Out-of-Time Ordered Correlator for a Chaotic Many-Body Quantum System

Hans A. Weidenm\"uller

REVIEW 5 major objections 5 minor 9 references

Out-of-Time Ordered Correlator for a Chaotic Many-Body Quantum System

T0 review · 5 major / 5 minor · reviewed 2026-08-12 · deepseek-v4-flash

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

Pith's one-line read The paper claims that for a chaotic many-body quantum system built on the Bohigas–Giannoni–Schmit random-matrix ensemble, the out-of-time ordered correlator decays at large times as a Gaussian with width ℏ/(2Δ), and conjectures that Δ…

desk verdict A careful, honest random-matrix calculation of late-time OTOC decay, with the advertised λ_max connection explicitly labeled as a conjecture rather than a derived result. read the letter →

arxiv 2411.16274 v1 pith:DSGE6V7Z submitted 2024-11-25 math-ph cond-mat.stat-mechmath.MPquant-ph

Hans A. Weidenmüller This is my paper · ORCID

classification math-phcond-mat.stat-mechmath.MPquant-ph MSC 81Q5015B52

keywords out-of-timeorderedcorrelatorquantumchaosrandommatrixtheoryBohigas-Giannoni-SchmitconjectureLyapunovexponentGaussianorthogonalensemblemany-bodysystemenergycorrelationwidth

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

This paper tries to pin down the large-time behavior of the out-of-time ordered correlator (OTOC) for a chaotic many-body quantum system, using a random-matrix parametrization of the Hamiltonian that implements the Bohigas–Giannoni–Schmit conjecture. The central result is that the ensemble-averaged OTOC decays as exp(-2t²Δ²/ℏ²), so the decay time is ℏ/(2Δ), where Δ is the width of the energy interval in which Gaussian orthogonal ensemble statistics holds. The paper further conjectures that Δ is universally related to the largest Lyapunov exponent of the corresponding classical system by Δ = ℏλ_max. If that conjecture is right, the late-time OTOC decay is governed by the dimensionless parameter λ_max t, making OTOC a direct quantum probe of classical chaos.

What carries the argument

The central machinery is the universal parametrization of the chaotic Hamiltonian: $H_{mn} = \sum_\alpha O_{m\alpha} E_\alpha O_{n\alpha}$, with zero-centered Gaussian orthogonal-matrix elements satisfying $\langle O_{m\alpha} O_{n\beta} \rangle = \delta_{mn} \delta_{\alpha\beta} F(E_m - \bar{E}_\alpha)$ and a normalized Gaussian $F$ of width $\Delta$; within $\Delta$ the eigenvalues obey Wigner-Dyson statistics. The argument is carried by the correlated-moment formula for products of $Y(\chi) = \exp(\chi H)$, which to leading order in $1/N$ keeps only non-crossing contraction patterns and yields explicit exponentials in the energy differences and in $\chi^2 \Delta^2/(2k)$. Applying that formula to the four $Y$-insertions in $C(t)$ and $F(t)$, and using the assumption that the test operators have vanishing thermal averages over a width-$\Delta$ window, reduces each average to three terms, all sharing the time factor $\exp(-2t^2\Delta^2/\hbar^2)$.

What would settle it

Measure the two-point energy correlation function of a chaotic quantum system (for instance the Sinai billiard or a kicked rotor) in the semiclassical regime and compare it to a Gaussian; if the correlation function is measurably non-Gaussian, or if the measured large-time OTOC decay is exponential rather than Gaussian with the predicted $\hbar/(2\Delta)$ width, the central claim fails. A cheaper test: check whether the conjectured relation $\Delta = \hbar \lambda_{\max}$ holds by computing $\Delta$ from spectral statistics and $\lambda_{\max}$ from the classical dynamics of the same system.

Watch

Extended reading notes

Core claim

For a time-reversal-invariant chaotic many-body system whose Hamiltonian H = H_HF + V is represented through random orthogonal matrices O and eigenvalues E with Wigner-Dyson statistics inside a correlation width Δ, the paper shows that the thermal OTOC C(t) and the symmetrized function F(t) are self-averaging: their variances are of order 1/N relative to the squared means, so almost every member of the ensemble exhibits the average behavior. The averages decay at large times as Gaussians: ⟨C(t)⟩ carries the factor exp(-2t²Δ²/ℏ²) and tends to a nonvanishing constant built from thermal averages of V² and W², while ⟨F(t)⟩ vanishes with the same Gaussian factor times exp(-3β²Δ²/8). Since the Gaussian time factor comes by Fourier transformation from the assumed Gaussian energy correlation function, the time scale ℏ/Δ is robust to the precise form of that function. The paper conjectures that Δ = ℏλ_max, which would make the late-time OTOC decay universally controlled by the classical Lyapunov exponent.

Load-bearing premise

The whole calculation rests on the assumption that the energy correlation function $F(E_m - \bar{E}_\alpha)$ is a Gaussian of width $\Delta$; the paper itself notes this form rests on numerical evidence and would be a Lorentzian for weak chaos, which would turn the Gaussian OTOC decay into exponential decay.

Editorial extensions

If this is right

If $\Delta = \hbar \lambda_{\max}$ is confirmed, the late-time OTOC decay directly measures the classical Lyapunov exponent in any chaotic many-body system that satisfies the Bohigas–Giannoni–Schmit conjecture.
The decay time scale $\hbar/(2\Delta)$ is independent of the detailed form of the energy correlation function, so it should be a universal signature of strong quantum chaos.
The self-averaging property (variances of order $1/N$) means a single realization of a chaotic many-body system, not only an ensemble average, should show the predicted Gaussian decay.
The asymptotic value of $\langle C(t)\rangle$ gives a relation between thermal averages of $V^2$ and $W^2$, a prediction checkable in numerical simulations.
The conjecture extends the Bohigas–Giannoni–Schmit conjecture by adding a quantitative link between the spectral correlation width and the classical dynamics.

Reading between the lines

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

If the Gaussian decay is confirmed, the Lorentzian alternative for weak chaos suggests a possible order-parameter-like distinction between strong and weak quantum chaos, with the shape of the energy correlation function determining the OTOC decay law.
The approach is restricted to time-reversal-invariant (GOE) systems; an analogous treatment for GUE or GSE ensembles might yield a different numerical coefficient in the Gaussian exponent, worth testing.
The conjectured equality $\Delta = \hbar \lambda_{\max}$, if combined with the BGS interval, could provide a way to extract Lyapunov exponents from spectral statistics without computing time evolution at all.
The discrepancy with exponential Ruelle–Pollicott decay in finite-size matrix models might be resolved by checking whether those models obey the BGS conjecture within a width $\Delta$; if they do not, the Gaussian prediction would not apply to them.

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

The real content here is the explicit ensemble-averaged OTOC for the parametric random-matrix model of Ref. [3]. That calculation is new: Ref. [4] only estimated the asymptotic value, and the Gaussian factor exp(-2t²Δ²/ℏ²) in Eqs. (27) and (28) is worked out rather than guessed. The contraction-counting in Section V is careful, the 1/N suppression of the variance in Section VIII is plausible, and the paper is candid about what it can and cannot do. Credit where due: this is a legitimate analytic result about a specific many-body random-matrix ensemble, not a hollow claim. The soft spots are real but they are not hidden. The Gaussian time decay is inherited from the assumed Gaussian form of F in Eq. (15). The author says so himself and notes that a Lorentzian would change the decay shape while preserving the ℏ/Δ time scale. That is a fair caveat. The more load-bearing issue is the conjecture Δ = ℏλ_max in Eq. (31). The abstract's physical punchline—late-time OTOC governed by λ_max t—rests entirely on that conjecture, which is supported only by Arve's 400 Sinai-billiard levels and is in tension with Ref. [9]. The stress-test note is right about this. But the paper does not dress the conjecture as a theorem; it is flagged as a conjecture, and the conflict with Ref. [9] is acknowledged rather than ignored. I do not think there is a load-bearing flaw in the calculation itself. If you drop Eq. (31), you still have a well-defined statement about the correlation width Δ, and the shape of the decay follows from the assumed F. The math in the moments section looks consistent to me, and the treatment of Tr(Z) fluctuations is reasonable for N ≫ 1. The main risk is not circularity—it is that the universal Lyapunov claim may be empirically wrong or restricted to a narrower class of systems than the BGS conjecture covers. That is a scientific question worth refereeing, not a reason to desk-reject. Who should read this: anyone working on OTOC in random-matrix or chaotic many-body settings, and anyone interested in the BGS conjecture's quantitative content. It deserves a serious referee—the calculation should be checked, the conjecture should be challenged, and the relation to Ref. [9] should be sharpened. My own verdict would be conditional, not accept, but this is exactly the kind of paper peer review exists for.

Referee Report

5 major / 5 minor

Summary. The paper considers a time-reversal-invariant chaotic many-body quantum system described by a Hartree-Fock term plus a residual interaction, and models its Hamiltonian with a random-matrix ensemble in which eigenvectors and eigenvalues obey local GOE statistics over an energy correlation width Δ. Using a non-crossing contraction expansion for moments of the evolution operator, the author computes the ensemble-averaged out-of-time-ordered correlator C(t) and the symmetrized function F(t) to leading order in 1/N, obtaining a Gaussian time decay exp(-2t^2Δ^2/ℏ^2) and a static asymptotic term for C(t). The author also bounds the variance by 1/N and conjectures Δ = ℏλ_max, so that the late-time OTOC is governed by the dimensionless parameter λ_max t.

Significance. If the calculation is correct, the paper provides an explicit analytic formula for the late-time OTOC in a random-matrix model of many-body chaos and identifies a concrete time scale, ℏ/Δ, set by the Bohigas-Giannoni-Schmit correlation width. The moment expansion in Section V and the 1/N bookkeeping are carefully organized, and the variance estimate in Section VIII is a useful self-averaging result. The author is also transparent about two major caveats: the Gaussian form of the correlation function in Eq. (15) is based on numerical evidence, and the connection to Lyapunov exponents through Eq. (31) is a conjecture. However, the internal inconsistency in Eq. (20), the very strong assumption in Eq. (26), and the mismatch between the variance bound and the almost-everywhere claim in Eq. (30) mean that the central results are not yet established as stated.

major comments (5)

[Section V, Eq. (20)] The printed connected k-th moment has prefactor (√(2π)ρ(E_{m_1})Δ)^{k-1} and a positive exponent +Σ_{j<l}(E_{m_j}-E_{m_l})^2/(2kΔ^2). These two features are internally inconsistent with the equations that are said to follow from Eq. (20): inserting k=2 and χ_1=χ_2=-β gives Eq. (24) only if the prefactor is inverted and the pairwise exponent has a minus sign, and the same negative Gaussian factors appear in Eqs. (27) and (33). As printed, the moment grows with energy separation and has the wrong 1/N scaling, so the derivation of the central results is not self-consistent.
[Section VII, Eq. (26)] The 'slightly more stringent' assumption is not slight. For a dense spectrum, requiring Σ_m A_{mm}e^{-βE_m}e^{-(E_m-E_n)^2/(2kΔ^2)}=0 for every E_n, β, and positive integer k forces the measure ρ(E_m)A_{mm}e^{-βE_m} to vanish identically, because the Gaussian factor has a positive Fourier transform. Thus V and W are effectively required to have vanishing diagonal matrix elements in the Hartree-Fock basis. The paper gives no argument that the simple operators of Ref. [4] satisfy this condition, and without Eq. (26) the terms discarded after possibility (ii) in Section VII contribute at an uncontrolled order to Eqs. (27) and (28).
[Section IX, Eq. (31)] The abstract's headline statement that the large-time OTOC is governed by λ_max t is not a consequence of the calculation: the derivation produces a decay with time scale ℏ/Δ, and the identification Δ = ℏλ_max is a conjecture imported from Ref. [8] and not derived within the ensemble of Section III. The paper itself notes the discrepancy with Ref. [9] and does not resolve it. The physical claim should be explicitly marked as conditional on Eq. (31), or Eq. (31) should be supported by independent evidence.
[Section VIII and Eq. (30)] Section VIII bounds the variance at fixed t by order 1/N relative to the squared mean. This yields convergence in probability for each fixed t, not the assertion in Eq. (30) that F(t)=⟨F(t)⟩ and C(t)=⟨C(t)⟩ for almost all members of the ensemble. To justify an almost-everywhere statement uniformly in t, the paper needs a continuity or tightness argument, or it should weaken Eq. (30) to a statement about ensemble averages and fixed-time fluctuations.
[Section IX and Eq. (15)] The explicit Gaussian factor exp(-2t^2Δ^2/ℏ^2) is the Fourier transform of the assumed Gaussian model for F, and the paper concedes that this form is based on numerical evidence and may fail for weak chaos. Since the Gaussian shape is the central quantitative prediction, the paper should either present Eq. (15) as a model assumption whose consequences are conditional, or repeat the calculation for a generic correlation function of width Δ in order to separate the robust time scale ℏ/Δ from the shape-dependent decay.

minor comments (5)

[Section III, Eqs. (14) and (15)] Equation (14) uses F(E_m - E_α) while Eq. (15) defines F(E_m - \bar E_α); the overline should appear consistently in both places.
[Throughout] The word 'Ljapunov' should be 'Lyapunov' (abstract and Section IX), and there are typographical errors such as 'advantange', 'occuring', and 'diffult' that should be corrected.
[Appendix, Eq. (33)] Equation (33) has a missing brace in the exponential: it should read exp{-(E_{m_1}-E_{m_3})^2/(4Δ^2)} rather than exp -(E_{m_1}-E_{m_3})^2/(4Δ^2)}.
[Section V, Eq. (20)] The notation m_{j+1} with j=k implicitly means m_1; this cyclic convention should be stated explicitly when the product of Kronecker deltas is introduced.
[Section V, Eq. (18)] The statement that the factors formally commute but their order must be respected is clear, but the claim that only one connected contraction pattern survives for general k would be easier to verify with an explicit k=3 example illustrating the non-crossing rule.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the OTOC decay in Δ is a conditional consequence of the assumed correlation width, and the Δ=ℏλmax link is an openly labeled conjecture, so the central claim is not forced by construction.

full rationale

The derivation is self-contained: given the random-matrix ensemble defined by Eqs. (12)-(15), the Gaussian decay factors in Eqs. (27)-(28) follow from the Gaussian integration in Eq. (20), and the paper explicitly identifies this as a Fourier-transform consequence of Eq. (15) rather than as an independent prediction. The time scale ℏ/Δ is a conditional consequence of the assumed correlation width Δ, not a fitted parameter renamed as a prediction, so the observation that the output inherits the input scale is not circularity. The advertised λmax t dependence rests entirely on Eq. (31), which the paper labels a conjecture based on Arve's Sinai-billiard data (Ref. [8]); this is an unsupported external link, not a derived result smuggled in by definition. The paper also flags the conflict with Ref. [9] and the uncertain Gaussian-vs-Lorentzian shape of F in Eq. (15), calling the Gaussian form 'very likely' but not on 'absolutely firm ground'; these are robustness and correctness concerns, not circular steps. Self-citation of Ref. [3] supplies the modeling assumption and the parametric representation, but not the OTOC result itself, so the citation is not load-bearing in a way that makes the present derivation circular.

Assumptions & free parameters 2 free parameters · 6 assumptions · 0 invented entities

The central claim rests on the BGS conjecture, the specific Gaussian correlation form imported from Ref. [3], the stronger zero-one-point condition (26), the large-N limit, and the new conjecture Δ = ℏλ_max. The only free parameter in the OTOC time scale is Δ, an input width, and the λ_max connection is an additional unproved axiom.

free parameters (2)

Δ (energy correlation width)
Energy correlation width within which BGS/GOE statistics apply; an input of the random-matrix ensemble, not derived. The OTOC decay time ℏ/(2Δ) is directly proportional to this parameter.
ρ(E) (average level density)
Average level density of the Hartree-Fock Hamiltonian, an input of the semiclassical regime; N = Δρ is assumed to be much larger than 1.

assumptions (6)

domain assumption BGS conjecture: eigenvalues and eigenvectors locally follow GOE statistics within an interval of width Δ.
Section III, central modeling input.
domain assumption Gaussian form of F(E_m - E̅_α) with width Δ (Eq. (15)), based on numerical evidence in Ref. [3].
Determines the Gaussian time decay via Fourier transform; the author flags it as not firm.
ad hoc to paper Stronger zero-one-point condition (Eq. (26)): thermal averages of V and W vanish over any interval of width Δ.
Imposed to eliminate terms and simplify the average; stronger than the standard condition (25) and not physically derived.
domain assumption Semiclassical limit N = Δρ >> 1, keeping leading order in 1/N.
Used throughout; justifies replacing Tr(Z) with its average and ignoring many contraction patterns.
ad hoc to paper Conjecture Δ = ℏλ_max (Eq. (31)).
Unproved conjecture generalizing the Sinai billiard relation from Ref. [8]; the headline statement that OTOC is governed by λ_max t depends on it.
domain assumption Statistical independence of eigenvalues and eigenvectors.
Postulated after Eq. (13) in the GOE setting.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Out-of-Time Ordered Correlator for a Chaotic Many-Body Quantum System." pith.science (2026). https://pith.science/paper/DSGE6V7Z

@misc{pith2026241116274,
  author       = {Pith},
  title        = {Pith review of: Out-of-Time Ordered Correlator for a Chaotic Many-Body Quantum System},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/DSGE6V7Z}},
  note         = {Machine review of arXiv:2411.16274}
}

abstract

Using the parametric representation of a chaotic many-body quantum system derived earlier, we calculate explicitly the large-time dependence and asymptotic value of the out-of-time correlator (OTOC) of that system. The dependence on time $t$ is determined by $\Delta t / \hbar$. Here $\Delta$ is the energy correlation width within which the Bohigas-Giannoni-Schmit conjecture applies. We conjecture that $\Delta$ is universally related to the leading Ljapunov coefficient of the corresponding classical system by $\Delta = \hbar \lambda_{\max}$. Then the large-time behavior of OTOC is given by the dimensionless parameter $\lambda_{\max} t$.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

9 extracted references · 9 canonical work pages

[3]

H. A. Weidenm¨ uller, J. Math. Phys. A: Math. Theor. 57, 165001 (2024)

work page 2024
[4]

Maldacena, S

J. Maldacena, S. H. Shenker, and D. Stanford, JHEP 2016, 1 06 (2016)

work page 2016
[9]

Garcia-Mata, M

I. Garcia-Mata, M. Saraceno, R. A. Jalabert, A. J. Rongag lia, and D. A. Wisniacki, Phys. Rev. Lett. 121, 210601 (2018)

work page 2018
[8]

Arve, Phys

P . Arve, Phys. Rev. A 44, 6920 (1991)

work page 1991
[1]

A. I. Larkin and Y . N. Ovchinnikov, JETP 28, 1200 (1969)

work page 1969
[2]

Garcia-Mata, R

I. Garcia-Mata, R. A. Jalabert, and D. A. Wisniacki, Scho larpe- dia 18(4), 55237 (2023)

work page 2023
[5]

Bohigas, M

O. Bohigas, M. J. Giannoni, and C. Schmit, Phys. Rev. Lett . 52, 1 (1984)

work page 1984
[6]

M. L. Mehta, Random Matrices (3rd ed.), Academic Press, New Y ork (2004)

work page 2004

Show all 9 references

[7]

K. K. Mon and J. B. French, Ann. Phys. (N.Y .) 95, 90 (1975)

1975

Pith tools

Reviewed August 12, 2026 · model on record in the stance chip above.

[1] [3]

H. A. Weidenm¨ uller, J. Math. Phys. A: Math. Theor. 57, 165001 (2024)

work page 2024

[2] [4]

Maldacena, S

J. Maldacena, S. H. Shenker, and D. Stanford, JHEP 2016, 1 06 (2016)

work page 2016

[3] [9]

Garcia-Mata, M

I. Garcia-Mata, M. Saraceno, R. A. Jalabert, A. J. Rongag lia, and D. A. Wisniacki, Phys. Rev. Lett. 121, 210601 (2018)

work page 2018

[4] [8]

Arve, Phys

P . Arve, Phys. Rev. A 44, 6920 (1991)

work page 1991

[5] [1]

A. I. Larkin and Y . N. Ovchinnikov, JETP 28, 1200 (1969)

work page 1969

[6] [2]

Garcia-Mata, R

I. Garcia-Mata, R. A. Jalabert, and D. A. Wisniacki, Scho larpe- dia 18(4), 55237 (2023)

work page 2023

[7] [5]

Bohigas, M

O. Bohigas, M. J. Giannoni, and C. Schmit, Phys. Rev. Lett . 52, 1 (1984)

work page 1984

[8] [6]

M. L. Mehta, Random Matrices (3rd ed.), Academic Press, New Y ork (2004)

work page 2004