Pith. sign in

REVIEW 5 major objections 217 references

The paper claims that treating an RL agent and its environment as a discrete-time autonomous dynamical system lets Finite-Time Lyapunov Exponent ridges serve as safety barriers, with repelling structures marking unsafe regions and attractin

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

The claimed RL safety verification framework is absent from the manuscript; the body text is an unrelated high-energy physics paper about de Sitter horizon chaos.

T0 review reviewed 2026-08-05 challenge →

load-bearing objection The abstract describes an RL safety framework, but the full text is a different hep-th paper, so there is nothing here to referee. the 5 major comments →

arxiv 2508.15588 v1 pith:I3JFEKNH submitted 2025-08-21 cs.AI

A Dynamical Systems Framework for Reinforcement Learning Safety and Robustness Verification

classification cs.AI
keywords reinforcement learning safetyFinite-Time Lyapunov ExponentLagrangian Coherent Structurespolicy verificationrobustness metricsdynamical systemssafety barrierstrap states
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

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 tries to establish that a learned policy's safety and robustness are not just reward statistics but geometric features of the closed-loop dynamics. By viewing the agent–environment loop as a discrete-time autonomous dynamical system, it uses finite-time Lyapunov exponents to extract Lagrangian Coherent Structures: repelling ridges act as separatrices that shield unsafe regions, while attracting ridges expose convergence behavior and hidden trap states. To make the picture quantitative, the paper introduces three metrics—Mean Boundary Repulsion, Aggregated Spurious Attractor Strength, and Temporally-Aware Spurious Attractor Strength—and argues these measure a policy's safety margin and robustness. If correct, this gives an interpretable, quantifiable safety audit for RL policies in settings where high reward hides critical failure modes.

Core claim

The central claim is that the closed-loop system formed by a policy and its environment has an observable geometric skeleton, and that this skeleton carries the safety-relevant information. Repelling Lagrangian Coherent Structures, computed as ridges of the finite-time Lyapunov exponent field, mark the boundaries that separate safe from unsafe regions of state space; attracting LCS reveal the sets toward which the system converges, including spurious attractors that act as traps. The paper further claims that the strength and temporal persistence of these structures can be condensed into scalars—MBR for the repulsion of unsafe boundaries, ASAS and TASAS for aggregated and temporally-aware at

What carries the argument

The central object is the finite-time Lyapunov exponent (FTLE) field, a per-state number measuring how quickly nearby starting conditions separate over a finite integration window. Ridges of this field are the Lagrangian Coherent Structures (LCS): repelling LCS act as separatrices that separate the state space into safe and unsafe basins, while attracting LCS outline invariant sets and failure traps. The quantitative metrics MBR, ASAS, and TASAS summarize the geometric picture: MBR quantifies how strongly unsafe basins push trajectories away, while ASAS and TASAS quantify the pull of spurious attractors, with TASAS adding dependence on when the attraction acts.

Load-bearing premise

The whole safety-barrier interpretation depends on the finite-time Lyapunov exponent ridges computed from a finite number of sampled trajectories being stable and coinciding with the true safety-relevant boundaries of the closed-loop system.

What would settle it

Run the method on a two-dimensional continuous control task with a known unsafe set and a policy trained to avoid it; compute the FTLE field over increasing integration times and perturb the sampling grid. If the resulting repelling ridges shift by more than the local grid scale as the integration window grows, or if the ridge lies outside the true boundary of the basin of attraction of the unsafe set, the claimed one-to-one correspondence between LCS and safety barriers fails.

Watch this falsifier. Get emailed when new claim-graph text bears on it.

If this is right

  • Policies that look identical under reward can be distinguished by the geometry of their repelling and attracting LCS.
  • Repelling LCS give a principled place to enforce safety: staying inside the ridge-bounded safe region should keep the policy away from known failure modes.
  • Attracting LCS identify trap states and convergence failures that would otherwise only surface in long rollouts.
  • MBR, ASAS, and TASAS provide a numeric safety margin that can be tracked during training or used to compare candidate policies.
  • The local stability guarantees and model-uncertainty extension let the verification claim survive a mismatched dynamics model.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • A natural next step the paper does not take is to use the same geometric skeleton online: repelling ridges could seed barrier-function candidates or safety filters that actively prevent the policy from entering unsafe basins during execution.
  • If the true loop is non-autonomous or the policy is stochastic, the discrete-time autonomous assumption becomes a modeling choice; lifting the state to include the randomness source or time index would be needed to keep the FTLE interpretation valid.
  • The safety-margin metrics could be tested as predictors of empirical failure: policies with lower MBR or higher TASAS should fail more often under bounded perturbations, which is a direct quantitative check of the framework.
  • The attractor view also suggests a training-time use: penalizing the strength of spurious attractors (ASAS/TASAS) during reward shaping may remove failure traps that rewards alone do not discourage.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

5 major / 0 minor

Summary. The submission's abstract claims a new dynamical-systems framework for reinforcement-learning safety verification: the agent and environment are modeled as a discrete-time autonomous dynamical system; Finite-Time Lyapunov Exponents and Lagrangian Coherent Structures are used to identify safety barriers and failure modes; and three metrics (MBR, ASAS, TASAS) are introduced to quantify safety margins and robustness, together with local stability guarantees and control experiments. However, the supplied full text is a completely different paper, arXiv:2508.15589, 'Cosmological pole-skipping, shock waves and quantum chaotic dynamics of de Sitter horizons', by different authors and in a different field. None of the RL safety framework, definitions, theorems, algorithms, or experiments described in the abstract appears anywhere in the body. The manuscript is therefore internally inconsistent: the central claim has no supporting text.

Significance. If the abstract's claims were substantiated, the framework would offer an interpretable, quantitative complement to reward-based RL evaluation, potentially identifying unsafe policies that reward alone misses. That is a worthwhile goal. But as submitted, the paper cannot be assessed on that basis: the body contains no definition of the closed-loop dynamical system, no construction of FTLE/LCS fields for an RL loop, no formal definitions or properties of MBR/ASAS/TASAS, and no experimental results. The only technical content is a holographic pole-skipping analysis whose conclusions are explicitly conditional on a holographic dual existing, and that content is irrelevant to the abstract's RL claims. No machine-checked proofs, reproducible code, or falsifiable RL predictions are present. The central claim is unverifiable, not merely contestable.

major comments (5)
  1. [Entire manuscript (title through references)] The full text is arXiv:2508.15589, 'Cosmological pole-skipping, shock waves and quantum chaotic dynamics of de Sitter horizons', by Ahn, Grozdanov, Jeong, and Pedraza. It has no section corresponding to the abstract's RL safety framework. The agent-environment map, the FTLE/LCS construction, the definitions of MBR/ASAS/TASAS, the local stability guarantee, and the control experiments are all absent. This is load-bearing: every central claim in the abstract lacks a supporting derivation or evaluation.
  2. [Abstract, first paragraph] The framework's premise is that the RL loop can be represented as a discrete-time autonomous dynamical system whose finite-time Lyapunov exponents yield stable LCS that correspond one-to-one with safety-relevant boundaries. This is asserted, not demonstrated. The body does not define the state space, the policy class, the environment class, or the conditions under which FTLE ridges are not finite-time integration artifacts. In particular, stochastic policies or nonstationary environments would violate the autonomous deterministic approximation, and no treatment is given.
  3. [Abstract, second paragraph (metrics)] MBR, ASAS, and TASAS are named as 'formal' quantitative measures of safety margin and robustness, but their definitions are never given. Without definitions, one cannot determine whether they measure what is claimed, whether they depend on simulation length or trajectory count, or whether they were tuned on known failure cases. The claim 'formally measure' is therefore unsupported.
  4. [Abstract, third paragraph (experiments)] The abstract reports experiments in discrete and continuous control environments and states that the framework 'successfully identifies critical flaws' in policies that appear successful by reward. No environments, baselines, hyperparameters, metrics values, figures, tables, or code are provided in the full text. The experimental claim is not reproducible or auditable.
  5. [Body, Section 5 (Conclusions)] The body's own caveat—that an explicit holographic dual for de Sitter space remains an open theoretical challenge—pertains to the physics paper and does not mitigate the absence of the RL content. The only limitation statement in the manuscript is about a different project; the abstract's framework has no stated limitations or validity conditions because it has no body.

Circularity Check

0 steps flagged

No circularity can be exhibited: the supplied full text is a different paper, so the claimed RL derivation chain is absent and no equation-level reduction to inputs exists to audit.

full rationale

The abstract claims a dynamical-systems RL safety framework (FTLE/LCS, MBR/ASAS/TASAS metrics, local stability guarantees, model uncertainty). However, the supplied full text is arXiv:2508.15589, a hep-th paper on pole-skipping in de Sitter horizons, with no definitions of the RL closed-loop system, FTLE/LCS computation, metrics, or experiments. Under the hard rule that circularity must be exhibited by quoting a specific reduction (Eq. X = Eq. Y by construction, or a fitted parameter renamed as a prediction), no such reduction can be identified from the abstract alone: the metric definitions are absent, so one cannot show that MBR/ASAS/TASAS encode the safety conclusions by construction, nor that the LCS safety barriers are defined in terms of the unsafe regions they are said to reveal. The mismatch is an internal inconsistency / unverifiability, not circularity. The physics text contains toy models explicitly constructed to reproduce superluminal/imaginary butterfly velocities (Sec. 4.4: 'we now develop a microscopic toy model capable of reproducing the enhancement of the butterfly velocity...' and 'we now propose a toy model that can capture the purely imaginary butterfly velocity'), but those are explicitly labeled toy models for future exploration and are not used to derive the pole-skipping/shock-wave results; they do not enter the RL abstract's derivation. Because no derivation chain from the abstract is present in the body, I set score 0, the honest non-finding, rather than guessing at hidden circularity.

Axiom & Free-Parameter Ledger

0 free parameters · 4 axioms · 0 invented entities

Because the full text does not correspond to the abstract, this ledger records the presuppositions of the abstract's framework rather than auditable derivations. The load-bearing assumptions are the dynamical-systems modeling of the RL loop and the interpretive link between LCS and safety barriers. No invented physical entities are claimed in the abstract.

axioms (4)
  • domain assumption The combination of an RL agent and its environment forms a discrete-time autonomous dynamical system
    Stated as the foundation of the framework in the abstract; a stochastic policy or non-stationary environment would break autonomy. No justification is present in the available text.
  • domain assumption FTLE ridges computed from finite trajectories correspond to Lagrangian Coherent Structures that act as safety barriers around unsafe regions
    This is the interpretive bridge from dynamical-systems observables to safety semantics claimed in the abstract; it requires finite-time estimates to converge to stable geometric structures, which the missing body would need to demonstrate.
  • domain assumption Attracting LCS reveal convergence properties and unintended trap states
    Interpretive claim in the abstract carrying the same verification burden as the repelling-LCS claim; no evidence is supplied.
  • ad hoc to paper The author-defined metrics MBR, ASAS, and TASAS formally measure a policy's safety margin and robustness
    These metrics are introduced in the abstract without definitions; their validity as formal measures is an assumption of the claimed framework and is unverifiable in this manuscript.

reviewed 2026-08-05 · how reviews work

0 comments
Cite this review

Pith. "Pith review of A Dynamical Systems Framework for Reinforcement Learning Safety and Robustness Verification." pith.science (2026). https://pith.science/paper/I3JFEKNH

@misc{pith2026250815588,
  author       = {Pith},
  title        = {Pith review of: A Dynamical Systems Framework for Reinforcement Learning Safety and Robustness Verification},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/I3JFEKNH}},
  note         = {Machine review of arXiv:2508.15588}
}
Share X Bluesky LinkedIn Reddit HN
read the original abstract

The application of reinforcement learning to safety-critical systems is limited by the lack of formal methods for verifying the robustness and safety of learned policies. This paper introduces a novel framework that addresses this gap by analyzing the combination of an RL agent and its environment as a discrete-time autonomous dynamical system. By leveraging tools from dynamical systems theory, specifically the Finite-Time Lyapunov Exponent (FTLE), we identify and visualize Lagrangian Coherent Structures (LCS) that act as the hidden "skeleton" governing the system's behavior. We demonstrate that repelling LCS function as safety barriers around unsafe regions, while attracting LCS reveal the system's convergence properties and potential failure modes, such as unintended "trap" states. To move beyond qualitative visualization, we introduce a suite of quantitative metrics, Mean Boundary Repulsion (MBR), Aggregated Spurious Attractor Strength (ASAS), and Temporally-Aware Spurious Attractor Strength (TASAS), to formally measure a policy's safety margin and robustness. We further provide a method for deriving local stability guarantees and extend the analysis to handle model uncertainty. Through experiments in both discrete and continuous control environments, we show that this framework provides a comprehensive and interpretable assessment of policy behavior, successfully identifying critical flaws in policies that appear successful based on reward alone.

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

217 extracted references · 10 canonical work pages · 9 internal anchors

  1. [1]

    ’t Hooft,Dimensional reduction in quantum gravity,gr-qc/9310026

    G. ’t Hooft,Dimensional reduction in quantum gravity,gr-qc/9310026

  2. [2]

    Susskind,The World as a hologram,J.Math.Phys.��(1995) 6377–6396, [hep-th/9409089]

    L. Susskind,The World as a hologram,J.Math.Phys.��(1995) 6377–6396, [hep-th/9409089]

  3. [3]

    J. D. Bekenstein,Black holes and the second law,Lett. Nuovo Cim.�(1972) 737–740

  4. [4]

    S. W. Hawking,Particle Creation by Black Holes,Commun. Math. Phys.��(1975) 199–220

  5. [5]

    J. M. Maldacena,The Large N limit of superconformal field theories and supergravity, Adv.Theor.Math.Phys.�(1998) 231–252, [hep-th/9711200]

  6. [6]

    Strominger,The dS / CFT correspondence,JHEP��(2001) 034, [hep-th/0106113]

    A. Strominger,The dS / CFT correspondence,JHEP��(2001) 034, [hep-th/0106113]

  7. [7]

    Weinberg,Cosmology, vol

    S. Weinberg,Cosmology, vol. 41. Oxford University Press, June, 2009. [8]Supernova Search Teamcollaboration, A. G. Riess et al.,Observational evidence from supernovae for an accelerating universe and a cosmological constant,Astron. J.���(1998) 1009–1038, [astro-ph/9805201]. [9]Supernova Cosmology Projectcollaboration, S. Perlmutter et al.,Measurements ofΩ ...

  8. [10]

    Spradlin, A

    M. Spradlin, A. Strominger and A. Volovich,Les Houches lectures on de Sitter space, inLes Houches Summer School: Session 76: Euro Summer School on Unity of Fundamental Physics: Gravity, Gauge Theory and Strings, pp. 423–453, 10, 2001.hep-th/0110007

  9. [11]

    Anninos,De Sitter Musings,Int

    D. Anninos,De Sitter Musings,Int. J. Mod. Phys. A��(2012) 1230013, [1205.3855]

  10. [12]

    D. A. Galante,Modave lectures on de Sitter space & holography,PoS����������(2023) 003, [2306.10141]

  11. [13]

    G. W. Gibbons and S. W. Hawking,Cosmological Event Horizons, Thermodynamics, and Particle Creation,Phys. Rev. D��(1977) 2738–2751

  12. [14]

    Chandrasekaran, R

    V. Chandrasekaran, R. Longo, G. Penington and E. Witten,An algebra of observables for de Sitter space,JHEP��(2023) 082, [2206.10780]

  13. [15]

    Witten,Algebras, regions, and observers.,Proc

    E. Witten,Algebras, regions, and observers.,Proc. Symp. Pure Math.���(2024) 247–276, [2303.02837]

  14. [16]

    Bousso,Adventures in de Sitter space, inWorkshop on Conference on the Future of Theoretical Physics and Cosmology in Honor of Steven Hawking’s 60th Birthday, pp

    R. Bousso,Adventures in de Sitter space, inWorkshop on Conference on the Future of Theoretical Physics and Cosmology in Honor of Steven Hawking’s 60th Birthday, pp. 539–569, 5, 2002.hep-th/0205177

  15. [17]

    Susskind,Entanglement and Chaos in De Sitter Space Holography: An SYK Example, JHAP�(2021) 1–22, [2109.14104]

    L. Susskind,Entanglement and Chaos in De Sitter Space Holography: An SYK Example, JHAP�(2021) 1–22, [2109.14104]

  16. [18]

    Shaghoulian,The central dogma and cosmological horizons,JHEP��(2022) 132, [2110.13210]

    E. Shaghoulian,The central dogma and cosmological horizons,JHEP��(2022) 132, [2110.13210]

  17. [19]

    Shaghoulian and L

    E. Shaghoulian and L. Susskind,Entanglement in De Sitter space,JHEP��(2022) 198, [2201.03603]

  18. [20]

    Bousso,The Holographic principle for general backgrounds,Class

    R. Bousso,The Holographic principle for general backgrounds,Class. Quant. Grav.�� (2000) 997–1005, [hep-th/9911002]. – 46 –

  19. [21]

    Banks and W

    T. Banks and W. Fischler,M theory observables for cosmological space-times, hep-th/0102077

  20. [22]

    Banks, W

    T. Banks, W. Fischler and S. Paban,Recurrent nightmares? Measurement theory in de Sitter space,JHEP��(2002) 062, [hep-th/0210160]

  21. [23]

    Dyson, J

    L. Dyson, J. Lindesay and L. Susskind,Is there really a de Sitter/CFT duality?,JHEP�� (2002) 045, [hep-th/0202163]

  22. [24]

    Banks, B

    T. Banks, B. Fiol and A. Morisse,Towards a quantum theory of de Sitter space,JHEP�� (2006) 004, [hep-th/0609062]

  23. [25]

    Banks and W

    T. Banks and W. Fischler,The holographic spacetime model of cosmology,Int. J. Mod. Phys. D��(2018) 1846005, [1806.01749]

  24. [26]

    Susskind,De Sitter Holography: Fluctuations, Anomalous Symmetry, and Wormholes, Universe�(2021) 464, [2106.03964]

    L. Susskind,De Sitter Holography: Fluctuations, Anomalous Symmetry, and Wormholes, Universe�(2021) 464, [2106.03964]

  25. [27]

    X. Dong, E. Silverstein and G. Torroba,De Sitter Holography and Entanglement Entropy, JHEP��(2018) 050, [1804.08623]

  26. [28]

    Lin and L

    H. Lin and L. Susskind,Infinite Temperature’s Not So Hot,2206.01083

  27. [29]

    Baiguera, R

    S. Baiguera, R. Berman, S. Chapman and R. C. Myers,The cosmological switchback effect, JHEP��(2023) 162, [2304.15008]

  28. [30]

    Lewkowycz, J

    A. Lewkowycz, J. Liu, E. Silverstein and G. Torroba,� �and EE, with implications for (A)dS subregion encodings,JHEP��(2020) 152, [1909.13808]

  29. [31]

    Coleman, E

    E. Coleman, E. A. Mazenc, V. Shyam, E. Silverstein, R. M. Soni, G. Torroba et al.,De Sitter microstates from T �+Λ � and the Hawking-Page transition,JHEP��(2022) 140, [2110.14670]

  30. [32]

    Anninos, F

    D. Anninos, F. Denef, Y. T. A. Law and Z. Sun,Quantum de Sitter horizon entropy from quasicanonical bulk, edge, sphere and topological string partition functions,JHEP��(2022) 088, [2009.12464]

  31. [33]

    Silverstein,Black hole to cosmic horizon microstates in string/M theory: timelike boundaries and internal averaging,JHEP��(2023) 160, [2212.00588]

    E. Silverstein,Black hole to cosmic horizon microstates in string/M theory: timelike boundaries and internal averaging,JHEP��(2023) 160, [2212.00588]

  32. [34]

    Freivogel, V

    B. Freivogel, V. E. Hubeny, A. Maloney, R. C. Myers, M. Rangamani and S. Shenker, Inflation in AdS/CFT,JHEP��(2006) 007, [hep-th/0510046]

  33. [35]

    D. A. Lowe and S. Roy,Punctuated eternal inflation via AdS/CFT,Phys. Rev. D��(2010) 063508, [1004.1402]

  34. [36]

    Fischetti, D

    S. Fischetti, D. Marolf and A. C. Wall,A paucity of bulk entangling surfaces: AdS wormholes with de Sitter interiors,Class. Quant. Grav.��(2015) 065011, [1409.6754]

  35. [37]

    Anninos and D

    D. Anninos and D. M. Hofman,Infrared Realization of dS � in AdS�,Class. Quant. Grav.�� (2018) 085003, [1703.04622]

  36. [38]

    Anninos, D

    D. Anninos, D. A. Galante and D. M. Hofman,De Sitter horizons & holographic liquids, JHEP��(2019) 038, [1811.08153]

  37. [39]

    Mirbabayi,Uptunneling to de Sitter,JHEP��(2020) 070, [2003.05460]

    M. Mirbabayi,Uptunneling to de Sitter,JHEP��(2020) 070, [2003.05460]

  38. [40]

    Anninos and D

    D. Anninos and D. A. Galante,Constructing AdS � flow geometries,JHEP��(2021) 045, [2011.01944]. – 47 –

  39. [41]

    Grozdanov, K

    S. Grozdanov, K. Schalm and V. Scopelliti,Black hole scrambling from hydrodynamics, Phys. Rev. Lett.���(2018) 231601, [1710.00921]

  40. [42]

    Blake, H

    M. Blake, H. Lee and H. Liu,A quantum hydrodynamical description for scrambling and many-body chaos,JHEP��(2018) 127, [1801.00010]

  41. [43]

    Blake, R

    M. Blake, R. A. Davison, S. Grozdanov and H. Liu,Many-body chaos and energy dynamics in holography,JHEP��(2018) 035, [1809.01169]

  42. [44]

    W. Z. Chua, T. Hartman and W. W. Weng,Replica manifolds, pole skipping, and the butterfly effect,2504.08139

  43. [45]

    Grozdanov, P

    S. Grozdanov, P. K. Kovtun, A. O. Starinets and P. Tadi´ c,The complex life of hydrodynamic modes,JHEP��(2019) 097, [1904.12862]

  44. [46]

    Blake, R

    M. Blake, R. A. Davison and D. Vegh,Horizon constraints on holographic Green ’s functions,JHEP��(2020) 077, [1904.12883]

  45. [47]

    Amado, C

    I. Amado, C. Hoyos-Badajoz, K. Landsteiner and S. Montero,Hydrodynamics and beyond in the strongly coupled N=4 plasma,JHEP��(2008) 133, [0805.2570]

  46. [48]

    Blake and H

    M. Blake and H. Liu,On systems of maximal quantum chaos,JHEP��(2021) 229, [2102.11294]

  47. [49]

    Jeong, K.-Y

    H.-S. Jeong, K.-Y. Kim and Y.-W. Sun,Bound of diffusion constants from pole-skipping points: spontaneous symmetry breaking and magnetic field,JHEP��(2021) 105, [2104.13084]

  48. [50]

    C. Choi, M. Mezei and G. S´ arosi,Pole skipping away from maximal chaos,2010.08558

  49. [51]

    F. M. Haehl and M. Rozali,Effective Field Theory for Chaotic CFTs,JHEP��(2018) 118, [1808.02898]

  50. [52]

    Jensen,Scrambling in nearly thermalized states at large central charge,1906.05852

    K. Jensen,Scrambling in nearly thermalized states at large central charge,1906.05852

  51. [53]

    S. Das, B. Ezhuthachan and A. Kundu,Real time dynamics from low point correlators in 2d BCFT,JHEP��(2019) 141, [1907.08763]

  52. [54]

    F. M. Haehl, W. Reeves and M. Rozali,Reparametrization modes, shadow operators, and quantum chaos in higher-dimensional CFTs,JHEP��(2019) 102, [1909.05847]

  53. [55]

    D. M. Ramirez,Chaos and pole skipping in CFT �,JHEP��(2021) 006, [2009.00500]

  54. [56]

    Natsuume and T

    M. Natsuume and T. Okamura,Nonuniqueness of scattering amplitudes at special points, Phys. Rev. D���(2021) 126007, [2108.07832]

  55. [57]

    Natsuume and T

    M. Natsuume and T. Okamura,Nonuniqueness of Green ’s functions at special points,JHEP ��(2019) 139, [1905.12015]

  56. [58]

    Natsuume and T

    M. Natsuume and T. Okamura,Holographic chaos, pole-skipping, and regularity,PTEP ����(2020) 013B07, [1905.12014]

  57. [59]

    Natsuume and T

    M. Natsuume and T. Okamura,Pole-skipping with finite-coupling corrections,Phys. Rev. D ���(2019) 126012, [1909.09168]

  58. [60]

    Ceplak, K

    N. Ceplak, K. Ramdial and D. Vegh,Fermionic pole-skipping in holography,JHEP�� (2020) 203, [1910.02975]

  59. [61]

    Y. Ahn, V. Jahnke, H.-S. Jeong and K.-Y. Kim,Scrambling in Hyperbolic Black Holes: shock waves and pole-skipping,JHEP��(2019) 257, [1907.08030]. – 48 –

  60. [62]

    Y. Ahn, V. Jahnke, H.-S. Jeong, K.-Y. Kim, K.-S. Lee and M. Nishida,Pole-skipping of scalar and vector fields in hyperbolic space: conformal blocks and holography,JHEP�� (2020) 111, [2006.00974]

  61. [63]

    Abbasi and S

    N. Abbasi and S. Tahery,Complexified quasinormal modes and the pole-skipping in a holographic system at finite chemical potential,JHEP��(2020) 076, [2007.10024]

  62. [64]

    Liu and A

    Y. Liu and A. Raju,Quantum Chaos in Topologically Massive Gravity,JHEP��(2020) 027, [2005.08508]

  63. [65]

    Y. Ahn, V. Jahnke, H.-S. Jeong, K.-Y. Kim, K.-S. Lee and M. Nishida,Classifying pole-skipping points,JHEP��(2021) 175, [2010.16166]

  64. [66]

    Natsuume and T

    M. Natsuume and T. Okamura,Pole-skipping and zero temperature,Phys. Rev. D��� (2021) 066017, [2011.10093]

  65. [67]

    Sil,Pole skipping and chaos in anisotropic plasma: a holographic study,JHEP��(2021) 232, [2012.07710]

    K. Sil,Pole skipping and chaos in anisotropic plasma: a holographic study,JHEP��(2021) 232, [2012.07710]

  66. [68]

    Ceplak and D

    N. Ceplak and D. Vegh,Pole-skipping and Rarita-Schwinger fields,Phys. Rev. D��� (2021) 106009, [2101.01490]

  67. [69]

    Blake and R

    M. Blake and R. A. Davison,Chaos and pole-skipping in rotating black holes,JHEP�� (2022) 013, [2111.11093]

  68. [70]

    D. S. Ageev,Butterfly velocity and chaos suppression in de Sitter space,Theor. Math. Phys. ���(2025) 863–871, [2105.02258]

  69. [71]

    Wang and W.-B

    Y.-T. Wang and W.-B. Pan,Pole-skipping of holographic correlators: aspects of gauge symmetry and generalizations,JHEP��(2023) 174, [2209.04296]

  70. [72]

    Wang and Z.-Y

    D. Wang and Z.-Y. Wang,Pole Skipping in Holographic Theories with Bosonic Fields, Phys. Rev. Lett.���(2022) 231603, [2208.01047]

  71. [73]

    M. A. G. Amano, M. Blake, C. Cartwright, M. Kaminski and A. P. Thompson,Chaos and pole-skipping in a simply spinning plasma,JHEP��(2023) 253, [2211.00016]

  72. [74]

    Grozdanov and M

    S. Grozdanov and M. Vrbica,Pole-skipping of gravitational waves in the backgrounds of four-dimensional massive black holes,Eur. Phys. J. C��(2023) 1103, [2303.15921]

  73. [75]

    Jeong, C.-W

    H.-S. Jeong, C.-W. Ji and K.-Y. Kim,Pole-skipping in rotating BTZ black holes,JHEP�� (2023) 139, [2306.14805]

  74. [76]

    Grozdanov, T

    S. Grozdanov, T. Lemut and J. F. Pedraza,Reconstruction of the quasinormal spectrum from pole skipping,Phys. Rev. D���(2023) L101901, [2308.01371]

  75. [77]

    Jeong,Thermal diffusivity and pole-skipping in the incoherent semilocally critical IR, Phys

    H.-S. Jeong,Thermal diffusivity and pole-skipping in the incoherent semilocally critical IR, Phys. Lett. B���(2025) 139591, [2309.13412]

  76. [78]

    Abbasi and K

    N. Abbasi and K. Landsteiner,Pole-skipping as order parameter to probe a quantum critical point,JHEP��(2023) 169, [2307.16716]

  77. [79]

    Pan, Y.-W

    W.-B. Pan, Y.-W. Sun and Y.-T. Wang,Pole-skipping for massive fields and the Stueckelberg formalism,JHEP��(2024) 256, [2404.17354]

  78. [80]

    Y. Ahn, V. Jahnke, H.-S. Jeong, C.-W. Ji, K.-Y. Kim and M. Nishida,On pole-skipping with gauge-invariant variables in holographic axion theories,JHEP��(2024) 020, [2402.12951]

  79. [81]

    Grozdanov and M

    S. Grozdanov and M. Vrbica,Duality Constraints on Thermal Spectra of 3D Conformal – 49 – Field Theories and 4D Quasinormal Modes,Phys. Rev. Lett.���(2024) 211601, [2406.19790]

  80. [82]

    Grozdanov and M

    S. Grozdanov and M. Vrbica,Duality and four-dimensional black holes: gravitational waves, algebraically special solutions, pole skipping, and the spectral duality relation in holographic thermal CFTs,2505.14229

Showing first 80 references.

This paper was first reviewed by deepseek-v4-flash on August 5, 2026.