e.g., in Eqs. (9), (10), and (12). These should be corrected and the adjoint placements made unambiguous.
[Eq. (14)] The entropy formula appears to have a missing parenthesis and the summation indices are unclear. Please rewrite the expression with explicit domains of summation.
[Abstract and Introduction] The name 'Hugenholz' is spelled inconsistently with the standard 'Hugenholtz' in the references. Please unify the spelling.
[Section IV] The discussion of the commuting automorphism (16) is heuristic. If it is intended as evidence for the generality of the assumptions, a precise statement of when such an automorphism exists would be helpful.

Circularity Check

0 steps flagged · score 2.0 of 10

No circular derivation: KMS condition is inferred via modular spectrum, not assumed. Self-citations are background only.

full rationale

The claimed derivation chain is not circular. The fixed-point condition (9) is derived from the Boltzmann-Hugenholz equation (7) under explicit multiclustering assumptions; the KMS condition is not among the inputs. The modular operator Δ is defined from the arbitrary time-invariant state, and the conclusion Δ = e^{-βH} is obtained via a spectral cancellation argument. Whether that argument is rigorous is a correctness question, not circularity: Eq. (11) is asserted, but the paper does not define δ(λ−βμ) to be the KMS condition and then plug it back in; it claims to derive it. The assumptions in Section II (existence of limits, multiclustering) are not equivalent to KMS, and the paper explicitly acknowledges (Section IV) that Lieb-Robinson bounds are insufficient and that cluster properties can fail [5],[6]. The self-citations [5],[6],[7],[14],[15],[20] appear in plausibility and background statements and are not used to substitute for the core derivation. Thus no load-bearing reduction to inputs is exhibited. Score set to 2 only because several self-citations are present, but they are not load-bearing.

Assumptions & free parameters 2 free parameters · 7 assumptions · 0 invented entities

The paper pulls three items from the shelf for free: (i) the Hugenholz quasifree framework [11], (ii) standard modular theory [15], and (iii) the external clustering and Lieb-Robinson bounds [16],[17],[19],[20]. Its own added axioms are the multiclustering of invariant states (6), the convergence of the scaled limits (4)-(5), and a formal regularization of the non-summable perturbation expansion (7)-(8). No genuinely new entities are introduced. Net contribution beyond the ledger is the spectral cancellation claim (11), asserting that fixed points are KMS, which is where the paper is least explicit.

free parameters (2)

gamma in commuting automorphism (16) = not a number; free positive constant
Introduced in Section IV to define the commuting evolution with derivative gamma*deltaK + gamma^-1*deltaV; used to construct candidate time-invariant states; any value yields a valid commuting automorphism and no physical selection rule is given.
choice of perturbation H' = not a number; chosen freely subject to being space-translation invariant and short range
Section II: 'we have some freedom and can choose it e.g. quasifree'. The theorem is claimed for 'only one type of perturbation', but no uniqueness or genericity statement is proven.

assumptions (7)

domain assumption Fermi lattice system with gauge-invariant Hamiltonian H = K + V + lambda H', K quadratic and V quartic; evolutions commute with space translations
Section II, first paragraph; the whole framework uses [2] and [11].
ad hoc to paper For the states under study the limits (4) and (5) exist, the limit is alpha0-invariant and space-translation invariant and independent of alpha0(t') for finite t', and the state is multi-clustering in t (eq. (6))
Section II, text below (5)-(6). This is the paper's key working hypothesis; Section IV concedes it is not established for interacting systems, citing [5],[6] and calling [17] 'promising but not sufficient'.
ad hoc to paper The formal perturbative expansion (7)-(8) defines the Hugenholz evolution: factorized terms cluster-cancel and only order-lambda^2n terms survive; the double integral in (7) reduces to the square of W = integral_0^inf H'_t dt
Section II, equations (7)-(9). The paper states the terms are not summable as operators and that multi-clustering makes most of them factorize; no proof of the summation and cancellation is given.
domain assumption The modular operator Delta of the invariant state commutes with the Hamiltonian H, and the joint spectral representation (mu, lambda) of H and log Delta is available
Section II after (10), citing [15]; standard modular theory for faithful invariant states.
ad hoc to paper W = integral_0^inf H'_t dt is a well-defined (unbounded) self-adjoint operator such that the GNS computations (9)-(11) are legitimate
The integral over all positive times of an interacting H' is never proved to converge in any sense; the paper works directly with formal operator distributions, as seen in the 1/(mu - mu' - i epsilon) kernels in (11).
domain assumption For the entropy statement, the invariant state can be replaced by omega_X^c tensor omega_X and W by W_X^c tensor W_X up to surface terms, and the bulk entropy-density change dominates
Section III after (14): 'The two states differ by a surface term... In the thermodynamic limit the increase is dominating.' No quantitative surface-bulk control is given.
standard math External bounds are valid as invoked: exponential clustering (18) of Matsui [16] for KMS states on 1D lattices, Lieb-Robinson bounds [17], the Radin vacuum bound [19], and the chaotic-delocalization result [20]
Section IV uses these as external benchmarks; the paper does not reprove them.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Global dynamical stability for KMS-states." pith.science (2026). https://pith.science/paper/TTHAOKQC

@misc{pith2026250810728,
  author       = {Pith},
  title        = {Pith review of: Global dynamical stability for KMS-states},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/TTHAOKQC}},
  note         = {Machine review of arXiv:2508.10728}
}

read the original abstract

The Hugenholz-Boltzmann-evolution is generalized to strongly interacting systems on the lattice. Under appropriate assumptions states stable under this evolution are shown to satify the KMS-condition. How far these assumptions are reasonable is discussed.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

20 extracted references · 20 canonical work pages

[5]

Interacting Fermi systems in the tracial state

H. Narnhofer: Interacting systems in the tracial state, arXiv 2406.04977

work page Pith review arXiv
[6]

Asymptotics for automorphisms of quantum systems on the lattice

H. Narnhofer: Asymptotics for quantum systems on the lattice, arXiv 2405.19696

work page Pith review arXiv
[1]

Haag, N.M

R. Haag, N.M. Hugenholz, M. Winnink: Commun. Math. Phys.5 215 (1967)

work page 1967
[2]

Bratteli, D.W

O. Bratteli, D.W. Robinson: Operator Algebras and Quantum Statistical Mechanics I and II, Springer (1981)

work page 1981
[3]

Pusz, S.L

W. Pusz, S.L. Woronowicz: Commun. Math. Phys.58 273 (1978)

work page 1978
[4]

R. Haag, D. Kastler, E. Trych-Pohlmeyer: Commun. Math. Phys.56214 (1977)

work page 1977
[7]

Narnhofer:Journ

H. Narnhofer:Journ. Phys. A:Math. Theor.41 335211 (2008)

work page 2008
[8]

Pulvirenti: Boltzmann equation (Classical and Quantum) Encyclopedia of mathematical Physics 306-312 (2006)

M. Pulvirenti: Boltzmann equation (Classical and Quantum) Encyclopedia of mathematical Physics 306-312 (2006)

work page 2006

Show all 20 references

[9]

Lanford III: in Springer Lecture Notes35 (1975) 1-111

O. Lanford III: in Springer Lecture Notes35 (1975) 1-111

1975
[10]

Illner, M

R. Illner, M. Pulvirenti: Commun. Math. Phys.121 134 (1989)

1989
[11]

Hugenholz: Journal of Statistical Physics32/2 231 (1983)

N.M. Hugenholz: Journal of Statistical Physics32/2 231 (1983)

1983
[12]

Lindblad: Commun

G. Lindblad: Commun. Math. Phys.48 119 (1976)

1976
[13]

Alicki, K

R. Alicki, K. Lendi: Quantumm Dynamical Semigroups and Applications. Lect. Notes Phys.717Springer-Verlag Berlin (2007)

2007
[14]

Benatti, F

F. Benatti, F. Carollo, R. Floreanini, H. Narnhofer: J. Phys.A: Math. Theor.50 (2017) 423001

2017
[15]

Narnhofer: Acta Phys

H. Narnhofer: Acta Phys. Austr.47 (1977) 1

1977
[16]

Henri Poincare4 (2003) 63-83

T.Matsui: Ann. Henri Poincare4 (2003) 63-83

2003
[17]

Lieb, D.W

E.H. Lieb, D.W. Robinson: Commun. Math. Phys.28 (1972) 251-257

1972
[18]

Araki: Publ.RIMS, Kyoto Univ20 277 (1984) 6

H. Araki: Publ.RIMS, Kyoto Univ20 277 (1984) 6

1984
[19]

Radin: Commun

C. Radin: Commun. Math. Phys.62 159 (1978)

1978
[20]

Baumgartner, F

B. Baumgartner, F. Benatti, H. Narnhofer: Journ. Phys.A: Math. Theor.43 115301 (2010)

2010

Pith tools

Reviewed August 5, 2026 · model on record in the stance chip above.

[1] [5]

Interacting Fermi systems in the tracial state

H. Narnhofer: Interacting systems in the tracial state, arXiv 2406.04977

work page Pith review arXiv

[2] [6]

Asymptotics for automorphisms of quantum systems on the lattice

H. Narnhofer: Asymptotics for quantum systems on the lattice, arXiv 2405.19696

work page Pith review arXiv

[3] [1]

Haag, N.M

R. Haag, N.M. Hugenholz, M. Winnink: Commun. Math. Phys.5 215 (1967)

work page 1967

[4] [2]

Bratteli, D.W

O. Bratteli, D.W. Robinson: Operator Algebras and Quantum Statistical Mechanics I and II, Springer (1981)

work page 1981

[5] [3]

Pusz, S.L

W. Pusz, S.L. Woronowicz: Commun. Math. Phys.58 273 (1978)

work page 1978

[6] [4]

R. Haag, D. Kastler, E. Trych-Pohlmeyer: Commun. Math. Phys.56214 (1977)

work page 1977

[7] [7]

Narnhofer:Journ

H. Narnhofer:Journ. Phys. A:Math. Theor.41 335211 (2008)

work page 2008

[8] [8]

Pulvirenti: Boltzmann equation (Classical and Quantum) Encyclopedia of mathematical Physics 306-312 (2006)

M. Pulvirenti: Boltzmann equation (Classical and Quantum) Encyclopedia of mathematical Physics 306-312 (2006)

work page 2006