arxiv: 2604.16232 · v1 · submitted 2026-04-17 · 💻 cs.LG · cs.AI· cs.CE· cs.SC

Recognition: unknown

Neuro-Symbolic ODE Discovery with Latent Grammar Flow

Karin Yu , Eleni Chatzi , Georgios Kissas

Authors on Pith no claims yet

Pith reviewed 2026-05-10 08:33 UTC · model grok-4.3

classification 💻 cs.LG cs.AIcs.CEcs.SC

keywords equationsflowlatentdatadifferentialdiscretegrammarneuro-symbolic

0 comments

The pith

Latent Grammar Flow discovers ODEs by placing grammar-based equation representations in a discrete latent space, using a behavioral loss to cluster similar equations, and sampling via a discrete flow model guided by data fit and constraints.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

Scientists often want equations that explain how things change over time, like how a population grows or a machine vibrates. Instead of guessing equations by hand or using opaque neural networks, this approach turns possible equations into strings following grammar rules. These strings are placed in a special hidden space where similar equations sit close together. A flow model then moves through this space to generate new equation candidates that match the observed data while respecting rules like stability. Domain knowledge can be added either as grammar constraints or as extra predictors.

Core claim

We introduce Latent Grammar Flow (LGF), a neuro-symbolic generative framework for discovering ordinary differential equations from data. LGF embeds equations as grammar-based representations into a discrete latent space and forces semantically similar equations to be positioned closer together with a behavioural loss. Then, a discrete flow model guides the sampling process to recursively generate candidate equations that best fit the observed data.

Load-bearing premise

That a behavioral loss can reliably place semantically similar equations closer in the discrete latent space and that the discrete flow model can efficiently sample equations that both fit data and satisfy domain constraints without exhaustive search.

Figures

Figures reproduced from arXiv: 2604.16232 by Eleni Chatzi, Georgios Kissas, Karin Yu.

**Figure 2.** Figure 2: Violin plots of Benchmark 1, where the distribution is shown in blue and single values as red dots. [PITH_FULL_IMAGE:figures/full_fig_p007_2.png] view at source ↗

**Figure 3.** Figure 3: The numerical trajectories of the ground truth and predicted ODEs of Benchmark 2. [PITH_FULL_IMAGE:figures/full_fig_p014_3.png] view at source ↗

**Figure 4.** Figure 4: The numerical trajectories of the ground truth and predicted ODEs of Benchmark 3. [PITH_FULL_IMAGE:figures/full_fig_p015_4.png] view at source ↗

read the original abstract

Understanding natural and engineered systems often relies on symbolic formulations, such as differential equations, which provide interpretability and transferability beyond black-box models. We introduce Latent Grammar Flow (LGF), a neuro-symbolic generative framework for discovering ordinary differential equations from data. LGF embeds equations as grammar-based representations into a discrete latent space and forces semantically similar equations to be positioned closer together with a behavioural loss. Then, a discrete flow model guides the sampling process to recursively generate candidate equations that best fit the observed data. Domain knowledge and constraints, such as stability, can be either embedded into the rules or used as conditional predictors.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Desk Editor's Note private letter to a colleague

LGF combines grammar latents, behavioral clustering, and discrete flow for ODE discovery in a way that looks new but stays underspecified without the full methods and results.

read the letter

The main takeaway is that this paper introduces Latent Grammar Flow as a neuro-symbolic way to discover ODEs: it encodes grammar-derived equations into a discrete latent space, applies a behavioral loss to bring semantically similar ones closer, and then uses a discrete flow model to sample data-fitting candidates while allowing domain constraints like stability to be built in or conditioned on. The specific combination of grammar-based discrete representations, behavioral loss for clustering, and flow-guided sampling appears fresh relative to the symbolic regression and neuro-symbolic ODE work it cites. It does a reasonable job framing the problem around interpretability and transferability, and the idea of guiding generation away from exhaustive search via the flow is a practical angle worth exploring. The soft spots are clear from the abstract alone. No loss formulations, training details, similarity metrics for the behavioral loss, or experimental results are shown, so it is impossible to verify whether the clustering actually produces a usable metric space or whether the flow improves sampling efficiency over grammar-constrained enumeration. The stress-test concern about the behavioral loss failing to enforce reliable semantic proximity holds until the full paper supplies those pieces. This is aimed at people working on hybrid symbolic-neural methods for scientific modeling. A reader focused on interpretable discovery from data would find the framing useful, but only if the implementation details check out. It deserves a serious referee to evaluate the derivations and experiments, even if heavy revision is likely needed for clarity and validation.

Referee Report

2 major / 0 minor

Summary. The paper introduces Latent Grammar Flow (LGF), a neuro-symbolic generative framework for discovering ordinary differential equations from data. LGF embeds grammar-based equation representations into a discrete latent space, applies a behavioral loss to position semantically similar equations closer together, and uses a discrete flow model to recursively sample candidate equations that fit observed data while allowing domain knowledge and constraints (such as stability) to be incorporated via rules or conditional predictors.

Significance. If the embedding and sampling mechanisms operate as described, the framework could advance neuro-symbolic scientific discovery by enabling efficient, constrained search over symbolic ODEs without exhaustive enumeration, combining interpretability of grammars with generative modeling. The approach has potential to improve upon black-box or purely symbolic methods in terms of transferability and incorporation of prior knowledge, but this remains speculative without demonstrated results.

major comments (2)

[Abstract] Abstract: The core claim that the behavioral loss forces semantically similar equations closer in the discrete latent space is load-bearing for the discrete flow model's ability to guide sampling efficiently, yet no formulation of the loss, similarity metric (e.g., trajectory distance versus symbolic distance), or validation of the resulting embedding is provided.
[Abstract] Abstract: The assertion that the discrete flow model recursively generates candidate equations that best fit the data (while satisfying constraints) lacks any description of the flow architecture, training objective, data-fitting loss, or sampling procedure, which prevents assessment of whether the method actually outperforms grammar-constrained enumeration.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for their careful reading and constructive feedback. We address each major comment below and will revise the manuscript accordingly to improve clarity and completeness.

read point-by-point responses

Referee: [Abstract] Abstract: The core claim that the behavioral loss forces semantically similar equations closer in the discrete latent space is load-bearing for the discrete flow model's ability to guide sampling efficiently, yet no formulation of the loss, similarity metric (e.g., trajectory distance versus symbolic distance), or validation of the resulting embedding is provided.

Authors: We agree that the abstract is high-level and does not include the mathematical formulation of the behavioral loss, the choice of similarity metric, or any validation of the resulting embedding. This omission limits the reader's ability to assess the claim. In the revised version we will expand the abstract with a concise description of the loss and metric, and we will ensure the main text supplies the full definition, the explicit similarity measure, and the corresponding validation experiments. revision: yes
Referee: [Abstract] Abstract: The assertion that the discrete flow model recursively generates candidate equations that best fit the data (while satisfying constraints) lacks any description of the flow architecture, training objective, data-fitting loss, or sampling procedure, which prevents assessment of whether the method actually outperforms grammar-constrained enumeration.

Authors: We agree that the abstract provides no technical description of the discrete flow model, its architecture, training objective, data-fitting loss, or sampling procedure. This prevents a proper evaluation of the claimed advantages. We will revise the abstract to include a brief overview of these elements and will add or clarify the corresponding details in the methods section so that readers can compare the approach against exhaustive enumeration. revision: yes

Circularity Check

0 steps flagged

No significant circularity in derivation chain

full rationale

The paper presents LGF as a neuro-symbolic framework that embeds grammar-based equation representations into a discrete latent space using a behavioral loss to cluster semantically similar equations, followed by a discrete flow model for guided sampling of data-fitting candidates. These mechanisms are introduced as independent architectural choices without any reduction of the claimed discovery process to a fitted parameter, self-referential definition, or load-bearing self-citation. The abstract and described components do not exhibit any step where a prediction or result is equivalent to its inputs by construction. The framework remains self-contained against external benchmarks.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

Review performed on abstract only; no explicit free parameters, axioms, or invented entities are stated in the provided text. The framework itself is the novel contribution rather than a new physical entity.

pith-pipeline@v0.9.0 · 5402 in / 1157 out tokens · 18990 ms · 2026-05-10T08:33:51.307611+00:00 · methodology

discussion (0)

Reference graph

Works this paper leans on

43 extracted references · 25 canonical work pages · 3 internal anchors

[1]

Discovery of equations: experimental evaluation of convergence,

R. Zembowicz and J. M. Zytkow, “Discovery of equations: experimental evaluation of convergence,” inProceed- ings of the Tenth National Conference on Artificial Intelligence, ser. AAAI’92. AAAI Press, 1992, pp. 70–75, place: San Jose, California. 8 Yu et al. Neuro-Symbolic ODE Discovery with Latent Grammar FlowPreprint April 2026

1992
[2]

, & author Lipson, H

M. Schmidt and H. Lipson, “Distilling Free-Form Natural Laws from Experimental Data,”Science, vol. 324, no. 5923, pp. 81–85, Apr. 2009. [Online]. Available: https://www.science.org/doi/10.1126/science.1165893

work page doi:10.1126/science.1165893 2009
[3]

Neural Ordinary Differential Equations,

R. T. Q. Chen, Y . Rubanova, J. Bettencourt, and D. K. Duvenaud, “Neural Ordinary Differential Equations,” inAdvances in Neural Information Processing Systems, S. Bengio, H. Wallach, H. Larochelle, K. Grauman, N. Cesa-Bianchi, and R. Garnett, Eds., vol. 31. Curran Associates, Inc., 2018. [Online]. Available: https://proceedings.neurips.cc/paper_files/pape...

2018
[4]

10 Preprint

G. Lample and F. Charton, “Deep Learning for Symbolic Mathematics,” 2019, version Number: 1. [Online]. Available: https://arxiv.org/abs/1912.01412

work page arXiv 2019
[5]

Deep symbolic regression for recurrent sequences

S. d’Ascoli, P.-A. Kamienny, G. Lample, and F. Charton, “Deep Symbolic Regression for Recurrent Sequences,” inProceedings of the 39th International Conference on Machine Learning. PMLR, 2022, version Number: 2. [Online]. Available: https://arxiv.org/abs/2201.04600

work page arXiv 2022
[6]

Deep Generative Symbolic Regression with Monte-Carlo-Tree-Search,

P.-A. Kamienny, G. Lample, S. Lamprier, and M. Virgolin, “Deep Generative Symbolic Regression with Monte-Carlo-Tree-Search,” inProceedings of the 40th International Conference on Machine Learning, ser. Proceedings of Machine Learning Research, A. Krause, E. Brunskill, K. Cho, B. Engelhardt, S. Sabato, and J. Scarlett, Eds., vol. 202. PMLR, Jul. 2023, pp. ...

2023
[7]

Predicting Ordinary Differential Equations with Transformers,

S. Becker, M. Klein, A. Neitz, G. Parascandolo, and N. Kilbertus, “Predicting Ordinary Differential Equations with Transformers,” inProceedings of the 40th International Conference on Machine Learning, ser. Proceedings of Machine Learning Research, A. Krause, E. Brunskill, K. Cho, B. Engelhardt, S. Sabato, and J. Scarlett, Eds., vol. 202. PMLR, Jul. 2023,...

2023
[8]

J. R. Koza,Genetic programming. 1: On the programming of computers by means of natural selection, ser. Complex adaptive systems. Cambridge, Mass.: MIT Press, 1992

1992
[9]

Automated reverse engineering of nonlinear dynamical systems,

J. Bongard and H. Lipson, “Automated reverse engineering of nonlinear dynamical systems,”Proceedings of the National Academy of Sciences, vol. 104, no. 24, pp. 9943–9948, Jun. 2007. [Online]. Available: https://pnas.org/doi/full/10.1073/pnas.0609476104

work page doi:10.1073/pnas.0609476104 2007
[10]

Solving differential equations with genetic programming,

I. G. Tsoulos and I. E. Lagaris, “Solving differential equations with genetic programming,”Genetic Programming and Evolvable Machines, vol. 7, no. 1, pp. 33–54, Mar. 2006. [Online]. Available: http://link.springer.com/10.1007/s10710-006-7009-y

work page doi:10.1007/s10710-006-7009-y 2006
[11]

Cranmer, A

M. Cranmer, A. Sanchez-Gonzalez, P. Battaglia, R. Xu, K. Cranmer, D. Spergel, and S. Ho, “Discovering Symbolic Models from Deep Learning with Inductive Biases,” 2020, version Number: 2. [Online]. Available: https://arxiv.org/abs/2006.11287

work page arXiv 2020
[12]

Interpretable Machine Learning for Science with PySR and SymbolicRegression.jl

M. Cranmer, “Interpretable Machine Learning for Science with PySR and SymbolicRegression.jl,” 2023, version Number: 3. [Online]. Available: https://arxiv.org/abs/2305.01582

work page internal anchor Pith review arXiv 2023
[13]

Probabilistic grammars for equation discovery,

J. Brence, L. Todorovski, and S. Džeroski, “Probabilistic grammars for equation discovery,”Knowledge-Based Systems, vol. 224, p. 107077, Jul. 2021. [Online]. Available: https://linkinghub.elsevier.com/retrieve/pii/ S0950705121003403

2021
[14]

URL https: //doi.org/10.1007/s10994-024-06522-1

N. Omejc, B. Gec, J. Brence, L. Todorovski, and S. Džeroski, “Probabilistic grammars for modeling dynamical systems from coarse, noisy, and partial data,”Machine Learning, vol. 113, no. 10, pp. 7689–7721, Oct. 2024. [Online]. Available: https://link.springer.com/10.1007/s10994-024-06522-1

work page doi:10.1007/s10994-024-06522-1 2024
[15]

arXiv preprint arXiv:1912.04871 (2019)

B. K. Petersen, M. Landajuela, T. N. Mundhenk, C. P. Santiago, S. K. Kim, and J. T. Kim, “Deep symbolic regression: Recovering mathematical expressions from data via risk-seeking policy gradients,” in Proceedings of the International Conference on Learning Representations (ICLR), 2019. [Online]. Available: https://arxiv.org/abs/1912.04871

work page arXiv 2019
[16]

A Reinforcement Learning Approach to Domain-Knowledge Inclusion Using Grammar Guided Symbolic Regression,

L. Crochepierre, L. Boudjeloud, and V . Barbesant, “A Reinforcement Learning Approach to Domain-Knowledge Inclusion Using Grammar Guided Symbolic Regression,”ArXiv, vol. abs/2202.04367, 2022. [Online]. Available: https://api.semanticscholar.org/CorpusID:246679968

work page arXiv 2022
[17]

Solving the symbolic regression problem with tree-adjunct grammar guided genetic programming: the comparative results,

N. Hoai, R. McKay, D. Essam, and R. Chau, “Solving the symbolic regression problem with tree-adjunct grammar guided genetic programming: the comparative results,” inProceedings of the 2002 Congress on Evolutionary Computation. CEC’02 (Cat. No.02TH8600), vol. 2. Honolulu, HI, USA: IEEE, 2002, pp. 1326–1331. [Online]. Available: http://ieeexplore.ieee.org/d...

work page arXiv 2002
[18]

Learning Equations for Extrapolation and Control,

S. Sahoo, C. Lampert, and G. Martius, “Learning Equations for Extrapolation and Control,” inProceedings of the 35th International Conference on Machine Learning, ser. Proceedings of Machine Learning 9 Yu et al. Neuro-Symbolic ODE Discovery with Latent Grammar FlowPreprint April 2026 Research, J. Dy and A. Krause, Eds., vol. 80. PMLR, Jul. 2018, pp. 4442–4...

2026
[19]

NOMTO: Neural Operator-based symbolic Model approximaTion and discOvery,

S. Garmaev, S. Mishra, and O. Fink, “NOMTO: Neural Operator-based symbolic Model approximaTion and discOvery,” 2025, version Number: 1. [Online]. Available: https://arxiv.org/abs/2501.08086

work page arXiv 2025
[20]

Abhimanyu Das, Weihao Kong, Rajat Sen, and Yichen Zhou

S. L. Brunton, J. L. Proctor, and J. N. Kutz, “Discovering governing equations from data by sparse identification of nonlinear dynamical systems,”Proceedings of the National Academy of Sciences, vol. 113, no. 15, pp. 3932–3937, Apr. 2016. [Online]. Available: https://pnas.org/doi/full/10.1073/pnas.1517384113

work page doi:10.1073/pnas.1517384113 2016
[21]

Kaheman, J

K. Kaheman, J. N. Kutz, and S. L. Brunton, “SINDy-PI: a robust algorithm for parallel implicit sparse identification of nonlinear dynamics,”Proceedings of the Royal Society A: Mathematical, Physical and Engineering Sciences, vol. 476, no. 2242, p. 20200279, Oct. 2020. [Online]. Available: https://royalsocietypublishing.org/doi/10.1098/rspa.2020.0279

work page doi:10.1098/rspa.2020.0279 2020
[22]

D-CIPHER: Discovery of Closed-form Partial Differential Equations,

K. Kacprzyk, Z. Qian, and M. van der Schaar, “D-CIPHER: Discovery of Closed-form Partial Differential Equations,” inAdvances in Neural Information Processing Systems, A. Oh, T. Naumann, A. Globerson, K. Saenko, M. Hardt, and S. Levine, Eds., vol. 36. Curran Associates, Inc., 2023, pp. 27 609–27 644. [Online]. Available: https://proceedings.neurips.cc/pape...

2023
[23]

Grammar-based ordinary differential equation discovery,

K. Yu, E. Chatzi, and G. Kissas, “Grammar-based ordinary differential equation discovery,”Mechanical Systems and Signal Processing, vol. 240, p. 113395, Nov. 2025. [Online]. Available: https://linkinghub.elsevier.com/ retrieve/pii/S0888327025010969

2025
[24]

GenSR: Symbolic Regression Based in Equation Generative Space,

Q. Li, Y . Hu, J. Liu, and Y . Chen, “GenSR: Symbolic Regression Based in Equation Generative Space,” in Proceedings of the International Conference on Learning Representations (ICLR), 2026, version Number: 1

2026
[25]

AlphaEvolve: A coding agent for scientific and algorithmic discovery,

A. Novikov, N. V ˜u, M. Eisenberger, E. Dupont, P.-S. Huang, A. Z. Wagner, S. Shirobokov, B. Kozlovskii, F. J. R. Ruiz, A. Mehrabian, M. P. Kumar, A. See, S. Chaudhuri, G. Holland, A. Davies, S. Nowozin, P. Kohli, and M. Balog, “AlphaEvolve: A coding agent for scientific and algorithmic discovery,” 2025, version Number:

2025
[26]

AlphaEvolve: A coding agent for scientific and algorithmic discovery

[Online]. Available: https://arxiv.org/abs/2506.13131

work page internal anchor Pith review arXiv
[27]

Llm-sr: Scientific equation discovery via programming with large language models

P. Shojaee, K. Meidani, S. Gupta, A. B. Farimani, and C. K. Reddy, “LLM-SR: Scientific Equation Discovery via Programming with Large Language Models,” inProceedings of the International Conference on Learning Representations (ICLR), 2025, version Number: 3. [Online]. Available: https://arxiv.org/abs/2404.18400

work page arXiv 2025
[28]

d’Ascoli, S

S. d’Ascoli, S. Becker, A. Mathis, P. Schwaller, and N. Kilbertus, “ODEFormer: Symbolic Regression of Dynamical Systems with Transformers,” inThe Twelfth International Conference on Learning Representations, ICLR 2024, Vienna, Austria, May 7-11, 2024. OpenReview.net, 2023, version Number: 1. [Online]. Available: https://arxiv.org/abs/2310.05573

work page arXiv 2024
[29]

Chomsky,Aspects of the theory of syntax

N. Chomsky,Aspects of the theory of syntax. The MIT Press, Mar. 1969. [Online]. Available: https://www.jstor.org/stable/j.ctt17kk81z

1969
[30]

Grammar Variational Autoencoder,

M. J. Kusner, B. Paige, and J. M. Hernández-Lobato, “Grammar Variational Autoencoder,” inProceedings of the 34th International Conference on Machine Learning, ser. Proceedings of Machine Learning Research, D. Precup and Y . W. Teh, Eds., vol. 70. PMLR, Aug. 2017, pp. 1945–1954. [Online]. Available: https://proceedings.mlr.press/v70/kusner17a.html

2017
[31]

Finite scalar quantization: Vq-vae made simple.arXiv preprint arXiv:2309.15505, 2023

F. Mentzer, D. Minnen, E. Agustsson, and M. Tschannen, “Finite Scalar Quantization: VQ-V AE Made Simple,” 2023, version Number: 2. [Online]. Available: https://arxiv.org/abs/2309.15505

work page arXiv 2023
[32]

Neural Discrete Representation Learning

A. v. d. Oord, O. Vinyals, and K. Kavukcuoglu, “Neural Discrete Representation Learning,” 2017, version Number: 2. [Online]. Available: https://arxiv.org/abs/1711.00937

work page Pith review arXiv 2017
[33]

Quantifying Behavioural Distance Between Mathematical Expressions,

S. Mežnar, S. Džeroski, and L. Todorovski, “Quantifying Behavioural Distance Between Mathematical Expressions,” 2024, version Number: 1. [Online]. Available: https://arxiv.org/abs/2408.11515

work page arXiv 2024
[34]

Mathematical Methods of Organizing and Planning Production,

L. V . Kantorovich, “Mathematical Methods of Organizing and Planning Production,”Management Science, vol. 6, no. 4, pp. 366–422, Jul. 1960. [Online]. Available: https://pubsonline.informs.org/doi/10.1287/mnsc.6.4. 366

work page doi:10.1287/mnsc.6.4 1960
[35]

Adam: A Method for Stochastic Optimization

D. P. Kingma and J. Ba, “Adam: A Method for Stochastic Optimization,” 2014, version Number: 9. [Online]. Available: https://arxiv.org/abs/1412.6980

work page internal anchor Pith review Pith/arXiv arXiv 2014
[36]

In: 29th ACM International Conference on Ar- chitectural Support for Programming Languages and Operating Systems, Volume 2 (ASPLOS ’24)

J. Ansel, E. Yang, H. He, N. Gimelshein, A. Jain, M. V oznesensky, B. Bao, P. Bell, D. Berard, E. Burovski, G. Chauhan, A. Chourdia, W. Constable, A. Desmaison, Z. DeVito, E. Ellison, W. Feng, J. Gong, M. Gschwind, B. Hirsh, S. Huang, K. Kalambarkar, L. Kirsch, M. Lazos, M. Lezcano, Y . Liang, J. Liang, Y . Lu, C. K. Luk, B. Maher, Y . Pan, C. Puhrsch, M....

work page doi:10.1145/3620665.3640366 2026
[37]

Unlocking Guidance for Discrete State-Space Diffusion and Flow Models,

H. Nisonoff, J. Xiong, S. Allenspach, and J. Listgarten, “Unlocking Guidance for Discrete State-Space Diffusion and Flow Models,” inInternational Conference on Learning Representations, Y . Yue, A. Garg, N. Peng, F. Sha, and R. Yu, Eds., vol. 2025, 2025, pp. 36 052–36 106. [Online]. Available: https: //proceedings.iclr.cc/paper_files/paper/2025/file/59725...

2025
[38]

Gener- ative flows on discrete state-spaces: Enabling multimodal flows with applications to protein co-design.arXiv preprint arXiv:2402.04997, 2024

A. Campbell, J. Yim, R. Barzilay, T. Rainforth, and T. Jaakkola, “Generative Flows on Discrete State-Spaces: Enabling Multimodal Flows with Applications to Protein Co-Design,” inProceedings of the 41st International Coference on Machine Learning. ICML.Vienna, Austria: PMLR 235, 2024. [Online]. Available: https://arxiv.org/abs/2402.04997

work page arXiv 2024
[39]

A Simplex Method for Function Minimization,

J. A. Nelder and R. Mead, “A Simplex Method for Function Minimization,”The Computer Journal, vol. 7, no. 4, pp. 308–313, Jan. 1965. [Online]. Available: https://academic.oup.com/comjnl/article-lookup/doi/10. 1093/comjnl/7.4.308

1965
[40]

SciPy 1.0: fundamental algorithms for scientific computing in Python,

P. Virtanen, R. Gommers, T. E. Oliphant, M. Haberland, T. Reddy, D. Cournapeau, E. Burovski, P. Peterson, W. Weckesser, J. Bright, S. J. Van Der Walt, M. Brett, J. Wilson, K. J. Millman, N. Mayorov, A. R. J. Nelson, E. Jones, R. Kern, E. Larson, C. J. Carey, I. Polat, Y . Feng, E. W. Moore, J. VanderPlas, D. Laxalde, J. Perktold, R. Cimrman, I. Henriksen,...

2020
[41]

The general problem of the stability of motion,

A. M. Lyapunov, “The general problem of the stability of motion,”International Journal of Control, vol. 55, no. 3, pp. 531–534, Mar. 1992. [Online]. Available: http://www.tandfonline.com/doi/abs/10.1080/ 00207179208934253

1992
[42]

H. K. Khalil and J. W. Grizzle,Nonlinear systems. Prentice hall Upper Saddle River, NJ, 2002, vol. 3

2002
[43]

Fast and Accurate Deep Network Learning by Exponential Linear Units (ELUs)

D.-A. Clevert, T. Unterthiner, and S. Hochreiter, “Fast and Accurate Deep Network Learning by Exponential Linear Units (ELUs),” 2015, version Number: 5. [Online]. Available: https://arxiv.org/abs/1511.07289 11 Yu et al. Neuro-Symbolic ODE Discovery with Latent Grammar FlowPreprint April 2026 Appendix A Details on the benchmarks This appendix presents the ...

work page Pith review arXiv 2015