Pith. sign in

REVIEW 4 major objections 5 minor 10 references

Machine-Precision Prediction of Low-Dimensional Chaotic Systems from Noise-Free Data

T0 review · 4 major / 5 minor · reviewed 2026-08-06 · deepseek-v4-flash

Pith's one-line read Noise-free forecasting of low-dimensional chaos is effectively solved: polynomial regression in 512-bit arithmetic matches or beats 64-bit solvers that know the true equations, reaching 105 Lyapunov times on Lorenz-63.

desk verdict Record VPTs from a high-precision polynomial propagator, but the benchmark is relative to a fixed RK4 integrator and the best numbers are selected on the test set; still deserves a serious referee. read the letter →

arxiv 2507.09652 v2 pith:GAXT4SWN submitted 2025-07-13 nlin.CD cs.LGmath.DS

classification nlin.CDcs.LGmath.DS MSC 37D4537M1065P20 PACS 05.45.-a05.45.Pq
keywords chaosforecastingvalidpredictiontimepolynomialregressionmachineprecisionmulti-precisionarithmeticLorenz-63Lorenz-96Thomascyclicallysymmetricattractor
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

Chaotic forecasting is usually framed as hard because tiny errors double on the Lyapunov timescale. The paper claims that with perfect (noise-free) data the difficulty mostly evaporates: fitting the system's one-step propagator by ordinary least squares on high-degree monomials, executed in 512-bit arithmetic, yields forecasts that are limited only by how accurately the training data itself represents the true trajectory. On Lorenz-63 the method reaches valid prediction times of 32 to 105 Lyapunov times, far beyond the prior ceiling of 13, and matches or exceeds the forecast accuracy of RK4 solvers that are given the governing equations. The same recipe works for the non-polynomial Thomas attractor and for Lorenz-96 in dimensions 5 through 9. The authors conclude that noise-free low-dimensional chaotic benchmarks are effectively solved problems, and too easy for evaluating machine-learning dynamics methods.

What carries the argument

The load-bearing object is the polynomial propagator: the map $\hat\Phi_{\Delta t}(s) = \hat\beta^\top \xi_p(s)$ obtained by ordinary least squares, where $\xi_p$ stacks all monomials of degree $\le p$ in the $d$ state coordinates and $\hat\beta = (X^\top X)^{-1}X^\top Y$ solves the normal equations. Its two supporting mechanisms are high-precision linear algebra (512-bit arithmetic) to handle the ill-conditioned design matrix, and an auxiliary structural fact: the exact propagator of the RK4 integrator applied $k$ times to the Lorenz-63 equations is a polynomial of degree $F_{4k+2}$, the $(4k+2)$-nd Fibonacci number, so a single RK4 step is exactly a degree-$8$ polynomial. That identity explains why moderate-degree fits saturate the solver's own accuracy when the data step equals the solver step, and why sub-sampling the solver output (two steps already give degree 55) makes the learning task intrinsically harder.

What would settle it

Generate training and test data from a qualitatively different high-accuracy reference, for example an adaptive-step solver or a Taylor-method integrator with error control, and rerun the default configuration: if the 32 to 105 Lyapunov-time valid predictions persist, they reflect the underlying dynamics, while a collapse toward the 13 Lyapunov-time regime would show the result is specific to imitating the RK4 map. A second, cheaper check is to inject deterministic rounding noise into the last stored bit of the 64-bit training data and measure the valid prediction time as a function of that noise level.

Watch

Extended reading notes

Core claim

The paper's central claim is that a system-agnostic forecaster, namely ordinary least squares regression of the propagator $u(t)\mapsto u(t+\Delta t)$ on the multivariate monomials of the current state, reaches the numerical accuracy of the RK4 integrator that generated the data, and in the best configurations slightly exceeds it. The achievement is not a clever statistical model but numerical hygiene: high-degree polynomial features make the normal equations extremely ill-conditioned, and the paper shows that 512-bit arithmetic with full data normalization removes that bottleneck, while a 32-bit or 64-bit pipeline caps out near 7 or 21 Lyapunov times. Because the training data is noise-free, regularization is counterproductive and OLS is the right estimator. In the default setup (512-bit solver, 64-bit stored data, 512-bit fitter) the learned propagator reaches a valid prediction time of 35.6 Lyapunov times on Lorenz-63 versus 32.1 for a 64-bit RK4 solver, and with 512 bits used end to end it reaches 105 Lyapunov times, a ceiling the authors attribute to their chosen degree bound rather than to the method.

Load-bearing premise

The whole comparison treats the output of one particular numerical solver, run at one fixed time step, as the true system: the polynomial learns to imitate that solver's discrete map rather than the exact continuous-time equations, so the claimed machine-precision ceiling is relative to that solver.

Editorial extensions

If this is right

  • On Lorenz-63, previously published methods top out near 13 Lyapunov times; this method reaches 32 to 36 in realistic precision configurations and 105 when data and computation are both carried at 512-bit precision, averaged over 100 initial conditions.
  • The learned propagator outperforms a 64-bit RK4 solver of the true equations in the default setup (35.6 versus 32.1 Lyapunov times), so a data-only forecast can be more accurate than a standard solver that is handed the exact vector field.
  • The results transfer to the non-polynomial Thomas' Cyclically Symmetric Attractor, where degree-25 polynomials match a 512-bit solver at 38.0 Lyapunov times, showing the success does not depend on the vector field being polynomial.
  • For the spatiotemporally chaotic Lorenz-96 model in dimensions 5 through 9, the method reaches single-precision machine accuracy, with valid prediction times statistically indistinguishable from the reference solver's.
  • Forecasting from a single state (random test mode) performs as well as continuing from the training trajectory, because the estimated propagator depends only on the current state.
  • The authors draw the practical conclusion that noise-free Lorenz-63-style benchmarks are too easy to discriminate among methods, and that future work on learning dynamical systems should focus on noisy data.

Reading between the lines

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

  • If the claim holds, the relevant question shifts from 'can we learn the system' to 'where does the precision floor sit': the error budget is set by the least precise component of the data pipeline, so adding deterministic rounding noise at the level of the last stored bit should visibly cut the achievable valid prediction time, a cheap testable extension of the paper's own logic.
  • Because the fitted object is the discrete integrator map, the method's machine-precision result is relative to whatever solver generated the data; training on RK4 output and validating against a radically different high-accuracy integrator would separate 'learning the continuous dynamics' from 'imitating the solver', which the paper does not test.
  • The observed preference for a medium data step ($\Delta t \approx 2^{-5}$) over finer sampling suggests a trade-off between propagator simplicity and state-space exploration; sweeping $n$ beyond $2^{15}$ and degree beyond 16 could map where the 105-Lyapunov-time ceiling actually moves.
  • The authors' own caveat that dimensions of order 100 are computationally infeasible implies the method's success is tied to low dimension; applying it to moderately high-dimensional or noisy settings is where the 'solved problem' conclusion is most likely to break.
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

4 major / 5 minor

Summary. The paper proposes a system-agnostic forecasting method that fits high-degree polynomial features to the discrete-time propagator of a chaotic system using ordinary least squares, executed in 512-bit arithmetic. On Lorenz-63, it reports valid prediction times of about 35.6 Lyapunov times in the default setup and up to 104.9 (or 322 at the solver step size) with 512-bit data, far exceeding prior ML results and slightly exceeding the 64-bit RK4 solver baseline. The method is also applied to Thomas' Cyclically Symmetric Attractor and Lorenz-96 in dimensions 5 through 9, where it matches the respective solver baselines. All training and test trajectories are generated by the authors' fixed RK4 integrator, and the VPT is evaluated against that same discrete reference.

Significance. If the claims are read as "the method reproduces the data-generating discrete map at machine precision," the results are substantial: the empirical design is thorough, with 100 repetitions per configuration, confidence intervals, a Zenodo data release, and a formal proof in Appendix G that k RK4 steps of L63 are a polynomial of Fibonacci degree. The method is simple, data-efficient, and makes falsifiable predictions that prior ML benchmarks consistently fail to reach. However, the current abstract and title claim accuracy relative to "the true underlying dynamical systems," which is not established because the ground truth is always the output of one fixed RK4 integrator. This overstatement materially tempers the significance and needs to be addressed.

major comments (4)
  1. [Abstract; §1; §2.4; §3.1] The load-bearing claim that the method "exceeds the accuracy of standard 64-bit numerical ODE solvers of the true underlying dynamical systems" is not supported by the experiments, because the ground truth is always the output of the authors' fixed RK4 integrator with dt0=2^-10 (L63/L96) or 2^-6 (TCSA), and the same solver output is used for training and for VPT evaluation (Section 2.4, Table 6). The polynomial propagator therefore fits the discrete RK4 map, not the continuous-time flow; Appendix G makes this explicit by proving that one RK4 step of L63 is exactly a degree-8 polynomial, and Section 3.2 reports more than 322 Lyapunov times in the dt=dt0 case "because one RK4 solver step of the L63 system amounts to a polynomial propagator of degree 8." I recommend either reframing the abstract and title claims as matching or exceeding 64-bit solvers on the same discrete data-generating map, or adding an experiment that trains on the RK4 data and evaluates against a different high-accuracy reference (e.g., a much smaller-step 512-bit integration or a different integration scheme) to show that the learned propagator tracks the continuous flow itself.
  2. [§3.4; Table 3] The Lorenz-96 claims of "single-precision machine accuracy" and of results "statistically indistinguishable" from the solver share the same discrete-map limitation: both the training data and the RK4 reference forecast are generated by the same RK4 solver at the same dt0, so the comparison quantifies how well the polynomial propagator emulates the RK4 map given single-precision initial data, not how well it approximates the continuous-time Lorenz-96 flow. The equivalence test is also informal: "statistically indistinguishable" is defined as the solver's VPT lying inside the 95% confidence interval of the method, which with 100 repetitions and interval widths around ±0.8 Lyapunov times provides only weak evidence of equivalence; a formal test or a stated equivalence margin would be more appropriate.
  3. [§4 (Discussion); §3.2 (Figure 7)] The sentence "chaotic systems can be predicted to arbitrary accuracy—and thus arbitrary valid times" overreaches the evidence. All reported VPT values are for a fixed polynomial degree bound (p <= 16, or 25 for TCSA) and for a fixed data precision; the paper's own analysis (Figure 5 and Appendix C) shows the forecast error floor is set by the 64-bit data rounding, and no convergence result in p and n is proved. The claim should be narrowed to "arbitrary accuracy in reproducing the discrete-time propagator, up to the precision of the stored data and within the chosen polynomial feature family."
  4. [§3.2; Appendix G] The statement that the dt=dt0 result "is not a concern for larger time steps" is insufficient, because every VPT value in Table 2, including the headline 35.6 and 104.9 values, is measured against the same RK4-generated reference trajectory. The distinction between (a) exactly recovering the RK4 map when dt=dt0 and (b) approximating the map of the continuous-time system needs to be drawn in the methods and reflected in the interpretation of the headline results; otherwise readers will take "machine precision" to refer to the ODE flow rather than to the fixed numerical scheme.
minor comments (5)
  1. [Appendix A] The heading "Layapunov Exponent" is a typo and should read "Lyapunov Exponent."
  2. [Table 1; Figure 1; Appendix I captions] Several captions and labels contain stray spaces such as "V alid", "T raining", "T est", and "T ruth"; these should be cleaned up.
  3. [§2.4, Eq. (13)] The diagonal normalization is written with \hat{\sigma}^2_i inside a diagonal matrix, but the notation is ambiguous; please define \hat{S} explicitly as diag(\hat{\sigma}^2_1, ..., \hat{\sigma}^2_d) and clarify how \hat{S}^{-1/2} is computed.
  4. [§3.1, Figure 5] The plateau of constant error for about 8 Lyapunov times is an artifact of comparing a 512-bit forecast against 64-bit stored ground truth; the text should state this explicitly so readers do not attribute the plateau to the physics of the Lorenz system.
  5. [§4.2] The statement "there is no noise to overfit to" is too categorical, since finite-sample variance and multicollinearity of high-degree monomial bases are present even for noise-free data; the poor performance at small n and high degree visible in the Appendix I tables suggests these effects matter and deserve a sentence of qualification.

Circularity Check

0 steps flagged · score 2.0 of 10

No significant circularity: VPT is measured on held-out continuations; the one self-citation is peripheral and not load-bearing.

full rationale

The paper's central derivation is self-contained. A polynomial propagator is estimated by OLS regression on noise-free observations (Section 2.3, Eq. 9) and then iterated from a starting state; the reported VPT is evaluated against a held-out continuation or, in random test mode, against a new trajectory from a randomly chosen attractor state (Appendix E). The target is not used in fitting, and VPT is not a fitted parameter. The extreme value of more than 322 Lyapunov times at dt = dt0 is explicitly explained in Section 3.2 and Appendix G as a consequence of the RK4 one-step map of L63 being exactly a degree-8 polynomial; the model class therefore contains the target, which explains the result but is not circular because the polynomial coefficients are obtained by regression rather than by construction. The comparison with 64-bit RK4 solvers uses independently integrated trajectories from the same step size and different arithmetic precision (Table 6), which is a legitimate empirical benchmark. The only self-citation that could be questioned, the claim about poor performance under noise attributed to Schotz et al. [Sch+25] in Section 4.2, is peripheral to the central claim and not load-bearing. The main caveat, that ground truth is the output of a fixed RK4 integrator rather than the exact continuous-time flow, is a validity limitation of the benchmark rather than a circularity of the derivation.

Assumptions & free parameters 3 free parameters · 5 assumptions · 0 invented entities

The central claim rests on the RK4-generated dataset serving as ground truth, on the polynomial regression having enough capacity and numerical precision to fit the propagator, and on the training trajectories covering the attractor. No new physical entities are introduced. The main tuned or selected ingredients are the polynomial degree, the sampling time step, and the normalization scheme, all chosen to maximize reported VPT.

free parameters (3)
  • polynomial degree p = 1 to 16 for L63/L96, up to 25 for TCSA
    The degree is chosen by the authors to maximize VPT on the test set for each n and Delta t (Section 2.5; label boxes in Figures 4, 6, 7, 8). The headline VPT values are the best result over this grid, so the degree acts as a free selection parameter rather than a fixed model choice.
  • data time step Delta t = 2^-10 to 2^-3 for L63, 2^-9 to 2^-5 for L96, 2^-6 to 2^1 for TCSA
    The reported best VPT for each setting is obtained by maximizing over Delta t (Section 2.5, Figures 4 and 6). Delta t materially changes the complexity of the propagator and the number of prediction steps, so the optimized value is part of the reported result.
  • data normalization scheme = none, diagonal, or full
    For single- and double-precision methods, the choice of normalization strongly affects VPT (Appendix F), and the authors report the best scheme per precision setting, with full normalization for low precision and none for 512-bit. This categorical choice is also selected by test performance.
assumptions (5)
  • domain assumption The numerical output of the RK4 solver with fixed time step Delta t0 is accepted as ground truth for the true dynamical system.
    Figures 3 and Tables 2 and 6 define training data, test data, and reference solver all through the same RK4 scheme. The claim of machine-precision prediction is therefore relative to this discrete propagator, not to the exact continuous solution.
  • domain assumption The training trajectory explores the attractor well enough for the OLS-fitted polynomial propagator to generalize to unseen test states.
    OLS is an empirical risk minimizer over the training samples; the paper provides no bound or proof that the fitted polynomial propagator approximates the true propagator on the whole attractor. The random test mode in Appendix E gives empirical support, but assumes the training data covers the attractor.
  • domain assumption The linear system X^T X can be solved accurately at the chosen arithmetic precision.
    High-degree polynomial features yield ill-conditioned matrices, as described in Section 2.4. The method depends on 512-bit arithmetic and specific linear solvers (MPLAPACK RGESV, Armadillo solve with options such as allow ugly) to produce an accurate solution; at lower precision without normalization the method fails.
  • domain assumption The propagator of the system is sufficiently smooth over the attractor to be approximated by a degree-p polynomial to the needed accuracy.
    For the polynomial L63 propagator with Delta t = Delta t0 this is exact, as shown in Appendix G, but for generic non-polynomial systems like TCSA the paper relies on polynomial approximation of a smooth function. No error bound is given, and degree 25 is needed to match the solver.
  • standard math Standard existence and uniqueness of ODE solutions holds for the studied systems.
    Invoked in Section 2.1 to justify the propagator formalism; this holds for the polynomial and smooth sine systems used here.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Machine-Precision Prediction of Low-Dimensional Chaotic Systems from Noise-Free Data." pith.science (2026). https://pith.science/paper/GAXT4SWN

@misc{pith2026250709652,
  author       = {Pith},
  title        = {Pith review of: Machine-Precision Prediction of Low-Dimensional Chaotic Systems from Noise-Free Data},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/GAXT4SWN}},
  note         = {Machine review of arXiv:2507.09652}
}
read the original abstract

Low-dimensional chaotic systems such as the Lorenz-63 model are commonly used to benchmark system-agnostic methods for learning dynamics from data. This study shows that learning from noise-free observations in such systems can be achieved up to machine precision: using ordinary least squares regression on high-degree polynomial features with 512-bit arithmetic, a system-agnostic method is introduced that matches the accuracy of standard 64-bit numerical ODE solvers using the systems' governing equations. For the Lorenz-63 system, the method achieves valid prediction times of 36 Lyapunov times, and even up to 105 Lyapunov times with favorable precision configurations, dramatically outperforming prior work, which reaches 13 Lyapunov times at most. The results are further validated on Thomas' Cyclically Symmetric Attractor, a non-polynomial chaotic system that is considerably more complex than the Lorenz-63 model, and similar results extend to higher dimensions using the spatiotemporally chaotic Lorenz-96 model. These findings suggest that forecasting low-dimensional chaotic systems from noise-free data is effectively a solved problem.

Figures

Figures reproduced from arXiv: 2507.09652 by the authors.

Figure 1
Figure 1. An example illustrating the forecast abilities of the polynomial propagator introduced in this study. The L63 system is solved with an RK4 ODE solver calculating at 512-bit precision with time step 2−10 ≈ 0.001 (system time). The result is stored only at 64-bit precision. It is sub-sampled to obtain a time series with time step ∆t = 2−6 ≈ 0.016. We use the first n = 210 = 1024 samples (red) as training data (16.0 sy… view at source ↗
Figure 2
Figure 2. State space view of the Lorenz-63 (L63), Thomas’ Cyclically Symmetric Attractor (TCSA), as well as 5-dimensional and 9-dimensional versions of the Lorenz￾96 system (L96D5 and L96D9). We integrate L63, L96D5, and L96D9 for 20 and TCSA for 200 system time units and project the state space to 2 dimensions using principal component analysis. 9 [PITH_FULL_IMAGE:figures/full_fig_p009_2.png] view at source ↗
Figure 3
Figure 3. Data processing diagram for the default setup. We solve the ODE system with a multi precision (512-bit) RK4 solver. The train and test data are then stored at double precision (64 bit). The polynomial propagator method internally calculates with multi precision. The evaluation and calculation of the error metric is executed at standard double precision. RK4 ODE solver 32bit RK4 ODE solver 64bit RK4 ODE solver 512bit… view at source ↗
Figures from the paper (69 more)
Figure 4
Figure 4. Figure 4: Valid prediction times for the L63 system using the polynomial propagator method with optimal degree under the default setup. We use a multi-precision solver (512- bit), double-precision data storage (64-bit), and a multi-precision polynomial propagator. The data time …
Figure 5
Figure 5. Figure 5: Euclidean distance between forecast and ground truth of a single L63 run. This is the best-performing repetition of the default setup (L63, 512-bit system, 64-bit data, 512-bit method, sequential test mode, n = 215, ∆t = 2−7 , p = 9). The horizontal locations of the ve…
Figure 6
Figure 6. Figure 6: Valid prediction time (VPT) for L63 with best polynomial degree and best time step in different precision setting. The precision is encoded by three letters. The first letter represents the system (ODE solver), the second the data storage, and the third the forecasting…
Figure 7
Figure 7. Figure 7: Valid Prediction Time (VPT) for L63 with best polynomial degree for maxi￾mum precision throughout the processing pipeline. We use multi precision (512-bit) for the system, data, and method. For each value of ∆t (indicated by color) and n, we take the VPT0.5 (in Lyapuno…
Figure 8
Figure 8. Figure 8: Valid prediction times (VPTs) for the TCSA system using the polynomial propagator method with optimal degree under the default setup. We use a multi-precision solver (512-bit), double-precision data storage (64-bit), and a multi-precision (512-bit) polynomial propagato…
Figure 9
Figure 9. Figure 9: Average compute time for single precision polynomial propagator estimation in dimension [PITH_FULL_IMAGE:figures/full_fig_p023_9.png]
Figure 10
Figure 10. Figure 10: Average compute time for double precision polynomial propagator estimation in dimen [PITH_FULL_IMAGE:figures/full_fig_p024_10.png]
Figure 11
Figure 11. Figure 11: Average compute time for multi precision polynomial propagator estimation in dimen [PITH_FULL_IMAGE:figures/full_fig_p024_11.png]
Figure 12
Figure 12. Figure 12: Average compute time for double precision polynomial propagator estimation on L96. [PITH_FULL_IMAGE:figures/full_fig_p025_12.png]
Figure 13
Figure 13. Figure 13: Valid prediction time for L63 with best polynomial degree and best time step in different test modes. We use the default setting of multi-precision system, double-precision data, and multi-precision method and compare random (start of test set is randomly chosen on th…
Figure 14
Figure 14. Figure 14: Valid prediction time for L63 with best polynomial degree and best time step for single and double precision as well as different normalizations. We use the setting of multi-precision system and double-precision data and sequential test mode. We compare the polynomial…
Figure 15
Figure 15. Figure 15: Valid prediction time for the L96 model with single precision solver opti￾mized over polynomial degree and time step. We use a single-precision RK4 ODE solver (32-bit), double-precision data storage (64-bit), and a double-precision polynomial propagator. The system di…
Figure 16
Figure 16. Figure 16: Valid prediction time for the L96 model with double precision solver rounded to single precision optimized over polynomial degree and time step. We use a double￾precision RK4 ODE solver (64-bit), single-precision data storage (32-bit), and a double-precision polynomia…
Figure 17
Figure 17. Figure 17: Best Plot for L63, dds, normalize none, test sequential. See the beginning of Appendix I for a description. n ∆t 2−10 2−9 2−8 2−7 2−6 2−5 2−4 2−3 2 3 = 8 0.1 (1) 0.1 (1) 0.1 (1) 0.1 (1) 0.2 (1) 0.1 (1) 0.2 (2) 0.3 (1) 2 4 = 16 0.1 (1) 0.1 (1) 0.1 (1) 0.2 (1) 0.1 (1) 0…
Figure 18
Figure 18. Figure 18: All Plot for L63, dds, normalize none, test sequential. See the beginning of Appendix I for a description. 34 [PITH_FULL_IMAGE:figures/full_fig_p034_18.png]
Figure 19
Figure 19. Figure 19: Best Plot for L63, dds, normalize diag, test sequential. See the beginning of Appendix I for a description. n ∆t 2−10 2−9 2−8 2−7 2−6 2−5 2−4 2−3 2 3 = 8 0.2 (1) 0.2 (1) 0.2 (1) 0.2 (1) 0.2 (1) 0.1 (2) 0.3 (2) 0.3 (1) 2 4 = 16 0.1 (1) 0.1 (1) 0.2 (1) 0.2 (1) 0.1 (1) 0…
Figure 20
Figure 20. Figure 20: All Plot for L63, dds, normalize diag, test sequential. See the beginning of Appendix I for a description. 36 [PITH_FULL_IMAGE:figures/full_fig_p036_20.png]
Figure 21
Figure 21. Figure 21: Best Plot for L63, dds, normalize full, test sequential. See the beginning of Appendix I for a description. n ∆t 2−10 2−9 2−8 2−7 2−6 2−5 2−4 2−3 2 3 = 8 0.1 (1) 0.1 (1) 0.2 (1) 0.2 (1) 0.2 (1) 0.1 (1) 0.1 (1) 0.3 (1) 2 4 = 16 0.1 (1) 0.2 (1) 0.2 (1) 0.2 (1) 0.1 (1) 0…
Figure 22
Figure 22. Figure 22: All Plot for L63, dds, normalize full, test sequential. See the beginning of Appendix I for a description. 38 [PITH_FULL_IMAGE:figures/full_fig_p038_22.png]
Figure 23
Figure 23. Figure 23: Best Plot for L63, ddd, normalize none, test sequential. See the beginning of Appendix I for a description. n ∆t 2−10 2−9 2−8 2−7 2−6 2−5 2−4 2−3 2 3 = 8 0.2 (1) 0.2 (1) 0.2 (1) 0.2 (1) 0.2 (2) 0.1 (2) 0.2 (2) 0.3 (1) 2 4 = 16 0.2 (1) 0.2 (1) 0.2 (1) 0.2 (2) 0.7 (2) 0…
Figure 24
Figure 24. Figure 24: All Plot for L63, ddd, normalize none, test sequential. See the beginning of Appendix I for a description. 40 [PITH_FULL_IMAGE:figures/full_fig_p040_24.png]
Figure 25
Figure 25. Figure 25: Best Plot for L63, ddd, normalize diag, test sequential. See the beginning of Appendix I for a description. n ∆t 2−10 2−9 2−8 2−7 2−6 2−5 2−4 2−3 2 3 = 8 0.2 (1) 0.2 (1) 0.2 (1) 0.2 (1) 0.2 (1) 0.1 (1) 0.1 (2) 0.3 (1) 2 4 = 16 0.2 (1) 0.2 (1) 0.2 (1) 0.3 (2) 1.0 (2) 0…
Figure 26
Figure 26. Figure 26: All Plot for L63, ddd, normalize diag, test sequential. See the beginning of Appendix I for a description. 42 [PITH_FULL_IMAGE:figures/full_fig_p042_26.png]
Figure 27
Figure 27. Figure 27: Best Plot for L63, ddd, normalize full, test sequential. See the beginning of Appendix I for a description. n ∆t 2−10 2−9 2−8 2−7 2−6 2−5 2−4 2−3 2 3 = 8 0.2 (1) 0.2 (1) 0.2 (1) 0.2 (1) 0.2 (1) 0.1 (1) 0.1 (1) 0.3 (1) 2 4 = 16 0.2 (1) 0.2 (1) 0.2 (1) 0.4 (2) 1.0 (2) 0…
Figure 28
Figure 28. Figure 28: All Plot for L63, ddd, normalize full, test sequential. See the beginning of Appendix I for a description. 44 [PITH_FULL_IMAGE:figures/full_fig_p044_28.png]
Figure 29
Figure 29. Figure 29: Best Plot for L63, sds, normalize full, test sequential. See the beginning of Appendix I for a description. n ∆t 2−10 2−9 2−8 2−7 2−6 2−5 2−4 2−3 2 3 = 8 0.1 (1) 0.1 (1) 0.2 (1) 0.2 (1) 0.2 (1) 0.1 (1) 0.1 (1) 0.3 (1) 2 4 = 16 0.1 (1) 0.2 (1) 0.2 (1) 0.2 (1) 0.1 (1) 0…
Figure 30
Figure 30. Figure 30: All Plot for L63, sds, normalize full, test sequential. See the beginning of Appendix I for a description. 46 [PITH_FULL_IMAGE:figures/full_fig_p046_30.png]
Figure 31
Figure 31. Figure 31: Best Plot for L63, sdd, normalize full, test sequential. See the beginning of Appendix I for a description. n ∆t 2−10 2−9 2−8 2−7 2−6 2−5 2−4 2−3 2 3 = 8 0.2 (1) 0.2 (1) 0.2 (1) 0.2 (1) 0.2 (1) 0.1 (1) 0.1 (1) 0.3 (1) 2 4 = 16 0.2 (1) 0.2 (1) 0.2 (1) 0.2 (2) 0.6 (2) 0…
Figure 32
Figure 32. Figure 32: All Plot for L63, sdd, normalize full, test sequential. See the beginning of Appendix I for a description. 48 [PITH_FULL_IMAGE:figures/full_fig_p048_32.png]
Figure 33
Figure 33. Figure 33: Best Plot for L63, mds, normalize full, test sequential. See the beginning of Appendix I for a description. n ∆t 2−10 2−9 2−8 2−7 2−6 2−5 2−4 2−3 2 3 = 8 0.1 (1) 0.1 (1) 0.2 (1) 0.2 (1) 0.2 (1) 0.2 (1) 0.1 (1) 0.3 (1) 2 4 = 16 0.1 (1) 0.2 (1) 0.2 (1) 0.2 (1) 0.2 (1) 0…
Figure 34
Figure 34. Figure 34: All Plot for L63, mds, normalize full, test sequential. See the beginning of Appendix I for a description. 50 [PITH_FULL_IMAGE:figures/full_fig_p050_34.png]
Figure 35
Figure 35. Figure 35: Best Plot for L63, mdd, normalize full, test sequential. See the beginning of Appendix I for a description. n ∆t 2−10 2−9 2−8 2−7 2−6 2−5 2−4 2−3 2 3 = 8 0.2 (1) 0.2 (1) 0.2 (1) 0.2 (1) 0.2 (1) 0.2 (1) 0.1 (1) 0.3 (1) 2 4 = 16 0.2 (1) 0.2 (1) 0.2 (1) 0.4 (2) 0.9 (2) 0…
Figure 36
Figure 36. Figure 36: All Plot for L63, mdd, normalize full, test sequential. See the beginning of Appendix I for a description. 52 [PITH_FULL_IMAGE:figures/full_fig_p052_36.png]
Figure 37
Figure 37. Figure 37: Best Plot for L63, dsd, normalize full, test sequential. See the beginning of Appendix I for a description. n ∆t 2−10 2−9 2−8 2−7 2−6 2−5 2−4 2−3 2 3 = 8 0.2 (1) 0.2 (1) 0.2 (1) 0.2 (1) 0.2 (1) 0.1 (1) 0.1 (1) 0.3 (1) 2 4 = 16 0.2 (1) 0.2 (1) 0.2 (1) 0.2 (2) 0.8 (2) 0…
Figure 38
Figure 38. Figure 38: All Plot for L63, dsd, normalize full, test sequential. See the beginning of Appendix I for a description. 54 [PITH_FULL_IMAGE:figures/full_fig_p054_38.png]
Figure 39
Figure 39. Figure 39: Best Plot for L63, msd, normalize full, test sequential. See the beginning of Appendix I for a description. n ∆t 2−10 2−9 2−8 2−7 2−6 2−5 2−4 2−3 2 3 = 8 0.2 (1) 0.2 (1) 0.2 (1) 0.2 (1) 0.2 (1) 0.2 (1) 0.1 (1) 0.3 (1) 2 4 = 16 0.2 (1) 0.2 (1) 0.2 (1) 0.2 (2) 0.7 (2) 0…
Figure 40
Figure 40. Figure 40: All Plot for L63, msd, normalize full, test sequential. See the beginning of Appendix I for a description. 56 [PITH_FULL_IMAGE:figures/full_fig_p056_40.png]
Figure 41
Figure 41. Figure 41: Best Plot for L63, sdm, normalize none, test sequential. See the beginning of Appendix I for a description. n ∆t 2−10 2−9 2−8 2−7 2−6 2−5 2−4 2−3 2 3 = 8 0.2 (1) 0.2 (1) 0.2 (1) 0.2 (1) 0.2 (1) 0.2 (2) 0.2 (2) 0.3 (1) 2 4 = 16 0.2 (1) 0.2 (1) 0.2 (1) 0.2 (2) 0.6 (2) 0…
Figure 42
Figure 42. Figure 42: All Plot for L63, sdm, normalize none, test sequential. See the beginning of Appendix I for a description. 58 [PITH_FULL_IMAGE:figures/full_fig_p058_42.png]
Figure 43
Figure 43. Figure 43: Best Plot for L63, ddm, normalize none, test sequential. See the beginning of Appendix I for a description. n ∆t 2−10 2−9 2−8 2−7 2−6 2−5 2−4 2−3 2 3 = 8 0.2 (1) 0.2 (1) 0.2 (1) 0.2 (1) 0.2 (2) 0.2 (2) 0.2 (2) 0.3 (1) 2 4 = 16 0.3 (2) 0.9 (2) 1.8 (2) 1.5 (2) 1.0 (2) 0…
Figure 44
Figure 44. Figure 44: All Plot for L63, ddm, normalize none, test sequential. See the beginning of Appendix I for a description. 60 [PITH_FULL_IMAGE:figures/full_fig_p060_44.png]
Figure 45
Figure 45. Figure 45: Best Plot for L63, mdm, normalize none, test sequential. See the beginning of Appendix I for a description. n ∆t 2−10 2−9 2−8 2−7 2−6 2−5 2−4 2−3 2 3 = 8 0.2 (1) 0.2 (1) 0.2 (1) 0.2 (1) 0.2 (1) 0.2 (2) 0.1 (1) 0.3 (1) 2 4 = 16 0.2 (2) 0.9 (2) 1.6 (2) 1.3 (2) 0.9 (2) 0…
Figure 46
Figure 46. Figure 46: All Plot for L63, mdm, normalize none, test sequential. See the beginning of Appendix I for a description. 62 [PITH_FULL_IMAGE:figures/full_fig_p062_46.png]
Figure 47
Figure 47. Figure 47: Best Plot for L63, mdm, normalize none, test random. See the beginning of Appendix I for a description. n ∆t 2−10 2−9 2−8 2−7 2−6 2−5 2−4 2−3 2 3 = 8 0.1 (1) 0.1 (1) 0.1 (1) 0.1 (1) 0.1 (1) 0.1 (1) 0.1 (1) 0.1 (1) 2 4 = 16 0.1 (1) 0.5 (2) 1.1 (2) 0.8 (2) 0.5 (2) 0.2 (…
Figure 48
Figure 48. Figure 48: All Plot for L63, mdm, normalize none, test random. See the beginning of Appendix I for a description. 64 [PITH_FULL_IMAGE:figures/full_fig_p064_48.png]
Figure 49
Figure 49. Figure 49: Best Plot for L63, mmm, normalize none, test sequential. See the beginning of Appendix I for a description. n ∆t 2−10 2−9 2−8 2−7 2−6 2−5 2−4 2−3 2 3 = 8 0.2 (1) 0.2 (1) 0.2 (1) 0.2 (1) 0.2 (1) 0.2 (2) 0.2 (2) 0.4 (1) 2 4 = 16 2.4 (2) 2.2 (2) 1.6 (2) 1.5 (2) 1.0 (2) 0…
Figure 50
Figure 50. Figure 50: All Plot for L63, mmm, normalize none, test sequential. See the beginning of Appendix I for a description. 66 [PITH_FULL_IMAGE:figures/full_fig_p066_50.png]
Figure 51
Figure 51. Figure 51: Best Plot for L96D5, sdd, normalize full, test sequential. See the beginning of Appendix I for a description. n ∆t 2−9 2−8 2−7 2−6 2−5 2 3 = 8 0.1 (1) 0.1 (1) 0.1 (1) 0.1 (1) 0.1 (1) 2 4 = 16 0.1 (1) 0.1 (1) 0.1 (1) 0.1 (1) 0.1 (1) 2 5 = 32 0.1 (1) 0.1 (1) 0.1 (1) 0.1…
Figure 52
Figure 52. Figure 52: All Plot for L96D5, sdd, normalize full, test sequential. See the beginning of Appendix I for a description. 68 [PITH_FULL_IMAGE:figures/full_fig_p068_52.png]
Figure 53
Figure 53. Figure 53: Best Plot for L96D6, sdd, normalize full, test sequential. See the beginning of Appendix I for a description. n ∆t 2−9 2−8 2−7 2−6 2−5 2 3 = 8 0.1 (1) 0.1 (1) 0.2 (1) 0.2 (1) 0.2 (1) 2 4 = 16 0.2 (1) 0.2 (1) 0.2 (1) 0.2 (1) 0.1 (1) 2 5 = 32 0.3 (1) 0.2 (1) 0.2 (1) 0.1…
Figure 54
Figure 54. Figure 54: All Plot for L96D6, sdd, normalize full, test sequential. See the beginning of Appendix I for a description. 70 [PITH_FULL_IMAGE:figures/full_fig_p070_54.png]
Figure 55
Figure 55. Figure 55: Best Plot for L96D7, sdd, normalize full, test sequential. See the beginning of Appendix I for a description. n ∆t 2−9 2−8 2−7 2−6 2−5 2 3 = 8 0.0 (1) 0.0 (1) 0.1 (1) 0.1 (1) 0.0 (1) 2 4 = 16 0.4 (1) 0.4 (1) 0.4 (1) 0.3 (1) 0.3 (1) 2 5 = 32 0.4 (1) 0.4 (1) 0.3 (1) 0.3…
Figure 56
Figure 56. Figure 56: All Plot for L96D7, sdd, normalize full, test sequential. See the beginning of Appendix I for a description. 72 [PITH_FULL_IMAGE:figures/full_fig_p072_56.png]
Figure 57
Figure 57. Figure 57: Best Plot for L96D8, sdd, normalize full, test sequential. See the beginning of Appendix I for a description. n ∆t 2−9 2−8 2−7 2−6 2−5 2 3 = 8 0.0 (1) 0.0 (1) 0.0 (1) 0.0 (1) 0.0 (1) 2 4 = 16 0.4 (1) 0.4 (1) 0.4 (1) 0.4 (1) 0.3 (1) 2 5 = 32 0.4 (1) 0.4 (1) 0.4 (1) 0.3…
Figure 58
Figure 58. Figure 58: All Plot for L96D8, sdd, normalize full, test sequential. See the beginning of Appendix I for a description. 74 [PITH_FULL_IMAGE:figures/full_fig_p074_58.png]
Figure 59
Figure 59. Figure 59: Best Plot for L96D9, sdd, normalize full, test sequential. See the beginning of Appendix I for a description. n ∆t 2−9 2−8 2−7 2−6 2−5 2 3 = 8 0.0 (1) 0.0 (1) 0.0 (1) 0.0 (1) 0.0 (1) 2 4 = 16 0.3 (1) 0.3 (1) 0.3 (1) 0.3 (1) 0.2 (1) 2 5 = 32 0.3 (1) 0.3 (1) 0.3 (1) 0.2…
Figure 60
Figure 60. Figure 60: All Plot for L96D9, sdd, normalize full, test sequential. See the beginning of Appendix I for a description. 76 [PITH_FULL_IMAGE:figures/full_fig_p076_60.png]
Figure 61
Figure 61. Figure 61: Best Plot for L96D5, dsd, normalize full, test sequential. See the beginning of Appendix I for a description. n ∆t 2−9 2−8 2−7 2−6 2−5 2 3 = 8 0.1 (1) 0.1 (1) 0.1 (1) 0.1 (1) 0.1 (1) 2 4 = 16 0.1 (1) 0.1 (1) 0.1 (1) 0.1 (1) 0.1 (1) 2 5 = 32 0.1 (1) 0.1 (1) 0.1 (1) 0.1…
Figure 62
Figure 62. Figure 62: All Plot for L96D5, dsd, normalize full, test sequential. See the beginning of Appendix I for a description. 78 [PITH_FULL_IMAGE:figures/full_fig_p078_62.png]
Figure 63
Figure 63. Figure 63: Best Plot for L96D6, dsd, normalize full, test sequential. See the beginning of Appendix I for a description. n ∆t 2−9 2−8 2−7 2−6 2−5 2 3 = 8 0.1 (1) 0.1 (1) 0.2 (1) 0.2 (1) 0.2 (1) 2 4 = 16 0.3 (1) 0.2 (1) 0.3 (1) 0.2 (1) 0.1 (1) 2 5 = 32 0.3 (1) 0.3 (1) 0.2 (1) 0.2…
Figure 64
Figure 64. Figure 64: All Plot for L96D6, dsd, normalize full, test sequential. See the beginning of Appendix I for a description. 80 [PITH_FULL_IMAGE:figures/full_fig_p080_64.png]
Figure 65
Figure 65. Figure 65: Best Plot for L96D7, dsd, normalize full, test sequential. See the beginning of Appendix I for a description. n ∆t 2−9 2−8 2−7 2−6 2−5 2 3 = 8 0.1 (1) 0.0 (1) 0.0 (1) 0.1 (1) 0.0 (1) 2 4 = 16 0.4 (1) 0.4 (1) 0.4 (1) 0.3 (1) 0.3 (1) 2 5 = 32 0.4 (1) 0.4 (1) 0.3 (1) 0.3…
Figure 66
Figure 66. Figure 66: All Plot for L96D7, dsd, normalize full, test sequential. See the beginning of Appendix I for a description. 82 [PITH_FULL_IMAGE:figures/full_fig_p082_66.png]
Figure 67
Figure 67. Figure 67: Best Plot for L96D8, dsd, normalize full, test sequential. See the beginning of Appendix I for a description. n ∆t 2−9 2−8 2−7 2−6 2−5 2 3 = 8 0.0 (1) 0.0 (1) 0.0 (1) 0.0 (1) 0.0 (1) 2 4 = 16 0.4 (1) 0.4 (1) 0.4 (1) 0.3 (1) 0.3 (1) 2 5 = 32 0.4 (1) 0.4 (1) 0.3 (1) 0.3…
Figure 68
Figure 68. Figure 68: All Plot for L96D8, dsd, normalize full, test sequential. See the beginning of Appendix I for a description. 84 [PITH_FULL_IMAGE:figures/full_fig_p084_68.png]
Figure 69
Figure 69. Figure 69: Best Plot for L96D9, dsd, normalize full, test sequential. See the beginning of Appendix I for a description. n ∆t 2−9 2−8 2−7 2−6 2−5 2 3 = 8 0.0 (1) 0.0 (1) 0.0 (1) 0.0 (1) 0.0 (1) 2 4 = 16 0.3 (1) 0.3 (1) 0.3 (1) 0.3 (1) 0.2 (1) 2 5 = 32 0.4 (1) 0.4 (1) 0.3 (1) 0.2…
Figure 70
Figure 70. Figure 70: All Plot for L96D9, dsd, normalize full, test sequential. See the beginning of Appendix I for a description. 86 [PITH_FULL_IMAGE:figures/full_fig_p086_70.png]
Figure 71
Figure 71. Figure 71: Best Plot for TCSA, mdm, normalize none, test sequential. See the beginning of Appendix I for a description. n ∆t 2−6 2−5 2−4 2−3 2−2 2−1 2 0 2 1 2 3 = 8 0.0 (1) 0.0 (1) 0.0 (1) 0.0 (1) 0.0 (1) 0.0 (1) 0.0 (1) 0.0 (2) 2 4 = 16 0.1 (2) 0.1 (2) 0.1 (2) 0.1 (2) 0.0 (2) 0…
Figure 72
Figure 72. Figure 72: All Plot for TCSA, mdm, normalize none, test sequential. See the beginning of Appendix I for a description. 88 [PITH_FULL_IMAGE:figures/full_fig_p088_72.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

10 extracted references · 6 canonical work pages

  1. [1]

    Computational Efficiency of Multi-Step Learning Echo State Networks for Nonlinear Time Series Prediction

    [AT22] T. Akiyama and G. Tanaka. “Computational Efficiency of Multi-Step Learning Echo State Networks for Nonlinear Time Series Prediction”. In: IEEE Access 10 (2022), pp. 28535–28544. doi: 10.1109/access.2022.3158755. [Ben+80a] G. Benettin, L. Galgani, A. Giorgilli, and J.-M. Strelcyn. “Lyapunov Characteristic Exponents for smooth dynamical systems and f...

  2. [7]

    PySINDy: A Python package for the sparse identification of nonlinear dynamical systems from data

    doi: 10 . 48550 / arXiv . 2407 . 20158. arXiv: 2407 . 20158 [cs.LG]. [Sil+20] B. M. de Silva, K. Champion, M. Quade, J.-C. Loiseau, J. N. Kutz, and S. L. Brunton. “PySINDy: A Python package for the sparse identification of nonlinear dynamical systems from data”. In: Journal of Open Source Software 5.49 (2020), p

  3. [9]

    Deterministic chaos seen in terms of feedback circuits: analysis, synthesis, “Labyrinth chaos

    doi: 10.1201/9780429398490. [Tho99] R. Thomas. “Deterministic chaos seen in terms of feedback circuits: analysis, synthesis, “Labyrinth chaos””. In: International Journal of Bifurcation and Chaos 09.10 (1999), pp. 1889–1905. doi: 10.1142/S0218127499001383. [Vis04] D. Viswanath. “The fractal property of the Lorenz attractor”. In: Phys. D 190.1-2 (2004), pp...

  4. [13]

    Chaotic attractor reconstruction using small reservoirs—the influence of topology

    [Jau24] L. Jaurigue. “Chaotic attractor reconstruction using small reservoirs—the influence of topology”. In: Machine Learning: Science and Technology 5.3 (Aug. 2024), p. 035058. doi: 10.1088/2632-2153/ad6ee8. [K¨ os+23] F. K¨ oster, D. Patel, A. Wikner, L. Jaurigue, and K. L¨ udge. “Data-informed reservoir computing for efficient time-series prediction”....

  5. [2019]
  6. [2020]

    Neural Networks for Lorenz Map Prediction: A Trip Through Time

    doi: 10.48550/arXiv.1903.07768. arXiv: 1903.07768 [cs.LG]. [SC25] C. Sanderson and R. Curtin. “Armadillo: an Efficient Framework for Numerical Linear Algebra”. In: 2025 17th International Conference on Computer and Automation En- gineering (ICCAE). 2025, pp. 303–307. doi: 10.1109/ICCAE64891.2025.10980539. 90 [Sch+25] C. Sch¨ otz, A. White, M. Gelbrecht, a...

  7. [2022]

    Using machine learning to replicate chaotic attractors and calculate Lyapunov exponents from data

    doi: https://doi.org/ 10.48550/arXiv.2109.13406. arXiv: 2109.13406 [cs.MS]. [Ott02] E. Ott. Chaos in dynamical systems. Second. Cambridge University Press, Cambridge, 2002, pp. xii+478. doi: 10.1017/CBO9780511803260. MR 1924000. [Pat+17] J. Pathak, Z. Lu, B. R. Hunt, M. Girvan, and E. Ott. “Using machine learning to replicate chaotic attractors and calcul...

  8. [2024]

    Theoretical basis and application of an analogue-dynamical model in the Lorenz system

    [Ren+09] H. Ren, J. Chou, J. Huang, and P. Zhang. “Theoretical basis and application of an analogue-dynamical model in the Lorenz system”. en. In: Adv. Atmos. Sci. 26.1 (Jan. 2009), pp. 67–77. [Rob20] D. Roberts. Neural Networks for Lorenz Map Prediction: A Trip Through Time

Show all 10 references
  1. [2025]

    arXiv: 2501.06661 [cs.LG]

    doi: 10.48550/arXiv.2501.06661. arXiv: 2501.06661 [cs.LG]. [Nak22] M. Nakata. MPLAPACK version 2.0.1 user manual

  2. [2104]

    Predicting three-dimensional chaotic systems with four qubit quantum systems

    doi: 10.21105/joss.02104. [SR25] J. Steinegger and C. R¨ ath. “Predicting three-dimensional chaotic systems with four qubit quantum systems”. In: Scientific Reports 15.1 (Feb. 2025). doi: 10 . 1038 / s41598-025-87768-0 . [Sti74] S. M. Stigler. “Gergonne’s 1815 paper on the des...

Pith tools

Reviewed August 6, 2026 · model on record in the stance chip above.