Pith. sign in

REVIEW 3 major objections 5 minor 41 references

Combining Machine Learning with Recurrence Analysis for resonance detection

T0 review · 3 major / 5 minor · reviewed 2026-08-10 · deepseek-v4-flash

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

Pith's one-line read Recurrence analysis plus a standard-map-trained LSTM can locate resonances even in a four-dimensional map where conventional visualization tools do not exist.

desk verdict Promising 2D resonance detector whose "regardless of dimensionality" claim outruns the evidence; the 4D success is one embedded-only configuration with no independent ground truth. read the letter →

arxiv 2412.19683 v1 pith:PYYLTO6F submitted 2024-12-27 gr-qc cs.LG

Ondřej Zelenka , Ondřej Kopáček , Georgios Lukes-Gerakopoulos This is my paper · ORCID
classification gr-qccs.LG MSC 37M1068T0783C57
keywords recurrencequantificationanalysisresonancedetectionlongshort-termmemorystandardmaptime-delayembeddingJohannsen-Psaltisspacetimeextrememassratioinspirals4D
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 proposes that recurrence plots contain enough information about orbital resonances, the periodic islands in weakly non-integrable systems, that a machine-learning network can learn to spot them from the patterns alone without needing the system's equations or frequencies. To test this, the authors train a bidirectional LSTM on sweeps of initial conditions of the standard map, using 70 recurrence-quantification indicators per orbit, and then apply it to other systems. The network detects resonances in the de Vogeleare map and in the Poincare section of a test particle in the deformed Kerr spacetime known as the Johannsen-Psaltis metric, where the target is the $\omega_r/\omega_\theta = 2/3$ resonance relevant for extreme-mass-ratio inspiral modeling. The key demonstration is the 4D map: a plain network trained on full phase-space data fails, but a network trained on time-delay-embedded data and applied to embedded 4D data produces peaks at the true resonance locations. The paper concludes that recurrence quantifiers encode resonance structure even where conventional rotation-number and visual methods are unavailable.

What carries the argument

The load-bearing object is the 70-dimensional recurrence-quantification input vector: for each orbit, seven RQA measures, $RR$, $DET$, $LAM$, $L$, $L_{\rm entr}$, $DIV$, and $V_{\max}$, computed at ten recurrence thresholds $\epsilon$. These vectors, arranged as a sweep across initial conditions, are fed into a bidirectional LSTM whose cells are stacked along the initial-condition direction rather than along time. Embedding is the second mechanism: when only a scalar time series is available, time-delay embedding with dimension $m$ and delay $\tau$ reconstructs the phase space, and the paper shows that matching the embedding between training and test data is what makes the 4D detection work.

What would settle it

Compute recurrence-quantification features for the 4D map under several embedding dimensions and delays, and compare the embedded network's output peaks with the APLE resonance map: if the peak locations shift whenever $m$ or $\tau$ changes, the detections are artifacts of the embedding rather than signatures of the resonances.

Watch

Extended reading notes

Core claim

The paper's central claim is that recurrence quantifiers, the statistical measures of the diagonal and vertical line structures in a recurrence plot, carry detectable imprints of resonant islands, and that an LSTM trained once on the standard map can serve as a resonance detector for other dynamical systems, including a four-dimensional map where rotation-number and visual methods are unavailable. The basic network, trained on full phase-space coordinates, cleanly localizes resonances in two two-dimensional maps and in the Poincare section of a test particle orbiting a deformed Kerr black hole. In the 4D map, the same architecture detects resonances only when both training and test data pass through time-delay embedding: the embedded network applied to embedded data produces output peaks that align with the resonance geography charted by the APLE indicator, while the two mismatched combinations show nothing. The paper thus claims a proof of concept for a dimensionality-agnostic, automated resonance-detection pipeline, with the preprocessing convention required to match between training and application.

Load-bearing premise

The load-bearing premise is that recurrence-quantifier patterns are similar enough across different dynamical systems, when computed with the same thresholds and the same embedding choice, for a network trained on the standard map to recognize resonances in other systems; the 4D results show that this similarity holds only for matching embedding conventions.

Editorial extensions

If this is right

  • Resonance localization can be automated for two-dimensional systems without manual rotation-number curves: the network trained on the standard map detects islands in the de Vogeleare map and in the black-hole Poincare section.
  • For higher-dimensional systems, the practical recipe is to embed the available scalar observables and apply a network trained on embedded data; the 4D map test shows that this combination recovers the known resonance geography.
  • In extreme-mass-ratio inspiral modeling, the method offers a way to identify extended resonances such as the $\omega_r/\omega_\theta = 2/3$ resonance in the Johannsen-Psaltis spacetime, helping to decide which perturbation parameters matter for waveform modeling.
  • Because recurrence quantifiers are computable from any trajectory and do not require a Poincare section of a specific dimension, the same pipeline can in principle be applied to continuous-time systems by taking Poincare sections.
  • The failure of the cross-combinations in the 4D experiment implies that preprocessing must be considered part of the method: a detector trained on one embedding convention is not assumed to transfer to data prepared differently.

Reading between the lines

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

  • An implication the authors leave implicit is that the 4D success may owe more to embedding-induced features of the recurrence statistics than to a truly system-independent resonance signature; a decisive check would be to train on embedded standard-map data and test on a different embedded 4D system whose resonances are independently mapped.
  • If the preprocessing-match requirement generalizes, then each new application domain will need its own embedding calibration, which weakens the 'train once, use anywhere' reading of the method.
  • A testable extension for EMRI work is to apply the embedded network to synthetic gravitational-wave snapshots of inspiraling small bodies crossing a resonance and ask whether the output peaks track the crossing time; if they do, the method could become a resonance-crossing detector in waveform data.
  • The hand-labeling protocol, which marks an island as resonant only if at least two neighboring initial conditions fall in it, shapes the learned target; changing the sweep spacing or labeling single-point islands could change which resonances are detected and is worth quantifying.
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

3 major / 5 minor

Summary. The paper proposes a resonance detection method that combines recurrence quantification analysis (RQA) with a bidirectional LSTM network. The network is trained on hand-labeled orbits of the standard map and then applied to other systems: the de Vogeleare map, geodesic motion in the Johannsen-Psaltis spacetime, and a 4D symplectic map. The authors report clear resonance detection in the two 2D test cases and a partial success in the 4D case when both training and test RQA features are computed from time-delay embedded data. The stated motivation is detecting resonances in EMRI systems, where higher-dimensional phase spaces prevent the use of rotation-number methods.

Significance. Resonance detection in higher-dimensional near-integrable systems is an open problem with direct relevance to EMRI waveform modeling, and the proposed RQA+ML pipeline is a plausible and potentially useful approach. The 2D demonstrations, especially the transfer from the standard map to the de Vogeleare map and to a black-hole spacetime, are visually convincing and illustrate a practical workflow. The paper also gives a clear description of the network architecture, training data, and recurrence-analysis parameters, and it builds on standard, widely used tools. However, the central claim that recurrence quantifiers carry imprints of resonant behavior 'regardless of the system's dimensionality' rests on a single configuration in the 4D map and lacks an independent quantitative ground truth for resonance identity in that test case. As a result, the significance for EMRI applications, where more than two degrees of freedom matter, is not yet firmly established.

major comments (3)
  1. [Section III B and Section IV] The claim in the abstract and conclusion that recurrence quantifiers carry imprints of resonant behavior 'regardless of the system's dimensionality' is not supported by the 4D experiment as presented. The authors state in Section III B that only the embedded network applied to embedded 4D map data produces recognizable resonance peaks; the basic network on full-state 4D data fails (Fig. 9), and the cross combinations also fail. This makes the higher-dimensional success conditional on a specific embedding preprocessing pipeline, not a demonstration of a dimensionality-independent RQA signature. The conclusion 'we conclude that this method is effective' therefore overstates the strength of the 4D evidence.
  2. [Section II F 3 and Section III B] There is no independent ground truth for resonance identity in the 4D map. The APLE indicator defined in Eq. (20) distinguishes regular from chaotic trajectories but does not label a regular torus as resonant or non-resonant. The statement in Section III B that 'the network is reacting to resonances in their true locations' is based on visual comparison with the APLE geography in Fig. 5, not on a quantitative resonance classification. Without an independent method to identify resonant tori in the 4D map, the peaks in Fig. 10 could be artifacts of the embedding or of the recurrence thresholds rather than universal resonance signatures.
  3. [Section II D and Section III B] The RQA features are computed with fixed absolute thresholds epsilon = 0.001, ..., 1 after rescaling the phase space to [0, 2*pi]^n, rather than with a fixed recurrence rate. Because the distribution of pairwise distances changes with dimensionality and after time-delay embedding, the 4D/embedded feature vectors may lie outside the feature manifold of the standard-map training data. The authors themselves show in Fig. 3 that RQA values depend strongly on the choice of epsilon. The paper should either demonstrate that the results are robust to threshold normalization, for example by using a fixed recurrence rate, or justify why absolute thresholds transfer across systems and dimensions.
minor comments (5)
  1. [Abstract] The sentence 'Although the tool that we are presenting here can be used is quite generic' is ungrammatical and should be revised.
  2. [Section II B, Eq. (5b)] In the definition of P(epsilon, v), the factors (1 - R_{i,j}) and R_{i,j+k} omit the explicit epsilon dependence that is written in Eq. (5a); the notation should be made consistent.
  3. [Fig. 8 caption] The caption refers to 'the 1/3 resonance is highlighted in red', while the text in Section II F 2 and Section III A 2 identifies the resonance as omega_r/omega_theta = 2/3. This inconsistency should be corrected.
  4. [Section II D] There is a typo in 'thetorch.nn.LSTM module'; it should read 'the torch.nn.LSTM module'.
  5. [Appendix A] The statement that a code release is 'under preparation' limits reproducibility; the authors should consider providing the code with the submission or at least a complete list of hyperparameters and data-generation scripts.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the LSTM is trained on externally hand-labeled standard-map resonances and tested on independent systems.

full rationale

The paper's derivation chain is a standard supervised-learning workflow: manual labels of resonant islands in the standard map (Sec. II E), RQA features computed from trajectories (Sec. II B), and an LSTM trained on those input-label pairs (Sec. II D). Predictions on the de Vogeleare map, the Johannsen-Psaltis Poincaré section, and the 4D map use systems and trajectories not present in training, so the test outputs are not fitted to the target data by construction. The recurrence thresholds are fixed a priori (epsilon = 0.001, ..., 1), not optimized against the test labels. The self-citations [2, 16, 17, 24, 25] provide background context, the JP metric parameters, and the 4D map definition; none is invoked as an unverified uniqueness theorem that forces the conclusion. The limitation disclosed in Sec. III B, that only the embedded-network/embedded-data combination produces recognizable peaks in the 4D map, is a generalization and validation concern (possible embedding or threshold artifacts, no quantitative ground truth for resonance identity in 4D), not a circular reduction: the network outputs are not defined in terms of the APLE curves or the recurrence quantifiers used to evaluate them. No fitted parameter is renamed as a prediction, and no derived quantity is defined in terms of the labels it is said to predict. Therefore no circularity step is present.

Assumptions & free parameters 6 free parameters · 4 assumptions · 0 invented entities

The paper introduces no new physical entities. The free parameters are mostly standard RQA settings and network hyperparameters. The key domain assumptions are that the standard map is representative and that RQA features transfer across systems. The embedding parameters and LSTM hyperparameters are chosen via standard heuristics, not fitted to the target labels, but the hand-labeled training set introduces a subjective element.

free parameters (6)
  • recurrence thresholds epsilon set = [0.001, 0.002, 0.005, 0.01, 0.02, 0.05, 0.1, 0.2, 0.5, 1.0]
    Chosen to cover recurrence rates ~1-20% as stated in Sec II B; not fitted to resonance labels.
  • RQA minimal line lengths lmin, vmin = 2
    Standard default values from recurrence analysis literature (cited in Sec II B).
  • embedding dimension m and delay tau for training = m=2, tau=1
    Determined using false nearest neighbor and mutual information on representative trajectories (Sec III B); not fitted to labels.
  • LSTM depth and dropout rate = depth=2, dropout=0.5
    Selected based on lowest validation loss on the standard map validation set (Sec III A).
  • trajectory length = 10000 points
    Fixed for all orbits; no sensitivity analysis provided.
  • hand-labeling island criterion = at least two neighboring initial conditions in island
    Defines the training labels; affects which islands are considered resonances (Sec II E).
assumptions (4)
  • domain assumption Standard map dynamics capture generic resonance behavior
    Authors state the standard map is a kicked pendulum and that pendulum dynamics encapsulates resonance dynamics (Sec II E).
  • standard math Takens embedding theorem guarantees faithful phase space reconstruction
    Invoked in Sec II C to justify time-delay embedding, citing Takens [18].
  • domain assumption RQA measures are sensitive to resonance identity
    The paper claims and shows in Fig 3 that RQA quantifiers respond to resonances, but the relationship is not proven and depends on thresholds.
  • ad hoc to paper LSTM can learn a mapping from RQA features to resonance membership
    No theoretical guarantee; the success is purely empirical and only demonstrated on specific systems.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Combining Machine Learning with Recurrence Analysis for resonance detection." pith.science (2026). https://pith.science/paper/PYYLTO6F

@misc{pith2026241219683,
  author       = {Pith},
  title        = {Pith review of: Combining Machine Learning with Recurrence Analysis for resonance detection},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/PYYLTO6F}},
  note         = {Machine review of arXiv:2412.19683}
}
read the original abstract

The width of a resonance in a nearly integrable system, i.e. in a non-integrable system where chaotic motion is still not prominent, can tell us how a perturbation parameter is driving the system away from integrability. Although the tool that we are presenting here can be used is quite generic and can be used in a variety of systems, our particular interest lies in binary compact object systems known as extreme mass ratio inspirals (EMRIs). In an EMRI a lighter compact object, like a black hole or a neutron star, inspirals into a supermassive black hole due to gravitational radiation reaction. During this inspiral the lighter object crosses resonances, which are still not very well modeled. Measuring the width of resonances in EMRI models allows us to estimate the importance of each perturbation parameter able to drive the system away from resonances and decide whether its impact should be included in EMRI waveform modeling or not. To tackle this issue in our study we show first that recurrence quantifiers of orbits carry imprints of resonant behavior, regardless of the system's dimensionality. As a next step, we apply a long short-term memory machine learning architecture to automate the resonance detection procedure. Our analysis is developed on a simple standard map and gradually we extend it to more complicated systems until finally we employ it in a generic deformed Kerr spacetime known in the literature as the Johannsen-Psaltis spacetime.

Figures

Figures reproduced from arXiv: 2412.19683 by the authors.

Figure 1
Figure 1. Phase portrait of the standard map (see Eqs. [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Examples of recurrence plots of trajectories of the standard map with [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Example of RQA curves of the standard map at [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figures from the paper (6 more)
Figure 4
Figure 4. Figure 4: LSTM network architecture. Taken from [20]. is defined as ⃗xi = [PITH_FULL_IMAGE:figures/full_fig_p006_4.png]
Figure 6
Figure 6. Figure 6: Evolution of the training (blue) and validation [PITH_FULL_IMAGE:figures/full_fig_p008_6.png]
Figure 7
Figure 7. Figure 7: Results on the de Vogeleare map. The Poincar´e section is shown in black, displaying several prominent resonances. [PITH_FULL_IMAGE:figures/full_fig_p009_7.png]
Figure 8
Figure 8. Figure 8: Results of the basic network on the Poincar´e section of geodesic motion in the Johannsen-Psaltis spacetime super [PITH_FULL_IMAGE:figures/full_fig_p010_8.png]
Figure 9
Figure 9. Figure 9: Results of the basic network on the 4D map. Each panel corresponds to 4001 initial conditions with [PITH_FULL_IMAGE:figures/full_fig_p010_9.png]
Figure 10
Figure 10. Figure 10: Results of the embedded network on the 4D map. Each panel corresponds to 4001 initial conditions with [PITH_FULL_IMAGE:figures/full_fig_p011_10.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

41 extracted references · 32 canonical work pages

  1. [1]

    With d degrees of freedom, it consists of • generalized coordinates qi, i= 1,

    Hamiltonian systems Many systems in physics can be written in the form of a Hamiltonian system. With d degrees of freedom, it consists of • generalized coordinates qi, i= 1, . . . , d, • canonical momenta pi, i= 1, . . . , d, • a Hamiltonian function H qi, pi . The equations of motion then have the form dqi dt = ∂H ∂pi , (1a) dpi dt = − ∂H ∂q i . (1b) In ...

    work page

  2. [2]

    Dynamical maps The structure of a d = 2 Hamiltonian system’s phase space is commonly visualized using a Poincar´ e section. A surface of section is defined in the phase space such that the Hamiltonian flow is not tangent at any point of the surface, and successive intersections of the orbit and the surface of section are recorded and form the Poincar´ e s...

    work page

  3. [3]

    The initial conditions are taken along the pr = 0 line on the Poincar´ e section in r ∈ [6.36M, 6.43M ] spaced at 0.001M

    For integrals of motion we choose the values E = 0.95µ, Lz = 2 .85µM . The initial conditions are taken along the pr = 0 line on the Poincar´ e section in r ∈ [6.36M, 6.43M ] spaced at 0.001M . This sweep of initial conditions includes the ωr/ωθ = 2/3 orbital resonance

    work page

  4. [4]

    Dynamical mapping: de Vogeleare map For the simpler test data, we use the de Vogeleare map [23], defined as xn+1 = −yn + Kx n + x2 n , (15a) yn+1 = xn − Kx n+1 − x2 n+1 , (15b) with K ∈ R a perturbation parameter

    work page

  5. [5]

    Knowledge of their locations will allow us to properly model the qual- itatively distinct behavior of the passage through these resonances

    Poincar´ e section: motion around a black hole One of the aims of the method developed in this pa- per is the search for extended resonances in the motion of a test particle around a black hole. Knowledge of their locations will allow us to properly model the qual- itatively distinct behavior of the passage through these resonances. We consider a particle...

    work page

  6. [6]

    Lukes-Gerakopoulos and V

    G. Lukes-Gerakopoulos and V. Witzany, in Handbook of Gravitational Wave Astronomy , edited by C. Bambi, S. Katsanevas, and K. D. Kokkotas (2021) p. 42

    work page 2021

  7. [7]

    However, already in a system with 3 degrees of freedom, the Poincar´ e section becomes 4-dimensional and these simple methods are no longer applicable

    Higher-dimensional example: 4D map In a Hamiltonian system with 2 degrees of freedom, alternative methods exist which can detect resonances, such as the rotation curve and its estimation using the 2-dimensional Poincar´ e section. However, already in a system with 3 degrees of freedom, the Poincar´ e section becomes 4-dimensional and these simple methods ...

    work page

  8. [8]

    de Vogeleare map To demonstrate the effectiveness and robustness of the resonance detection network, we apply it to test data generated by a different system than the standard map. Fig. 7 shows the results of testing on the de Vo- geleare map data. The perturbation parameter is set to K = 0 .56 and 2001 initial conditions are used with linear spacing in x...

    work page 2001

Show all 41 references
  1. [9]

    with the machine learning approach. In the current paper, we first describe the method and demonstrate its abilities on simple systems (standard and de Voge- arXiv:2412.19683v1 [gr-qc] 27 Dec 2024 2 0 1 2 3 4 5 6 x −3 −2 −1 0 1 2 3 y Figure 1. Phase portrait of the standard ma...

    2024 arXiv

  2. [10]

    8 shows the result of the basic network applied to the Poincar´ e section of a test particle following a geodesic in the Johannsen-Psaltis spacetime metric (see Sec

    Motion around a black hole Fig. 8 shows the result of the basic network applied to the Poincar´ e section of a test particle following a geodesic in the Johannsen-Psaltis spacetime metric (see Sec. II F 2). The section contains a single resonance ωr/ωθ = 2 /3, which is detecte...

    2001

  3. [11]

    4D map Finally, let us focus on the 4-dimensional map. Fig. 9 shows the output of the basic network over the two curves described in Sec. II F 3. For better orientation, the corresponding APLE curves are also shown. In this case, no resonances are clearly recognizable in the n...

  4. [12]

    S. H. Strogatz, Nonlinear Dynamics and Chaos , 2nd ed. (CRC Press, London, England, 2019)

    2019

  5. [13]

    Mukherjee, O

    S. Mukherjee, O. Kop´ aˇ cek, and G. Lukes-Gerakopoulos, Phys. Rev. D 107, 064005 (2023), arXiv:2206.10302 [gr- qc]

    2023 arXiv

  6. [14]

    Zelenka, G

    O. Zelenka, G. Lukes-Gerakopoulos, V. Witzany, and O. Kop´ aˇ cek, Phys. Rev. D101, 024037 (2020)

    2020

  7. [15]

    Amaro-Seoane, Living Reviews in Relativity 21, 4 (2018), arXiv:1205.5240 [astro-ph.CO]

    P. Amaro-Seoane, Living Reviews in Relativity 21, 4 (2018), arXiv:1205.5240 [astro-ph.CO]. 12

    2018 arXiv

  8. [16]

    Amaro-Seoane, J

    P. Amaro-Seoane, J. Andrews, M. Arca Sedda, A. Askar, Q. Baghi, R. Balasov, I. Bartos, S. S. Bav- era, J. Bellovary, C. P. L. Berry, E. Berti, S. Bianchi, L. Blecha, S. Blondin, T. Bogdanovi´ c,et al., Living Re- views in Relativity 26, 2 (2023), arXiv:2203.06016 [gr- qc]

    2023 arXiv

  9. [17]

    Sukov´ a and O

    P. Sukov´ a and O. Semer´ ak, MNRAS436, 978 (2013), arXiv:1308.4306 [gr-qc]

    2013 arXiv

  10. [18]

    M. D. Hartl, Phys. Rev. D 67, 024005 (2003), arXiv:gr- qc/0210042 [gr-qc]

    2003

  11. [19]

    Marwan, M

    N. Marwan, M. Carmen Romano, M. Thiel, and J. Kurths, Physics Reports 438, 237 (2007)

    2007

  12. [20]

    Johannsen and D

    T. Johannsen and D. Psaltis, Phys. Rev. D 83, 124015 (2011), arXiv:1105.3191 [gr-qc]

    2011 arXiv

  13. [21]

    A. N. Kolmogorov, Doklady Akademii Nauk SSSR 98, 527 (1954)

    1954

  14. [22]

    V. I. Arnold, Russian Mathematical Surveys 18, 9 (1963)

    1963

  15. [23]

    Moser, Nachrichten der Akademie der Wissenschaften in G¨ ottingen

    J. Moser, Nachrichten der Akademie der Wissenschaften in G¨ ottingen. II. Mathematisch-Physikalische Klasse , 1 (1962)

    1962

  16. [24]

    G. D. Birkhoff, Transactions of the American Mathe- matical Society 14, 14 (1913)

    1913

  17. [25]

    Marwan, International Journal of Bifurcation and Chaos 21, 1003 (2011), arXiv:1007.2215 [nlin.CD]

    N. Marwan, International Journal of Bifurcation and Chaos 21, 1003 (2011), arXiv:1007.2215 [nlin.CD]

    2011 arXiv

  18. [26]

    Lukes-Gerakopoulos and O

    G. Lukes-Gerakopoulos and O. Kop´ aˇ cek, International Journal of Modern Physics D 27, 1850010 (2018), arXiv:1709.08446 [gr-qc]

    2018 arXiv

  19. [27]

    Kop´ aˇ cek, V

    O. Kop´ aˇ cek, V. Karas, J. Kov´ aˇ r, and Z. Stuchl ´ ık, ApJ 722, 1240 (2010), arXiv:1008.4650 [astro-ph.HE]

    2010 arXiv

  20. [28]

    Takens, in Dynamical Systems and Turbulence, War- wick 1980, edited by D

    F. Takens, in Dynamical Systems and Turbulence, War- wick 1980, edited by D. Rand and L.-S. Young (Springer Berlin Heidelberg, Berlin, Heidelberg, 1981) pp. 366– 381

    1980

  21. [29]

    Hochreiter and J

    S. Hochreiter and J. Schmidhuber, Neural Computation 9, 1735 (1997), https://direct.mit.edu/neco/article- pdf/9/8/1735/813796/neco.1997.9.8.1735.pdf

    1997

  22. [30]

    Alfaidi and C

    R. Alfaidi and C. Messenger, (2024), arXiv:2402.04589 [gr-qc]

    2024 arXiv

  23. [31]

    B. V. Chirikov, Research concerning the theory of non- linear resonance and stochasticity (1971)

    1971

  24. [32]

    Morbidelli, Modern celestial mechanics: aspects of solar system dynamics , 1st ed

    A. Morbidelli, Modern celestial mechanics: aspects of solar system dynamics , 1st ed. (CRC Press, 2002)

    2002

  25. [33]

    Reichl, The Transition to Chaos: Conservative Clas- sical Systems and Quantum Manifestations , Institute for Nonlinear Science (Springer, 2004)

    L. Reichl, The Transition to Chaos: Conservative Clas- sical Systems and Quantum Manifestations , Institute for Nonlinear Science (Springer, 2004)

    2004

  26. [34]

    Zelenka and G

    O. Zelenka and G. Lukes-Gerakopoulos, in Workshop on Black Holes and Neutron Stars (2017) arXiv:1711.02442 [gr-qc]

    2017 arXiv

  27. [35]

    Lukes-Gerakopoulos, N

    G. Lukes-Gerakopoulos, N. Voglis, and C. Efthymiopou- los, Physica A: Statistical Mechanics and its Applica- tions 387, 1907 (2008)

    2008

  28. [36]

    Froeschl´ e, M

    C. Froeschl´ e, M. Guzzo, and E. Lega, Celestial Mechan- ics and Dynamical Astronomy 92, 243–255 (2005)

    2005

  29. [37]

    D. P. Kingma and J. Ba, in 3rd International Con- ference on Learning Representations, ICLR 2015, San Diego, CA, USA, May 7-9, 2015, Conference Track Pro- ceedings, edited by Y. Bengio and Y. LeCun (2015) arXiv:1412.6980 [cs.LG]

    2015 arXiv

  30. [38]

    Sukov´ a, M

    P. Sukov´ a, M. Grzedzielski, and A. Janiuk, A& A586, A143 (2016)

    2016

  31. [39]

    Hegger, H

    R. Hegger, H. Kantz, and T. Schreiber, Chaos: An Interdisciplinary Journal of Nonlinear Science 9, 413 (1999), https://pubs.aip.org/aip/cha/article- pdf/9/2/413/18301636/413 1 online.pdf

    1999

  32. [40]

    Marwan, Commandline recurrence plots (2006), https://tocsy.pik-potsdam.de/commandline-rp

    N. Marwan, Commandline recurrence plots (2006), https://tocsy.pik-potsdam.de/commandline-rp. php

    2006

  33. [41]

    Marwan, Recurrence plots and cross recurrence plots (2024), http://www.recurrence-plot.tk

    N. Marwan, Recurrence plots and cross recurrence plots (2024), http://www.recurrence-plot.tk. Appendix A: Codes Code used to generate trajectories, train and evaluate the networks has been implemented in Python 3.11.5. The machine learning part has been done using the Py- Torc...

    2024

Pith tools

Reviewed August 10, 2026 · model on record in the stance chip above.