Pith. sign in

REVIEW 3 major objections 4 minor 37 references

Adaptive Stability-Constrained Neural Differential Equations for Controlled Dynamical Systems with Unknown Inputs

T0 review · 3 major / 4 minor · reviewed 2026-08-11 · deepseek-v4-flash

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

Pith's one-line read An input-dependent contraction metric gives controlled neural ODEs a quantitative bound on prediction sensitivity to differing controls and disturbances.

desk verdict The theorem isn't proven as written, but the transparent protocol and the input-conditioned metric idea make it worth a rigorous referee. read the letter →

arxiv 2608.09404 v1 pith:FB34A3RD submitted 2026-08-10 eess.SY cs.SY

Syed Pouladi This is my paper · ORCID
classification eess.SYcs.SY MSC 34D2393D2593B3068T07
keywords neuralordinarydifferentialequationscontractionanalysisincrementalstabilityinput-to-statenonlinearsystemidentificationpermanent-magnetsynchronousmotorinput-dependentmetricunknowndisturbances
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 an adaptive stability-constrained neural differential equation (AS-NDE) that jointly learns a controlled vector field and a state- and input-dependent Riemannian metric, so that predicted trajectories stay close even when the inputs used at rollout differ from those seen in training. Its central analytical claim is Theorem 1, an incremental input-to-state contraction bound: the metric distance between two trajectories decays exponentially at rate $\lambda$, up to additive gains from differences in the measured controls and the unknown disturbances. The paper's distinctive move is to keep the total derivative of the input-dependent metric, including the $\partial M/\partial u$ term, inside the training residual, a term it says heuristic stability regularizers omit. The draft is explicit that its numerical tables and curves are illustrative synthetic placeholders and that no measured data or executed training runs were available, so the contribution is the method and the proof, not empirical evidence. A sympathetic reader would care because, if the theorem's premise can be made to hold, neural models of nonlinear systems would come with a quantitative prediction-sensitivity guarantee instead of an ad hoc regularizer.

What carries the argument

The load-bearing object is the input-conditioned Riemannian metric $M_\phi(x,u) = L_\phi(x,u) L_\phi(x,u)^\top + \epsilon_M I$, where $L_\phi$ is a lower-triangular network output whose diagonal uses softplus plus a floor, so positive definiteness is guaranteed by construction. Alongside it, the contraction residual $S_\Theta = \dot M + A_\theta^\top M + M A_\theta + 2\lambda M$ is evaluated at collocation points, with the total derivative $\dot M = \sum_i \partial M/\partial x_i f_{\theta,i} + \sum_j \partial M/\partial u_j \dot u_j$ retained. Training penalizes $\mathrm{softplus}(\lambda_{\max}(S_\Theta)/\tau)\tau$, and Theorem 1's premise is exactly the negative semidefiniteness of $S_\Theta$ on the whole domain. The metric's job is to define a distance in which the variational dynamics are contractive; the residual's job is to encode the differential inequality that makes the exponential bound hold.

What would settle it

Train the model on the forced Duffing benchmark exactly as the protocol specifies, then from the same initial condition run two rollouts under different test command sequences and disturbances and compare the measured Riemannian distance to the right side of (12); a persistent overshoot of the bound would show that the contraction premise, or the way it is enforced, does not hold.

Watch

Extended reading notes

Core claim

The core discovery is Theorem 1. If Assumption 1 holds and the contraction inequality $\dot M + A_\theta^\top M + M A_\theta \preceq -2\lambda M$ is satisfied throughout a forward-invariant domain, then the Riemannian distance between two trajectories of the learned model obeys $d_M(t) \le e^{-\lambda t} d_M(0) + \int_0^t e^{-\lambda(t-\tau)} [b_u \|u_1-u_2\| + b_w \|w_1-w_2\|]\,d\tau$, with gains $b_u = \sup \|M^{1/2} B_\theta\|_2$ and $b_w = \sup \|M^{1/2} E\|_2$. The Euclidean bound (13) follows from metric equivalence. The deliberate inclusion of the total derivative of the metric, especially the $\partial M/\partial u$ term, is what makes the inequality correct for an input-dependent metric; the paper identifies this as easily omitted in heuristic regularizers. The stated object of the certificate is the learned model, not the unknown plant; Remark 1 says that a plant-level claim would need a verified bound on the modeling error $f^\star - f_\theta$ or independent validation.

Load-bearing premise

The load-bearing premise is that the contraction inequality $\dot M + A^\top M + M A \preceq -2\lambda M$ holds throughout the forward-invariant domain, including along paths connecting two different input trajectories, yet the training loss only samples this condition at finitely many collocation points and the metric's total derivative is not well-defined for two differing inputs.

Editorial extensions

If this is right

  • If the contraction premise can be verified, rollout error under new inputs is no longer an open-ended risk: the Euclidean form (13) gives a closed-form bound in terms of initial mismatch, input mismatch, disturbance mismatch, the rate $\lambda$, and the metric condition number.
  • The certificate is for the learned model, not the real plant; the paper explicitly says that claiming robustness of the plant requires an additional verified bound on $f^\star - f_\theta$.
  • The sampled contraction penalty is an empirical regularizer, and a zero violation rate on finite samples is not a proof; the prescribed certificate audit on a dense held-out grid plus an exterior shell is needed before calling the bound a guarantee.
  • For piecewise-constant commands, the $\partial M/\partial u$ term forces a hybrid treatment at jumps, and the paper proposes interval-by-interval checking rather than a single smooth certificate.
  • A negative result is meaningful under the paper's protocol: if residual violations drop without an improvement in held-out rollout error, the honest conclusion is that the certificate regularizes sensitivity without demonstrated predictive benefit.

Reading between the lines

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

  • Not stated in the paper, but the same proof could be re-run with a time-varying rate $\lambda(t)$ or with metric-dependent gains; the gain structure $b_w/\lambda$ already suggests a concrete trade-off between disturbance attenuation and stiffness.
  • Not stated in the paper, but the theorem's premise could be verified ex post on the Duffing and PMSM benchmarks using interval bound propagation, turning the sampled penalty into a real certificate; the paper names this tool in its certificate audit.
  • Not stated in the paper, but the bound (13) applies only to the learned model; combining it with an independently validated bound on $f^\star - f_\theta$ would yield a plant-level prediction-sensitivity guarantee, which is the result an end user would ultimately need.
  • Not stated in the paper, but the proposed ablations would settle a key ambiguity of the method: whether the adaptive metric itself buys the improvement, or whether the same gain comes from the contraction penalty acting on a fixed identity metric.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

3 major / 4 minor

Summary. The paper proposes Adaptive Stability-Constrained Neural Differential Equations (AS-NDE), a continuous-time controlled neural model that jointly learns a vector field f_theta(x,u) and an input-dependent Riemannian metric M_phi(x,u). Positive definiteness of the metric is enforced by a Cholesky-type construction with a floor term, and a sampled differential inequality penalizes violations of a prescribed contraction rate. The main theoretical claim is Theorem 1, an incremental input-to-state bound stating that the Riemannian distance between two trajectories of the learned model decays exponentially up to gains proportional to the differences in their controls and disturbances. An experimental protocol is specified for a forced Duffing oscillator and a PMSM model, with all numerical tables and figures explicitly labeled as illustrative placeholders pending actual experiments.

Significance. If Theorem 1 is established rigorously, the paper would provide a useful quantitative sensitivity bound for learned controlled neural ODEs, explicitly accounting for the total derivative of an input-dependent metric, a term often omitted in heuristic stability regularizers. The paper is unusually transparent: it separates the method from the unexecuted experiments, labels all placeholder data as such, and includes a thoughtful threats-to-validity section. These are strengths. However, the central proof as written is incomplete, and the premise of the theorem is never certified for the learned model; therefore the contribution is currently a plausible methodology with a conditional guarantee, not a proven result. The reproducibility protocol and the honest treatment of limitations are commendable and should be preserved in a revised version.

major comments (3)
  1. [Section V, Eq. (10) and proof of Theorem 1] The proof states that for two solutions with different inputs (u1,w1) and (u2,w2), the variational dynamics are delta x_dot = A_theta delta x + B_theta delta u + E delta w, and then differentiates V = delta x^T M delta x. For delta x = x1 - x2, this equation is not exact: the difference f(x1,u1) - f(x2,u2) is a path integral of A_theta and B_theta over the connecting segment plus a remainder that is not controlled by the pointwise contraction inequality. To make the argument rigorous, one must define a virtual flow x_s(t) satisfying x_s_dot = f(x_s,u_s) + E w_s with u_s = (1-s)u1 + s u2 and w_s = (1-s)w1 + s w2, and set delta x_s = partial x_s / partial s; only then is the linearized equation exact. The proof as written is a linearization argument and does not establish the claimed bound for finite trajectory differences. This is the paper's central contribution, so the proof must be repaired and the virtual-flow construction stated explicitly.
  2. [Section V, Eq. (5) and Theorem 1 premise] The contraction inequality is assumed to hold throughout a forward-invariant domain, but because the metric depends on u, its total derivative in Eq. (5) contains the term partial M / partial u times u_dot. When comparing two different input trajectories, the connecting path must specify the input derivative u_s_dot = (1-s) u1_dot + s u2_dot, and the inequality must hold for all s in [0,1] along the virtual flow. The manuscript does not state this strengthened premise, and the training loss (7) only samples the inequality at a finite set of collocation points. Consequently, Theorem 1's bound is conditional on a property that is never certified for the learned model. The paper's own Remark 1 concedes this point, but the theorem as stated is not established for arbitrary admissible inputs unless the premise is explicitly extended to the virtual paths and verified.
  3. [Section VI-E and Table I] Because the contraction loss L_ctr directly penalizes lambda_max(S), any violation rate reported for AS-NDE is a fitted quantity and cannot serve as independent evidence that the theorem's premise holds. The certificate audit described in Section VI-E is the appropriate check, but it must report the full distribution of lambda_max(S) on dense held-out grids and exterior shells, together with the metric condition numbers, rather than a single violation percentage. The placeholder numbers in Table I are correctly labeled as illustrative, but the relationship between the penalty and the reported violation metric should be made explicit so that readers do not interpret low violation rates as a verification of the theoretical condition.
minor comments (4)
  1. [Section VI-A, Eq. (15)] The second line of the Duffing equation contains a stray 'quad' that should be removed.
  2. [Section V, proof of Theorem 1] The notation 'D+ sqrt(V)' is used without definition; please define the upper Dini derivative and spell out the integration step from the differential inequality to Eq. (12).
  3. [Section V, Theorem 1 statement] The Riemannian distance d_M(t) is computed between points (x1(t),u1(t)) and (x2(t),u2(t)) using a metric that depends on both arguments; the paper should define this distance explicitly as the infimum over smooth paths in the joint (x,u) space to avoid ambiguity.
  4. [Section VI-D] The phrase 'four-layer multilayer perceptron' is redundant; replace it with 'four-layer perceptron' or 'four-layer MLP' for clarity.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: Theorem 1 is a conditional derivation from an explicit contraction inequality, and all numerical evidence is explicitly labeled as placeholder.

full rationale

The paper's derivation chain is self-contained and transparently conditional. The metric M is defined by construction to be positive definite (Eq. 4), and the contraction residual S is defined in Eq. 6. Theorem 1 states: if Assumption 1 holds and the matrix inequality Mdot + A^T M + M A <= -2lambda M holds throughout a forward-invariant domain, then the Riemannian distance satisfies the stated bound. This is a mathematical implication, not a prediction obtained by fitting. The proof uses a variational calculation and standard contraction/ISS arguments; no fitted parameter is renamed as a prediction. The experimental section contains no executed results: Table I and all figures are explicitly marked as 'illustrative synthetic draft placeholders', and Remark 1 concedes that 'sampled residual penalties are empirical regularizers unless completed by formal verification.' There are no load-bearing self-citations: references [35] and [36] are prior work by other authors, and they are cited as motivation, not as justification for the theorem's premises. The main technical concern is a proof gap: Eq. (10) is exact for tangent variations along a connecting path, not for the finite difference x1-x2, and the metric's total derivative at interpolated inputs is not verified; however, that is a correctness issue, not circularity. The manuscript therefore receives a circularity score of 0.

Assumptions & free parameters 4 free parameters · 5 assumptions · 1 invented entities

The central claim rests on the unverified premise that the contraction inequality holds over the whole domain, and on hyperparameters that are not tuned. The metric is a learned construct rather than an independently measured quantity.

free parameters (4)
  • Target contraction rate lambda = Chosen by validation in [0.05, 0.5]
    Appears in the contraction residual Eq. (6) and final bound Eq. (12); selected on validation set per Table II.
  • Metric spectral bounds m and bar m = Not fixed numerically in the draft; enforced through L_metric in Eq. (9)
    Constants in Eq. (3) and in the Euclidean bound Eq. (13); chosen by hand, not learned.
  • Metric floor epsilon_M = 1e-3
    Added in Eq. (4) to guarantee positive definiteness; hand-chosen constant.
  • Loss weights rho_c, rho_M, rho_J = Not specified numerically
    Weights in the objective Eq. (8); hyperparameters that must be tuned on validation data.
assumptions (5)
  • domain assumption f_theta and M_phi are continuously differentiable on a compact training domain, u is absolutely continuous, and mI <= M_phi <= bar m I.
    Assumption 1 in Section III; used for differentiating V and for metric equivalence in Eq. (13).
  • domain assumption The learned model dot x = f_theta(x,u) represents the plant and E is a known or conservative disturbance injection matrix.
    Section III; the certificate applies to the learned model, and no bound on the plant mismatch f_star - f_theta is provided or verified.
  • ad hoc to paper The contraction inequality dot M + A^T M + M A <= -2 lambda M holds throughout a forward-invariant domain.
    Premise of Theorem 1 in Section V; only sampled via the softplus penalty L_ctr in Eq. (7), and for u-dependent metrics the condition is under-specified along paths connecting different input trajectories.
  • domain assumption Solutions remain inside the forward-invariant domain for all times considered.
    Theorem 1 requires trajectories to stay in the domain; no invariance certificate is provided.
  • standard math Standard differential calculus, Cauchy-Schwarz, and geodesic distance minimization are valid along the connecting path.
    Used in the proof of Theorem 1 from Eq. (14) to Eq. (12); not machine-checked.
invented entities (1)
  • Input-dependent Riemannian metric M_phi(x,u)
    purpose: Defines the time-varying distance measure used for the contraction certificate and incremental ISS bound
    A learned mathematical object, not a physical entity. It is trained on the same data as the vector field and its contraction property is only sampled, so it provides no independent evidence outside the paper.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Adaptive Stability-Constrained Neural Differential Equations for Controlled Dynamical Systems with Unknown Inputs." pith.science (2026). https://pith.science/paper/FB34A3RD

@misc{pith2026260809404,
  author       = {Pith},
  title        = {Pith review of: Adaptive Stability-Constrained Neural Differential Equations for Controlled Dynamical Systems with Unknown Inputs},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/FB34A3RD}},
  note         = {Machine review of arXiv:2608.09404}
}
read the original abstract

Continuous-time neural models are attractive for identifying nonlinear systems, but a small one-step error can grow rapidly when a learned vector field is rolled out under inputs that differ from those used for training. This paper develops an adaptive stability-constrained neural differential equation (AS-NDE) for systems with measured controls and unmatched, unknown perturbations. The nominal vector field and a state--input-dependent Riemannian metric are learned jointly. Positive definiteness is enforced by construction, while a sampled differential inequality penalizes violations of a prescribed contraction rate. An incremental input-to-state bound is derived: the distance between two trajectories decays exponentially up to gains determined by differences in their controls and disturbances. The statement explicitly accounts for the time derivative of an input-dependent metric, a term that is easily omitted in heuristic stability regularizers. We give a reproducible evaluation protocol for a forced Duffing oscillator and a permanent-magnet synchronous motor (PMSM) model. Because no measured data or executed training runs accompany this draft, all numerical curves and tables are clearly identified as illustrative synthetic placeholders; their PGFPlots coordinates are embedded in the source for direct replacement. The resulting manuscript is intended as a technically consistent starting point, not as evidence of empirical superiority before the prescribed experiments are run.

Figures

Figures reproduced from arXiv: 2608.09404 by the authors.

Figure 1
Figure 1. Illustrative synthetic Duffing rollout. Coordinates are placeholders, [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. Illustrative synthetic horizon-error curves. Replace the embedded [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

37 extracted references · 26 canonical work pages

  1. [1]

    Neural ordinary differential equations,

    R. T. Q. Chen, Y . Rubanova, J. Bettencourt, and D. Duvenaud, “Neural ordinary differential equations,” inAdv. Neural Inf. Process. Syst., vol. 31, 2018

    work page 2018

  2. [2]

    Augmented neural ODEs,

    E. Dupont, A. Doucet, and Y . W. Teh, “Augmented neural ODEs,” in Adv. Neural Inf. Process. Syst., vol. 32, 2019

    work page 2019

  3. [3]

    Latent ordinary differential equations for irregularly-sampled time series,

    Y . Rubanova, R. T. Q. Chen, and D. Duvenaud, “Latent ordinary differential equations for irregularly-sampled time series,” inAdv. Neural Inf. Process. Syst., vol. 32, 2019

    work page 2019

  4. [4]

    Neural controlled differential equations for irregular time series,

    P. Kidger, J. Morrill, J. Foster, and T. Lyons, “Neural controlled differential equations for irregular time series,” inAdv. Neural Inf. Process. Syst., vol. 33, 2020, pp. 6696–6707

    work page 2020

  5. [5]

    Universal differential equations for scientific machine learning,

    C. Rackauckaset al., “Universal differential equations for scientific machine learning,”arXiv:2001.04385, 2020

    arXiv 2001

  6. [6]

    Physics-informed neural networks: A deep learning framework for solving forward and inverse problems involving nonlinear partial differential equations,

    M. Raissi, P. Perdikaris, and G. E. Karniadakis, “Physics-informed neural networks: A deep learning framework for solving forward and inverse problems involving nonlinear partial differential equations,”J. Comput. Phys., vol. 378, pp. 686–707, 2019

    2019

  7. [7]

    Physics-informed machine learning,

    G. E. Karniadakis, I. G. Kevrekidis, L. Lu, P. Perdikaris, S. Wang, and L. Yang, “Physics-informed machine learning,”Nat. Rev. Phys., vol. 3, pp. 422–440, 2021

    work page 2021

  8. [8]

    Discovering governing equations from data by sparse identification of nonlinear dynamical systems,

    S. L. Brunton, J. L. Proctor, and J. N. Kutz, “Discovering governing equations from data by sparse identification of nonlinear dynamical systems,”Proc. Natl. Acad. Sci. USA, vol. 113, no. 15, pp. 3932–3937, 2016

    work page 2016

Show all 37 references
  1. [9]

    Data-driven discovery of partial differential equations,

    S. H. Rudy, S. L. Brunton, J. L. Proctor, and J. N. Kutz, “Data-driven discovery of partial differential equations,”Sci. Adv., vol. 3, no. 4, Art. no. e1602614, 2017

    2017

  2. [10]

    Deep learning for universal linear embeddings of nonlinear dynamics,

    B. Lusch, J. N. Kutz, and S. L. Brunton, “Deep learning for universal linear embeddings of nonlinear dynamics,”Nat. Commun., vol. 9, Art. no. 4950, 2018

    2018

  3. [11]

    Linearly recurrent autoencoder networks for learning dynamics,

    S. E. Otto and C. W. Rowley, “Linearly recurrent autoencoder networks for learning dynamics,”SIAM J. Appl. Dyn. Syst., vol. 18, no. 1, pp. 558–593, 2019

    2019

  4. [12]

    Learning Koopman invariant subspaces for dynamic mode decomposition,

    N. Takeishi, Y . Kawahara, and T. Yairi, “Learning Koopman invariant subspaces for dynamic mode decomposition,” inAdv. Neural Inf. Pro- cess. Syst., vol. 30, 2017

    2017

  5. [13]

    Hamiltonian neural net- works,

    S. Greydanus, M. Dzamba, and J. Yosinski, “Hamiltonian neural net- works,” inAdv. Neural Inf. Process. Syst., vol. 32, 2019

    2019

  6. [14]

    Lagrangian neural networks,

    M. Cranmer, S. Greydanus, S. Hoyer, P. Battaglia, D. Spergel, and S. Ho, “Lagrangian neural networks,” inICLR Workshop on Integration of Deep Neural Models and Differential Equations, 2020

    2020

  7. [15]

    On contraction analysis for non-linear systems,

    W. Lohmiller and J.-J. E. Slotine, “On contraction analysis for non-linear systems,”Automatica, vol. 34, no. 6, pp. 683–696, 1998

    1998

  8. [16]

    A differential Lyapunov framework for contraction analysis,

    F. Forni and R. Sepulchre, “A differential Lyapunov framework for contraction analysis,”IEEE Trans. Autom. Control, vol. 59, no. 3, pp. 614–628, 2014

    2014

  9. [17]

    Control contraction metrics: Convex and intrinsic criteria for nonlinear feedback design,

    I. R. Manchester and J.-J. E. Slotine, “Control contraction metrics: Convex and intrinsic criteria for nonlinear feedback design,”IEEE Trans. Autom. Control, vol. 62, no. 6, pp. 3046–3053, 2017

    2017

  10. [18]

    A Lyapunov approach to incremental stability properties,

    D. Angeli, “A Lyapunov approach to incremental stability properties,” IEEE Trans. Autom. Control, vol. 47, no. 3, pp. 410–421, 2002

    2002

  11. [19]

    Smooth stabilization implies coprime factorization,

    E. D. Sontag, “Smooth stabilization implies coprime factorization,”IEEE Trans. Autom. Control, vol. 34, no. 4, pp. 435–443, 1989

    1989

  12. [20]

    H. K. Khalil,Nonlinear Systems, 3rd ed. Upper Saddle River, NJ, USA: Prentice Hall, 2002

    2002

  13. [21]

    J.-J. E. Slotine and W. Li,Applied Nonlinear Control. Englewood Cliffs, NJ, USA: Prentice Hall, 1991

    1991

  14. [22]

    Ljung,System Identification: Theory for the User, 2nd ed

    L. Ljung,System Identification: Theory for the User, 2nd ed. Upper Saddle River, NJ, USA: Prentice Hall, 1999

    1999

  15. [23]

    S. A. Billings,Nonlinear System Identification: NARMAX Methods in the Time, Frequency, and Spatio-Temporal Domains. Chichester, U.K.: Wiley, 2013

    2013

  16. [24]

    Stable architectures for deep neural net- works,

    E. Haber and L. Ruthotto, “Stable architectures for deep neural net- works,”Inverse Problems, vol. 34, no. 1, Art. no. 014004, 2018

    2018

  17. [25]

    Stable recurrent models,

    J. Miller and M. Hardt, “Stable recurrent models,” inProc. Int. Conf. Learn. Representations, 2019

    2019

  18. [26]

    Lipschitz bounded equilibrium net- works,

    M. Revay and I. R. Manchester, “Lipschitz bounded equilibrium net- works,”arXiv:2010.01732, 2020

    2010 arXiv

  19. [27]

    The Lyapunov neural network: Adaptive stability certification for safe learning of dynamical systems,

    S. M. Richards, F. Berkenkamp, and A. Krause, “The Lyapunov neural network: Adaptive stability certification for safe learning of dynamical systems,” inProc. Conf. Robot Learn., 2018, pp. 466–476

    2018

  20. [28]

    Safe control with learned certificates: A survey of neural Lyapunov, barrier, and contraction methods for robotics and control,

    C. Dawson, S. Gao, and C. Fan, “Safe control with learned certificates: A survey of neural Lyapunov, barrier, and contraction methods for robotics and control,”IEEE Trans. Robot., vol. 39, no. 3, pp. 1749–1767, 2023

    2023

  21. [29]

    Control barrier function based quadratic programs for safety critical systems,

    A. D. Ames, X. Xu, J. W. Grizzle, and P. Tabuada, “Control barrier function based quadratic programs for safety critical systems,”IEEE Trans. Autom. Control, vol. 62, no. 8, pp. 3861–3876, 2017

    2017

  22. [30]

    Modeling, simulation, and analysis of permanent-magnet motor drives, Part I: The permanent-magnet syn- chronous motor drive,

    P. Pillay and R. Krishnan, “Modeling, simulation, and analysis of permanent-magnet motor drives, Part I: The permanent-magnet syn- chronous motor drive,”IEEE Trans. Ind. Appl., vol. 25, no. 2, pp. 265– 273, 1989

    1989

  23. [31]

    Krishnan,Permanent Magnet Synchronous and Brushless DC Motor Drives

    R. Krishnan,Permanent Magnet Synchronous and Brushless DC Motor Drives. Boca Raton, FL, USA: CRC Press, 2010

    2010

  24. [32]

    Control barrier functions: Theory and applications,

    A. D. Ames, S. Coogan, M. Egerstedt, G. Notomista, K. Sreenath, and P. Tabuada, “Control barrier functions: Theory and applications,” inProc. Eur. Control Conf., 2019, pp. 3420–3431

    2019

  25. [33]

    AntisymmetricRNN: A dynamical system view on recurrent neural networks,

    B. Chang, M. Chen, E. Haber, and E. H. Chi, “AntisymmetricRNN: A dynamical system view on recurrent neural networks,” inProc. Int. Conf. Learn. Representations, 2019

    2019

  26. [34]

    Dissecting neural ODEs,

    S. Massaroli, M. Poli, J. Park, A. Yamashita, and H. Asama, “Dissecting neural ODEs,” inAdv. Neural Inf. Process. Syst., vol. 33, 2020, pp. 3952–3963

    2020

  27. [35]

    ICODE: Modeling dynamical systems with extrinsic input information,

    Z. Li, W. Mei, K. Yu, Y . Bai, and S. Li, “ICODE: Modeling dynamical systems with extrinsic input information,”IEEE Trans. Autom. Sci. Eng., vol. 22, pp. 14358–14370, 2025, doi: 10.1109/TASE.2025.3560450

    2025

  28. [36]

    ControlSynth neural ODEs: Modeling dynamical systems with guaranteed convergence,

    W. Mei, D. Zheng, and S. Li, “ControlSynth neural ODEs: Modeling dynamical systems with guaranteed convergence,” inAdv. Neural Inf. Process. Syst., vol. 37, 2024, doi: 10.52202/079017-3148

    2024 doi

  29. [37]

    Learning and current prediction of PMSM drive via differential neural networks,

    W. Mei, X. Wang, Y . Lu, K. Yu, and S. Li, “Learning and current prediction of PMSM drive via differential neural networks,”IEEE Trans. Circuits Syst. II, Exp. Briefs, vol. 72, no. 3, pp. 489–493, 2025, doi: 10.1109/TCSII.2025.3527024

    2025

Pith tools

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