Pith. sign in

REVIEW 2 major objections 4 minor 37 references

Local integrability breaking and exponential localization of leading Lyapunov vectors

T0 review · 2 major / 4 minor · reviewed 2026-08-11 · deepseek-v4-flash

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

Pith's one-line read The paper shows that a single-site integrability-breaking perturbation in a discrete space-time spin lattice makes the Lyapunov spectrum singular in the thermodynamic limit: the density of exponents collapses to a delta function while a…

desk verdict Genuinely new numerical result on local integrability breaking, with extensive numerics and a plausible but unproven central relation; deserves serious refereeing, conditional on error bars and honest labeling of the conjecture. read the letter →

arxiv 2412.16887 v2 pith:KJCYYH6M submitted 2024-12-22 cond-mat.stat-mech nlin.CD

Jiaozi Wang , Tomaž Prosen , Giulio Casati This is my paper · ORCID
classification cond-mat.stat-mechnlin.CD
keywords LyapunovspectrumvectorslocalintegrabilitybreakingexponentiallocalizationKardar-Parisi-Zhangscalingspace-timelatticeLandau-Lifshitzmagnetdynamicalexponent
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 studies what happens to deterministic chaos and transport when a discrete space-time lattice that is integrable is perturbed at a single site. It claims that in the thermodynamic limit the density of Lyapunov exponents becomes a delta function $\delta(\lambda)$ at zero, so that almost all exponents vanish, while a vanishing fraction of non-zero exponents survives and converges. The Lyapunov vectors belonging to those surviving exponents are exponentially localized around the impurity, with localization length equal to the inverse of the corresponding exponent. The authors also find that this local impurity does not change the bulk transport universality class: spin-spin and current-current correlations keep the Kardar-Parisi-Zhang scaling with dynamical exponent $z=3/2$ at zero magnetization, and ballistic scaling at non-zero magnetization. A sympathetic reader would care because it suggests a simple, universal structure for how local integrability breaking affects many-body chaos.

What carries the argument

The machinery is the local perturbation combined with a light-cone argument: an infinitesimal deviation seeded at a site $x^*$ first grows with the maximal Lyapunov exponent $\lambda_1$ until the influence of the perturbation at $x=0$ reaches it after a time $|x^*|/v$ (with sound speed $v=1$ here); the resulting time-delayed growth, compared with the free-growth formula, forces the overlap of the first Lyapunov vector with site $x^*$ to decay as $e^{-\lambda_1 |x^*|}$, hence $\xi_1 = 1/\lambda_1$, and the argument is extended to subleading vectors $k>1$. The numerical workhorse is a standard QR-decomposition scheme for extracting Lyapunov exponents and backward Lyapunov vectors from long trajectories. For transport, the load-bearing objects are the spin-spin correlation $C(x_1,x_2,t)$ and the current-current correlation $C_{JJ}(t)$, whose scaling collapses under $t^{1/z}$ with $z=3/2$ ($\mu=0$) or $z=1$ ($\mu\neq 0$).

What would settle it

Compute, for system sizes up to $L=2^{14}$ and long times, the logarithm of the spatial profile of the second and third backward Lyapunov vectors; if the slope $-\log \rho_k(x)/|x|$ is not numerically equal to $\lambda_k$ for large $|x|$ (or deviates systematically as the perturbation strength $\phi\to 0$), the exponent of localization is not the Lyapunov exponent and Eq. (7) fails. Alternatively, measure $C(x,-x,t)$ at fixed small $x$ and increasing $t$: a crossover from $t^{-1/3}$ (KPZ) to $t^{-1/2}$ (diffusive) scaling would falsify the claim that the local impurity leaves the bulk transport universality class unchanged.

Watch

Extended reading notes

Core claim

Consider the SO(3)-symmetric discrete space-time Landau-Lifshitz magnet with a local rotation applied to a single site at $x=0$. The paper's central discovery is that this local integrability breaking produces a Lyapunov spectrum whose thermodynamic limit $L\to\infty$ is singular: the spectral density $P(\lambda)$ tends to $\delta(\lambda)$, so the overwhelming majority of Lyapunov exponents vanish, while a sub-extensive sequence of finite exponents $\lambda_k$ survives. The corresponding backward Lyapunov vectors are exponentially localized around the impurity, with spatial profile $\rho_k(x) \asymp e^{-\lambda_k |x|}$, that is, the localization length is $\xi_k = 1/\lambda_k$. The same perturbation leaves the transport properties of the bulk unchanged in the scaling limit: the two-point spin and current correlators obey KPZ scaling with $z=3/2$ at zero magnetization (and ballistic scaling at non-zero magnetization), with the spread of the perturbation's influence governed by the same dynamical exponent. The authors conjecture this structure is universal for local integrability breaking in locally interacting many-body systems.

Load-bearing premise

The argument for $\xi_k=1/\lambda_k$ assumes that a small deviation near site $x$ grows at the unperturbed maximal rate $\lambda_1$ both before and after the local impurity's influence reaches it, the influence acting only as a time delay $|x|/v$, and that this same single-rate picture carries over to every subleading vector $k$.

Editorial extensions

If this is right

  • In the thermodynamic limit, the Lyapunov spectral density collapses to $\delta(\lambda)$; all but a sub-extensive set of exponents vanish.
  • Each surviving non-zero exponent $\lambda_k$ corresponds to a Lyapunov vector exponentially localized around the impurity, with localization length $\xi_k=1/\lambda_k$; vectors of zero exponents become delocalized.
  • The local impurity does not change the bulk transport universality class: the KPZ dynamical exponent $z=3/2$ at zero magnetization and ballistic transport at non-zero magnetization persist in the thermodynamic limit.
  • A two-parameter scaling function for the spin correlator exists, whose single-parameter slice is the standard KPZ scaling function.
  • The same structure is observed in a locally perturbed classical unitary circuit map, suggesting universality across different locally perturbed integrable systems.

Reading between the lines

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

  • A testable extension the paper leaves open: if the number of impurities grows as a finite density, the singular spectrum should break down, with the number of non-zero exponents becoming extensive and localization lengths shortening; this crossover could be probed numerically.
  • A natural quantum analogue: in spin chains with a single impurity, the response of out-of-time-order correlators or operator spreading might show a similar exponential localization with length set by the local Lyapunov rate, if such a rate can be defined.
  • The sharp light-cone assumption could be relaxed: in models with sub-ballistic front propagation, the relation $\xi_k=1/\lambda_k$ would acquire corrections set by the front's spreading exponent, a modification not covered by the paper's heuristic derivation.
  • The two-parameter scaling function is presented as a limit; whether its shape is universal or changes with the perturbation strength $\phi$ and the coupling $\tau$ is not tested in the paper and could be checked by data collapse at different parameters.
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

2 major / 4 minor

Summary. The paper studies the discrete space-time Landau-Lifshitz magnet with a single-site local rotation that breaks integrability. It reports three main findings: (i) the Lyapunov spectrum becomes singular in the thermodynamic limit, with a vanishing fraction of nonzero exponents and a density converging to a delta function; (ii) the Lyapunov vectors corresponding to the nonzero exponents are exponentially localized around the impurity, with localization lengths satisfying ξ_k = 1/λ_k; and (iii) spin-spin and current-current correlations retain KPZ scaling with dynamical exponent z=3/2 at zero magnetization and ballistic scaling at nonzero magnetization, so the local impurity does not change bulk transport universality. The theoretical support for the localization-length relation is a heuristic delay-growth argument, while the transport and spectral claims are supported by extensive numerical simulations, including a second classical unitary circuit model in the supplemental material.

Significance. If the results hold, they identify a new and potentially universal mechanism for local integrability breaking: a single-site impurity produces a singular Lyapunov spectrum and exponentially localized Lyapunov vectors, while leaving bulk transport in the integrable universality class unchanged. The numerical work is substantial and carefully executed: long integration times (up to 10^6–10^7), large ensembles (up to 5×10^7 initial states), and scaling collapses that are visually convincing. The paper also contains a useful consistency check: the localization decay rate is taken from the independently measured Lyapunov spectrum rather than fitted from the spatial profile. The universality conjecture is supported by a second model in the supplemental material. The main weakness is that the central quantitative relation ξ_k = 1/λ_k is not derived from the tangent-space dynamics and is presented with a degree of confidence that the heuristic argument does not fully justify.

major comments (2)
  1. [Lyapunov spectrum and Lyapunov vectors, Eq. (7) and paragraph after Eq. (8)] The heuristic argument does not establish Eq. (7) as a derived result. For k=1, the argument yields only the upper bound ξ1 ≤ v/λ1, not the observed equality; the saturation to v/λ1 is assumed rather than explained. More importantly, the post-front growth of the tangent-space deviation is assumed to occur at the unperturbed global rate λ1, but the local rotation at x=0 modifies the Jacobian at that site, so a deviation passing through the impurity can be scattered into other directions and may grow at a different rate. If the effective post-front rate were λ' ≠ λ1, the same comparison would give ξ1 = v/λ', and the numerical agreement in Fig. 2(a) would be parametric. The extension to subleading vectors k>1 is even less secure because the orthogonalized subspace dynamics defining ψ_k is not described by the same delay-growth picture. I recommend either deriving the equality from the tangent-space dynamics or explicitly presenting Eq. (7) as a numerically supported conjecture, with the upper bound as the only proven statement.
  2. [Fig. 1 and text after 'Lyapunov spectrum and Lyapunov vectors'] The claim that lim_{L→∞} P(λ) = δ(λ) is an extrapolation from visual narrowing of the central spectral peak for L = 2^7, ..., 2^10. The paper does not provide error bars or a quantitative finite-size scaling analysis of the peak width, peak height, or the fraction of exponents inside a fixed window. The convergence of the five largest exponents in Fig. 2(c) does not by itself imply that the density becomes singular. Please add a scaling collapse for the central peak (e.g., width w(L) ∝ L^{-α} with α>0, or an equivalent measure) or soften the statement to a finite-size-supported conjecture.
minor comments (4)
  1. [Paragraph after Eq. (6)] 'Lyapunov exponense' should be 'Lyapunov exponents'.
  2. [Fig. 2 caption and text] The text says 'Fig. 2 (b) shows Ik for the first 100 Lyapunov vectors', but the IPR panel is Fig. 2 (d); please correct the reference.
  3. [Eq. (S7) in the Supplemental Material] Please clarify the normalization convention in the QR-based estimate of λ_k: the expression uses Rjj(2n) only, and it should be stated whether this corresponds to a full map step or a half step and how the time average is normalized.
  4. [Eq. (8) and following paragraph] The notation d(t) ≡ |δX^t| and the subsequent comparison with Eq. (8) would benefit from a statement that the norm is taken after projecting onto the tangent sphere; otherwise the argument is ambiguous about whether δX^t is the full tangent vector or its magnitude.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the central scaling relation ξ_k=1/λ_k is an independently measured consistency check, not a fitted input.

full rationale

The derivation chain is self-contained. The Lyapunov exponents λ_k are computed by the standard Benettin QR method (Eqs. S5-S7), and the spatial profiles ρ_k(x) are measured from the QR basis (Eq. 6 and Fig. 2); Eq. (7) is a comparison between these two independently computed quantities, not a fit of one from the other. The heuristic argument after Eq. (8) uses the measured λ_1 to argue for a bound ξ_1 ≲ v/λ_1 and 'suggests' the scaling (7); even though the argument is not rigorous and the claimed equality is stronger than the derived bound, this is a plausibility argument rather than a circular reduction. The transport conclusions are based on direct spatio-temporal spin-spin and current-current correlation measurements (Figs. 3-4) showing KPZ scaling, with the integrable KPZ behavior of Ref. [23] used as a benchmark, not as the derived result. Self-citations to Refs. [23-25,33] concern the unperturbed model and are not load-bearing for the new claims; no uniqueness theorem is imported from prior work, and no parameter is fitted and then renamed a prediction. The main limitations are the unproven light-cone delay assumption and the extension to subleading vectors, but these are correctness/rigor concerns, not circularity.

Assumptions & free parameters 0 free parameters · 4 assumptions · 0 invented entities

No quantities were fitted to data. The fixed model parameters τ=1 and ϕ=π/2 are inputs; the central local-length scaling is checked across other values in the supplemental material. The sound speed v=1 is a property of the unperturbed map, not fitted. No new entities are introduced; the only postulated object is the universality conjecture itself, which is explicitly labelled as conjectural.

assumptions (4)
  • domain assumption The unperturbed map Ψτ (with ϕ=0) is integrable and its zero-magnetization spin transport is KPZ with z=3/2 (Ref. [23]).
    Used as the baseline model; the paper does not re-derive these properties and relies on Ref. [23] for the definition of the integrable dynamics and its transport exponents.
  • domain assumption The canonical invariant ensemble ρtot_µ of Eq. (5) correctly samples fixed-magnetization states, and total z-magnetization S is the only conserved quantity after perturbation.
    The ensemble is standard for a single conserved charge; the uniqueness of the conserved quantity is stated as 'verified' numerically but not proven.
  • domain assumption The Lieb-Robinson or sound speed for the discrete space-time lattice is v=1, so a local perturbation at site 0 affects site x only after time t > |x|/v.
    Invoked in the heuristic derivation of Eq. (7); the value v=1 is stated without derivation, relying on the nearest-neighbor structure of the map.
  • standard math The Benettin/QR procedure at finite integration time tmax up to 10^7 yields converged Lyapunov exponents and backward Lyapunov vectors.
    The method is standard and cited [31]; convergence is assumed rather than demonstrated with explicit convergence tests, though the long times make it plausible.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Local integrability breaking and exponential localization of leading Lyapunov vectors." pith.science (2026). https://pith.science/paper/KJCYYH6M

@misc{pith2026241216887,
  author       = {Pith},
  title        = {Pith review of: Local integrability breaking and exponential localization of leading Lyapunov vectors},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/KJCYYH6M}},
  note         = {Machine review of arXiv:2412.16887}
}
abstract

We study integrability breaking and transport in a discrete space-time lattice with a local integrability breaking perturbation. We find a singular distribution of the Lyapunov spectrum where the majority of Lyapunov exponents vanish in the thermodynamic limit. The sub-extensive sequence of nonzero exponents, converging in the thermodynamic limit, correspond to Lyapunov vectors that are exponentially localized with localization lengths proportional to inverse Lyapunov exponents. Moreover, we investigate the transport behavior of the system by considering the spin-spin and current-current spatio-temporal correlation functions. Our results indicate that the overall transport behavior, similarly as in the purely integrable case, conforms to Kardar-Parisi-Zhang scaling in the thermodynamic limit and at vanishing magnetization. The same dynamical exponent $z=3/2$ governs the effect of local perturbation spreading in the bulk.

Figures

Figures reproduced from arXiv: 2412.16887 by the authors.

Figure 1
Figure 1. FIG. 1. The Lyapunov density [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. FIG. 2. Spatial distribution ¯ρ [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. FIG. 3. Spin-spin correlation function [PITH_FULL_IMAGE:figures/full_fig_p003_3.png] view at source ↗
Figures from the paper (1 more)
Figure 4
Figure 4. Figure 4: FIG. 4. Current-Current correlation function [PITH_FULL_IMAGE:figures/full_fig_p004_4.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

37 extracted references · 36 canonical work pages

  1. [1]

    A. N. Kolmogorov, in Stochastic Behavior in Classical and Quantum Hamiltonian Systems , edited by G. Casati and J. Ford (Springer Berlin Heidelberg, Berlin, Heidel- berg, 1979) pp. 51–56

    work page 1979

  2. [2]

    Considering a lattice of an even number L ∈ 2N of unit vectors [29] St x, we define a dynamical map Ψτ : (St+1 0 , St+1 1 ,

    = Φτ (S1, S2), S′ 1 = 1 σ2 + τ 2 (σ2S1 + τ 2S2 + τ S1 × S2), S′ 2 = 1 σ2 + τ 2 (σ2S2 + τ 2S1 + τ S2 × S1), (1) where σ2 := 1 2 (1 + S1 · S2). Considering a lattice of an even number L ∈ 2N of unit vectors [29] St x, we define a dynamical map Ψτ : (St+1 0 , St+1 1 , . . . ,St+1 L−1) = Ψτ (St 0, St 1, . . . ,St L−1), Ψτ = Ψ odd τ ◦ Ψ even τ = L/2−1Y x=0 Φ2x...

    work page

  3. [3]

    D. K. Campbell, P. Rosenau, and G. M. Zaslavsky, Chaos: An Interdisciplinary Journal of Nonlinear Science 15 (2005)

    work page 2005

  4. [4]

    Proof of a theorem of a. n. kolmogorov on the invariance of quasi-periodic motions under small perturbations of the hamiltonian,

    V. I. Arnold, “Proof of a theorem of a. n. kolmogorov on the invariance of quasi-periodic motions under small perturbations of the hamiltonian,” in Collected Works: Representations of Functions, Celestial Mechanics and KAM Theory, 1957–1965 , edited by A. B. Givental, B. A. Khesin, J. E. Marsden, A. N. Varchenko, V. A. Vassiliev, O. Y. Viro, and V. M. Zak...

    work page 1957

  5. [5]

    Politi, Scholarpedia 8, 2722 (2013), revision no

    A. Politi, Scholarpedia 8, 2722 (2013), revision no. 137286

    work page 2013

  6. [6]

    Nekhoroshev theory,

    L. Niederman, “Nekhoroshev theory,” in Encyclopedia of Complexity and Systems Science , edited by R. A. Meyers (Springer New York, New York, NY, 2009) pp. 5986– 5998

    work page 2009

  7. [7]

    Mithun, C

    T. Mithun, C. Danieli, Y. Kati, and S. Flach, Phys. Rev. Lett. 122, 054102 (2019)

    work page 2019

  8. [8]

    Gaspard, Chaos, Scattering and Statistical Mechanics, Cambridge Nonlinear Science Series (Cambridge Univer- sity Press, 1998)

    P. Gaspard, Chaos, Scattering and Statistical Mechanics, Cambridge Nonlinear Science Series (Cambridge Univer- sity Press, 1998)

    work page 1998

Show all 37 references
  1. [9]

    Mithun, C

    T. Mithun, C. Danieli, M. V. Fistul, B. L. Altshuler, and S. Flach, Phys. Rev. E 104, 014218 (2021)

    2021

  2. [10]

    Danieli, T

    C. Danieli, T. Mithun, Y. Kati, D. K. Campbell, and S. Flach, Phys. Rev. E 100, 032217 (2019)

    2019

  3. [11]

    G. M. Lando and S. Flach, Phys. Rev. E 108, L062301 (2023)

    2023

  4. [12]

    Malishava and S

    M. Malishava and S. Flach, Phys. Rev. Lett. 128, 134102 (2022)

    2022

  5. [13]

    Kaneko, Physica D: Nonlinear Phenomena 23, 436 (1986)

    K. Kaneko, Physica D: Nonlinear Phenomena 23, 436 (1986)

    1986

  6. [14]

    Zhang, G

    W. Zhang, G. M. Lando, B. Dietz, and S. Flach, Phys. Rev. Res. 6, L012064 (2024)

    2024

  7. [15]

    D. A. Egolf, I. V. Melnikov, W. Pesch, and R. E. Ecke, Nature 404, 733 (2000)

    2000

  8. [16]

    Chat´ e, Europhysics Letters21, 419 (1993)

    H. Chat´ e, Europhysics Letters21, 419 (1993)

    1993

  9. [17]

    Ginelli, H

    F. Ginelli, H. Chat´ e, R. Livi, and A. Politi, Journal of Physics A: Mathematical and Theoretical 46, 254005 (2013)

    2013

  10. [18]

    De Monte, F

    S. De Monte, F. d’Ovidio, H. Chat´ e, and E. Mosekilde, Phys. Rev. Lett. 92, 254101 (2004)

    2004

  11. [19]

    Ginelli, P

    F. Ginelli, P. Poggi, A. Turchi, H. Chat´ e, R. Livi, and A. Politi, Phys. Rev. Lett. 99, 130601 (2007)

    2007

  12. [20]

    Pikovsky and A

    A. Pikovsky and A. Politi, Nonlinearity 11, 1049 (1998)

    1998

  13. [21]

    Skokos, I

    C. Skokos, I. Gkolias, and S. Flach, Phys. Rev. Lett. 111, 064101 (2013)

    2013

  14. [22]

    K. A. Takeuchi, F. Ginelli, and H. Chat´ e, Phys. Rev. Lett. 103, 154103 (2009)

    2009

  15. [23]

    with a local magnetic field (impurity). In the ab- sence of impurity, the model is integrable, possesses a non-abelian (SO(3)) symmetry and has been suggested to exhibit Kardar-Parisi-Zhang (KPZ) scaling of 2-point dynamical correlations [23–26]. It becomes nonintegrable after...

    2025 arXiv

  16. [24]

    Taniguchi and G

    T. Taniguchi and G. P. Morriss, Phys. Rev. E 68, 046203 (2003)

    2003

  17. [25]

    Krajnik and T

    ˇZ. Krajnik and T. Prosen, Journal of Statistical Physics 179, 110 (2020)

    2020

  18. [26]

    Ljubotina, M

    M. Ljubotina, M. ˇZnidariˇ c, and T. Prosen, Phys. Rev. Lett. 122, 210602 (2019)

    2019

  19. [27]

    Krajnik, E

    ˇZ. Krajnik, E. Ilievski, and T. Prosen, SciPost Phys. 9, 038 (2020)

    2020

  20. [28]

    K. A. Takeuchi, K. Takasan, O. Busani, P. L. Fer- rari, R. Vasseur, and J. De Nardis, arXiv preprint arXiv:2406.07150 (2024)

    2024 arXiv

  21. [29]

    McCarthy, S

    C. McCarthy, S. Gopalakrishnan, and R. Vasseur, Phys. Rev. B 110, L180301 (2024)

    2024

  22. [30]

    A. J. McRoberts and R. Moessner, Phys. Rev. Lett. 133, 256301 (2024)

    2024

  23. [31]

    Each Sx takes values on 3 D unit sphere and constitute two canonical degree of freedom, hence N ≡ 2L

  24. [32]

    See the Supplemental Material for details on numerical simulations, additional results, verification of the model’s ergodicity, and numerical results for a classical unitary circuit map with local perturbation

  25. [33]

    Benettin, L

    G. Benettin, L. Galgani, A. Giorgilli, and J.-M. Strelcyn, Meccanica 15, 9 (1980)

    1980

  26. [34]

    Pr¨ ahofer and H

    M. Pr¨ ahofer and H. Spohn, Journal of statistical physics 115, 255 (2004)

    2004

  27. [35]

    Krajnik, E

    ˇZ. Krajnik, E. Ilievski, T. Prosen, and V. Pasquier, Sci- Post Phys. 11, 051 (2021)

    2021

  28. [36]

    O. S. Bariˇ si´ c, P. Prelovˇ sek, A. Metavitsiadis, and X. Zo- tos, Phys. Rev. B 80, 125118 (2009)

    2009

  29. [37]

    Brenes, E

    M. Brenes, E. Mascarenhas, M. Rigol, and J. Goold, Phys. Rev. B 98, 235128 (2018). 6 SUPPLEMENT AL MA TERIAL In the supplemental material, we provide more details on numerical simulations, additional numerical results, as well as results verification of ergodicity of the model...

    2018

Pith tools

Reviewed August 11, 2026 · model on record in the stance chip above.