Pith. sign in

REVIEW 4 major objections 6 minor 44 references

Deep Operator Neural Network Model Predictive Control

T0 review · 4 major / 6 minor · reviewed 2026-08-07 · deepseek-v4-flash

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

Pith's one-line read This paper claims that a modified DeepONet architecture computes all multi-step predictions of a nonlinear MIMO system in one forward pass and is a universal approximator for such mappings.

desk verdict A genuinely useful architecture and a fair empirical study, but the universal approximation proof has a missing bridge from scalar outputs to the shared-trunk multi-output setting. read the letter →

arxiv 2505.18008 v1 pith:M4DPJDJX submitted 2025-05-23 math.OC cs.LG

Thomas Oliver de Jong , Khemraj Shukla , Mircea Lazar This is my paper · ORCID
classification math.OCcs.LG MSC 41A3068T0793C1093C35
keywords modelpredictivecontrolDeepONetoperatorlearningmulti-steppredictionuniversalapproximationnonlinearMIMOsystemsneuralbasisrepresentation
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 tries to establish that a modified DeepONet architecture, the multi-step DeepONet (MS-DeepONet), can serve as the prediction model inside model predictive control for continuous-time nonlinear multi-input, multi-output systems. The authors prove that the MS-DeepONet is a universal approximator for the operator that maps an initial state and a piecewise-constant multi-step input sequence to the corresponding multi-step output sequence, so no recurrent network or iterative one-step simulation is needed to forecast a horizon. They also show empirically that, on the van der Pol oscillator, the quadruple tank process, and a cart-pendulum swing-up task, the MS-DeepONet learns faster, achieves lower validation loss, and yields lower mean tracking error than the standard DeepONet when both are embedded in MPC. If these claims hold, neural predictive control gains a one-shot, feedforward predictor with a formal representational guarantee.

What carries the argument

The load-bearing object is the MS-DeepONet product-layer predictor $\bar{y}_k = B^\top(u_k) t(x_k)$, in which one branch network encodes the full multi-step input sequence $u_k = \mathrm{col}(u_k,\ldots,u_{k+N-1})$ and one trunk network encodes the measured state $x(kT_s)$; the branch output layer carries $p$ linear features per output per step, so a single forward pass produces all $N n_y$ future outputs. The proof machinery is a modified universal approximation theorem (Theorem 2) that replaces the infinite-dimensional function space of the classical operator theorem with a compact Euclidean input domain, so the operator $G$ maps compact input sequences directly to $C(K_2)$, and then shows the resulting one-layer network is a special case of the unstacked DeepONet. The supporting structural result is the basis representation $y_k = \Theta_o \Phi_\otimes(u_k, x_k)$, obtained by expanding the product of affine branch and trunk layers into a Kronecker-feature vector, which turns the trained network into a linear-in-features predictor and connects it to data-enabled predictive control.

What would settle it

Train the MS-DeepONet and a standard DeepONet on a linear MIMO system with exactly computable sampled outputs, starting from the same data and matching widths, and compare multi-step prediction error to machine precision: for the universal-approximation claim to survive, the MS-DeepONet error must shrink toward zero as width and training epochs grow; a persistent error floor above the standard linear predictor's residual would falsify it.

Watch

Extended reading notes

Core claim

The central claim is that the solution operator of a continuous-time MIMO system (1) restricted to piecewise-constant inputs and sampled outputs, written as $y_k = G(u_k)(x_k)$, can be approximated arbitrarily well by a single product-layer network: $\sum_{i=1}^p b_{i,j}^q(u_k) t_i(x_k)$, where the branch net sees the entire multi-step input sequence $u_k$ and the trunk net sees only the initial state $x(kT_s)$. Theorem 2, adapting the Chen-Chen operator approximation theorem to the finite-dimensional domain of input sequences, is the proof that such a network exists for any $\epsilon > 0$, and Remark 1 carries that result over to the MIMO operator (6). The paper further derives an equivalent basis form $y_k = \Theta_o \Phi_\otimes(u_k, x_k)$ using a Kronecker product of branch and trunk features, which exposes the network as an adaptive linear basis and motivates a data-enabled predictive control variant. On the numerical side, the paper reports consistent improvements over the standard DeepONet/MimoONet predictor in training loss, validation loss, tracking error, and training time across three benchmark systems, including closed-loop swing-up data for the unstable pendulum on a cart.

Load-bearing premise

The method assumes the full state $x(kT_s)$ is measured at every sampling instant; if states are unobserved, the MS-DeepONet predictor as implemented cannot be applied and the numerical results do not cover that setting.

Editorial extensions

If this is right

  • An MPC controller using MS-DeepONet solves Problem 3 with a single matrix-vector evaluation $B^\top(u_k)t(x_k)$ per candidate input, avoiding recurrent unrolling and multiple branch networks; measured computation times stay below the sampling period in the presented benchmarks.
  • The universal approximation result means representational completeness: for any continuous-time nonlinear MIMO system with piecewise-constant inputs over a compact operating region, a sufficiently large MS-DeepONet can in principle represent the exact sampled multi-step input-output map, so prediction error within MPC is a training issue rather than an architectural ceiling.
  • The basis form $y_k = \Theta_o \Phi_\otimes(u_k, x_k)$ yields a DeePC-style formulation by constructing a data matrix $\Phi_\otimes$ from learned features, with the predictor solving a regularized problem in the coefficient vector $g_k$; the paper proposes but does not implement this variant.
  • The architecture's lower training loss and up to several-fold shorter training time compared with standard DeepONet (55 vs 14 minutes on van der Pol; 164 vs 40 minutes on quadruple tank) suggest the one-shot multi-step parametrization is an easier learning problem.

Reading between the lines

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

  • If the full-state assumption (Remark 5) is relaxed to measured input-output histories, the same product-layer form should be trainable as an output-feedback predictor; the paper sketches this but gives no experiment, so a natural test is to re-run the three benchmarks with the history window replacing $x(kT_s)$.
  • The theorem's compact-domain premise suggests that the approximation guarantee degrades gracefully rather than catastrophically outside the training region; a testable extension would be to quantify worst-case multi-step error as the prediction horizon $N$ grows beyond the trained value and to compare with one-step-recursive DeepONet at the same total width.
  • The candidate basis might transfer between similar systems: features $\Phi_t$ learned on one plant could be frozen and only $\Theta_o$ re-fitted for a neighboring plant, which would turn MS-DeepONet into a reusable linear-in-features plant model for fast MPC re-tuning.
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 / 6 minor

Summary. The paper proposes the MS-DeepONet, an adaptation of the DeepONet architecture that maps an initial state x(kT_s) and a length-N piecewise-constant input sequence to all N future outputs of a continuous-time MIMO system in a single forward pass. One branch network encodes the stacked input sequence and one trunk network encodes only the state; the branch output dimension is p·n_y·N, so the product layer produces the complete multi-step output vector. The authors state a universal-approximation result for the architecture (Section III, Theorem 2 and Remark 1), derive an equivalent Kronecker-product basis representation (Lemma 1), propose a grid-search hyperparameter ablation (Algorithm 2), formulate MPC problems for both the standard and MS-DeepONet predictors (Problems 3 and 4), and validate the approach on the van der Pol oscillator, the quadruple tank process, and a cart-pendulum swing-up task. The implementation is released on GitHub.

Significance. If the universality claim is fully established, the architecture is practically attractive: it avoids repeated network evaluations and separate branch networks per input, and the basis representation points toward data-enabled predictive control. Theorem 2 is a credible adaptation of the Chen-and-Chen result, and the numerical protocol is fair: both architectures are trained on identical data with a grid search, and the reported losses, AME, and computation times support the comparison on the first two benchmarks. The publicly available PyTorch code and the explicit ablation study are strengths. The main weaknesses are that the step from Theorem 2 to the exact MS-DeepONet architecture is asserted in Remark 1 rather than proved, that the basis representation in Lemma 1 contains a Kronecker-ordering inconsistency, and that the claimed consistent outperformance on the pendulum benchmark is not supported by a standard-DeepONet baseline. These issues are repairable but currently prevent the paper from supporting its strongest claims.

major comments (4)
  1. [Section III-A, Remark 1 (Eq. (15))] Remark 1 is the only place where universality of the exact MS-DeepONet architecture is claimed, but it is asserted rather than proved. Theorem 2 supplies, for each scalar output component G_{q+(j-1)n_y}, an approximation whose trunk basis functions σ(w^k·x+ζ^k) are chosen for that component. In Eq. (14), however, all output components must share the same p trunk functions t_i(x). To close the gap one must construct a single p-dimensional trunk and p·n_y·N branch outputs, for example by enlarging p to the total number of basis terms and zero-padding the branch matrix B(u_k), and then verify that feedforward networks of the form (12)-(13) can realize the resulting representations. This construction is absent; as written, the proof does not establish the abstract's claim that the MS-DeepONet is a universal approximator for multi-step sequence prediction.
  2. [Section III-B, Lemma 1 (Eqs. (22)-(23))] The proof of Lemma 1 expands the first term of (21) as vec^T(Wb_i^T Wt)(Φ_t(x_k)⊗Φ_b(u_k)), but the basis vector Φ⊗ in Eq. (22) uses Φ_b(u_k)⊗Φ_t(x_k) in its first block. With the column-major vec defined in the Notation, vec^T(A)(v⊗u) equals u^T A v while vec^T(A)(u⊗v) equals v^T A^T u, so the two orderings are not interchangeable. Since Eq. (23) lists vec^T(Wb_i^T Wt) as the coefficient of Φ_b⊗Φ_t, the expression does not equal (21) as written; a Kronecker permutation must be introduced. This inconsistency affects the basis representation used in Remark 3 and the data-enabled formulation sketched in Remark 6.
  3. [Section VI-C and Table 3] The abstract and conclusions state that MS-DeepONet consistently outperforms the standard DeepONet across the three benchmarks, but the pendulum-on-cart experiment contains no standard DeepONet baseline. Table 3 reports only an MS-DeepONet column for the pendulum, and Section VI-C evaluates no comparison against Problem 4. The consistently-outperforms claim is therefore supported only for the van der Pol and quadruple-tank examples; either add the standard DeepONet comparison or qualify the claim to those two examples.
  4. [Section III, Eq. (6) and Theorem 2] Theorem 2 assumes G is a continuous operator from a compact subset of R^l into C(K2), but Section III does not state the assumptions on f and h under which the multi-step operator in Eq. (6) is continuous. Standard well-posedness of the system (1) would supply this, but the hypotheses (for example, Lipschitz continuity of f and continuity of h) and a short argument are missing. Without them, Theorem 2 cannot be invoked for the operator (6).
minor comments (6)
  1. [Figure 4 caption] The caption says the output is the predicted multi-step output sequence \bar{u}_0; this should be \bar{y}_0 or the equivalent output notation, since the architecture predicts outputs, not inputs.
  2. [Section IV-B] After defining t(x_k, Ξ_i, Θ_i^t), the text repeats the branch definitions: 'Note in this case that Φ_b(u_k)=h^b_{l_i^b} and h^b_0=u_k' should instead state the trunk analogue, i.e., Φ_t(x_k)=h^t_{l_i^t} and h^t_0=x_k.
  3. [Appendix, Proposition 1 proof] In the proof of Proposition 1, the trunk argument is written as σ(w^k·u+ζ^k), but it should be σ(w^k·z+ζ^k) to match the operator input z; the surrounding text also contains a few typos such as 'from(6)' and 'MS-DeepoNet'.
  4. [Section VI-B] The statement that the MS-DeepONet MPC controller achieved performance comparable to the model-based MPC controller, with results omitted due to space limitations, is not verifiable; either include the comparison in the paper or remove the claim.
  5. [Section VI and Table 3] The acronym AME is used in Table 3 but never defined; please define it (for example, absolute mean tracking error) and state over which trajectories and time horizons it is computed, especially for the pendulum swing-up task.
  6. [Section V-A, Remark 5] The full-state-feedback assumption is disclosed in Remark 5, but all benchmark controllers rely on it and the alternative input-output history variant is not implemented; a sentence in the introduction or abstract stating that the method currently requires full-state feedback would set expectations more accurately.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the universal-approximation claim is adapted from external theorems and the empirical comparisons are genuine closed-loop benchmarks.

full rationale

The central theoretical claim, that MS-DeepONet is a universal approximator for multi-step prediction, rests on Theorem 2, whose proof is an adaptation of Chen and Chen's external Theorem 3 [23], and on Proposition 1, which relates the stacked and unstacked DeepONet forms; no output quantity is substituted into an input by construction. The main mathematical weakness is an omitted proof detail, not circularity: Remark 1 (Section III-A) asserts inequality (15) 'from Proposition 1 and Theorem 2', but the text does not spell out how the scalar, output-specific trunk functions of Theorem 2 are merged into the single shared trunk of equations (13)-(14); the standard enlargement-of-p construction would close the gap, and the missing argument does not fold the target result into the assumptions. The basis representation of Lemma 1 (Section III-B) is algebraic manipulation of the architecture's own product-layer equation, not a prediction derived from fitted data. The numerical comparisons train both architectures on the same data and evaluate closed-loop MPC tracking, so the claimed outperformance is an external benchmark rather than a fitted quantity; the validation loss in Table 3 is selected by the ablation study, which weakens the learning comparison statistically but is not circular. Self-citations ([8], [34], [39]) are used for background or future extensions and are not load-bearing. The full-state feedback assumption in Remark 5 (Section V-A) and the unoptimized pendulum training noted in Section VI-C are genuine limitations, flagged here explicitly, but they concern applicability and completeness, not circularity.

Assumptions & free parameters 6 free parameters · 5 assumptions · 0 invented entities

The central theoretical claim (universal approximation) rests on standard approximation results and continuity assumptions; the empirical claims rest on a set of hand-chosen hyperparameters and MPC weights that are not derived from the problem data.

free parameters (6)
  • Output feature count p = 20 (van der Pol, quadruple tank), 40 (pendulum)
    Chosen via grid search (ablation study, Algorithm 2) over p in {20,30,40} for van der Pol; selected as 20/40 for other systems; affects approximation capacity.
  • Hidden layer depth and width (l_b, l_t, n_b,i, n_t,i) = vdP: 3x{40,40,40}; 4-tank: 2x{20,20}; pendulum: 3x{128,256,128}
    Selected via ablation (vdP) or by hand (4-tank, pendulum); these hyperparameters define the network capacity.
  • Prediction horizon N = 10 (vdP), 20 (quadruple tank), 40 (pendulum)
    Chosen by the authors; directly sets the number of output steps and the dimension of the branch input/output, so it is a hand-picked parameter the results depend on.
  • Sampling time Ts = 0.1 s (vdP, pendulum), 5 s (quadruple tank)
    Chosen by the authors; determines the discretization of the continuous-time system and the data used for training.
  • MPC cost weights Q, R (and P) = vdP: Q=100, R=1; 4-tank: Q=100I, R=I; pendulum: Q=diag(1,1000,1,1,1), R=0.01
    Tuning knobs for the controller; different choices would change the closed-loop performance and the comparison.
  • Weight decay lambda (regularization) = not specified
    Mentioned in Eq. (30) as a regularizer but its value is never given, making it an unspecified free parameter of the training procedure.
assumptions (5)
  • standard math Theorem 3 of Chen and Chen (1995) [23] holds as stated
    Used as the foundational approximation result in the proof of Theorem 2 (Section III).
  • domain assumption The solution operator G in Eq. (6) is continuous on compact input/state sets
    Required for Theorem 2's compactness argument; follows from well-posedness of the ODE but is not proven in the paper.
  • domain assumption Plant dynamics f and output map h are continuous
    Assumed in the system description (1) and needed for the operator to be continuous.
  • domain assumption Full state x(kTs) is measured at each sampling instant
    Stated in Remark 5, Section V-A; the trunk network takes the state as input.
  • domain assumption Inputs are piecewise constant (ZOH) over sampling intervals
    Stated in Eq. (2) and embedded in the operator definition (6).

how reviews work

0 comments
Cite this review

Pith. "Pith review of Deep Operator Neural Network Model Predictive Control." pith.science (2026). https://pith.science/paper/M4DPJDJX

@misc{pith2026250518008,
  author       = {Pith},
  title        = {Pith review of: Deep Operator Neural Network Model Predictive Control},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/M4DPJDJX}},
  note         = {Machine review of arXiv:2505.18008}
}
read the original abstract

In this paper, we consider the design of model predictive control (MPC) algorithms based on deep operator neural networks (DeepONets). These neural networks are capable of accurately approximating real and complex valued solutions of continuous time nonlinear systems without relying on recurrent architectures. The DeepONet architecture is made up of two feedforward neural networks: the branch network, which encodes the input function space, and the trunk network, which represents dependencies on temporal variables or initial conditions. Utilizing the original DeepONet architecture as a predictor within MPC for Multi Input Multi Output (MIMO) systems requires multiple branch networks, to generate multi output predictions, one for each input. Moreover, to predict multiple time steps into the future, the network has to be evaluated multiple times. Motivated by this, we introduce a multi step DeepONet (MS-DeepONet) architecture that computes in one shot multi step predictions of system outputs from multi step input sequences, which is better suited for MPC. We prove that the MS DeepONet is a universal approximator in terms of multi step sequence prediction. Additionally, we develop automated hyper parameter selection strategies and implement MPC frameworks using both the standard DeepONet and the proposed MS DeepONet architectures in PyTorch. The implementation is publicly available on GitHub. Simulation results demonstrate that MS-DeepONet consistently outperforms the standard DeepONet in learning and predictive control tasks across several nonlinear benchmark systems: the van der Pol oscillator, the quadruple tank process, and a cart pendulum unstable system, where it successfully learns and executes multiple swing up and stabilization policies.

Figures

Figures reproduced from arXiv: 2505.18008 by the authors.

Figure 1
Figure 1. FIGURE 1 [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. FIGURE 2 [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. FIGURE 3 [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figures from the paper (7 more)
Figure 4
Figure 4. Figure 4: FIGURE 4 [PITH_FULL_IMAGE:figures/full_fig_p006_4.png]
Figure 5
Figure 5. Figure 5: FIGURE 5 [PITH_FULL_IMAGE:figures/full_fig_p008_5.png]
Figure 6
Figure 6. Figure 6: FIGURE 6 [PITH_FULL_IMAGE:figures/full_fig_p010_6.png]
Figure 7
Figure 7. Figure 7: FIGURE 7 [PITH_FULL_IMAGE:figures/full_fig_p011_7.png]
Figure 8
Figure 8. Figure 8: FIGURE 8 [PITH_FULL_IMAGE:figures/full_fig_p012_8.png]
Figure 10
Figure 10. Figure 10: FIGURE 10 [PITH_FULL_IMAGE:figures/full_fig_p013_10.png]
Figure 11
Figure 11. Figure 11: FIGURE 11 [PITH_FULL_IMAGE:figures/full_fig_p014_11.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

44 extracted references · 34 canonical work pages

  1. [1]

    Learning nonlinear operators via DeepONet based on the universal approxima- tion theorem of operators,

    L. Lu, P. Jin, G. Pang, Z. Zhang, and G. E. Karniadakis, “Learning nonlinear operators via DeepONet based on the universal approxima- tion theorem of operators,”Nature Machine Intelligence, vol. 3, pp. 218–229, 2021

    work page 2021

  2. [2]

    Complex valued deep operator network (deeponet)[G]for three dimensional maxwell’s equations:G ∈C m×n,

    Q. Jiang, M. Salvadori, D. Ota, V . Shankar, and K. Shukla, “Complex valued deep operator network (deeponet)[G]for three dimensional maxwell’s equations:G ∈C m×n,” 2024. [Online]. Available: https://arxiv.org/abs/2411.18733

    work page arXiv 2024

  3. [3]

    A neural predictive controller for non- linear systems,

    M. Lazar and O. Pastravanu, “A neural predictive controller for non- linear systems,”Mathematics and Computers in Simulation, vol. 60, no. 3-5, pp. 315–324, 2002

    work page 2002

  4. [4]

    Neural networks in model predictive control,

    M. Ławry ´nczuk, “Neural networks in model predictive control,”Intel- ligent systems for knowledge management, pp. 31–63, 2009

    work page 2009

  5. [5]

    Nørregard, O

    M. Nørregard, O. Ravn, N. K. Poulsen, and L. K. Hansen,Neural Networks for modelling and control of dynamic systems. Springer, London, 2000

    work page 2000

  6. [6]

    Learning affine predictors for mpc of nonlinear systems via artificial neural networks,

    D. Masti, F. Smarra, A. D’Innocenzo, and A. Bemporad, “Learning affine predictors for mpc of nonlinear systems via artificial neural networks,”IFAC-PapersOnLine, vol. 53, no. 2, pp. 5233–5238, 2020

    work page 2020

  7. [7]

    Deep prediction networks,

    A. Dalla Libera and G. Pillonetto, “Deep prediction networks,”Neu- rocomputing, vol. 469, pp. 321–329, 2022

    work page 2022

  8. [8]

    Nonlinear data-driven predictive control using deep subspace prediction networks,

    M. Lazar, M.-S. Popescu, and M. Schoukens, “Nonlinear data-driven predictive control using deep subspace prediction networks,” in2023 62nd IEEE Conference on Decision and Control (CDC), 2023, pp. 3770–3775

    work page 2023

Show all 44 references
  1. [9]

    Long short-term memory,

    S. Hochreiter and J. Schmidhuber, “Long short-term memory,”Neural Computation, vol. 9, no. 8, pp. 1735–1780, 1997

    1997

  2. [10]

    Learning phrase representations using rnn encoder-decoder for statistical machine translation,

    K. Cho, B. Van Merri ¨enboer, C. Gulcehre, D. Bahdanau, F. Bougares, H. Schwenk, and Y . Bengio, “Learning phrase representations using rnn encoder-decoder for statistical machine translation,” in Proceedings of the 2014 Conference on Empirical Methods in Natural Language Proc...

    2014 arXiv

  3. [11]

    Nonlinear mpc for offset-free tracking of systems learned by gru neural networks,

    F. Bonassi, C. F. O. da Silva, and R. Scattolini, “Nonlinear mpc for offset-free tracking of systems learned by gru neural networks,” IFAC-PapersOnLine, vol. 54, no. 14, pp. 54–59, 2021, 3rd IFAC Conference on Modelling, Identification and Control of Nonlinear Systems MICNON 2021

    2021

  4. [12]

    Advanced predictive control for gru and lstm networks,

    K. Zarzycki and M. Ławry ´nczuk, “Advanced predictive control for gru and lstm networks,”Information Sciences, vol. 616, pp. 229–254, 2022. 14 VOLUME

    2022

  5. [13]

    Robust constrained nonlinear model predictive control with gated recurrent unit model,

    I. Schimperna and L. Magni, “Robust constrained nonlinear model predictive control with gated recurrent unit model,”Automatica, vol. 161, p. 111472, 2024

    2024

  6. [14]

    Robust offset-free constrained model predictive control with long short-term memory networks,

    ——, “Robust offset-free constrained model predictive control with long short-term memory networks,”IEEE Transactions on Automatic Control, vol. 69, no. 12, pp. 8172–8187, 2024

    2024

  7. [15]

    Estimation and mpc control based on gated recurrent unit neural networks with unknown disturbances,

    E. Masero, F. Bonassi, A. L. Bella, and R. Scattolini, “Estimation and mpc control based on gated recurrent unit neural networks with unknown disturbances,” in2024 IEEE 63rd Conference on Decision and Control (CDC), 2024, pp. 120–125

    2024

  8. [16]

    Recurrent neural network-based mpc for systems with input and incremental input constraints,

    I. Schimperna, G. Galuppini, and L. Magni, “Recurrent neural network-based mpc for systems with input and incremental input constraints,”IEEE Control Systems Letters, vol. 8, pp. 814–819, 2024

    2024

  9. [17]

    On recurrent neural networks for learning-based control: Recent results and ideas for future developments,

    F. Bonassi, M. Farina, J. Xie, and R. Scattolini, “On recurrent neural networks for learning-based control: Recent results and ideas for future developments,”Journal of Process Control, vol. 114, pp. 92–104, 2022

    2022

  10. [18]

    A tutorial review of neural network modeling approaches for model predictive control,

    Y . M. Ren, M. S. Alhajeri, J. Luo, S. Chen, F. Abdullah, Z. Wu, and P. D. Christofides, “A tutorial review of neural network modeling approaches for model predictive control,”Computers & Chemical Engineering, vol. 165, p. 107956, 2022

    2022

  11. [19]

    Physics-informed neural networks-based model predictive control for multi-link manip- ulators,

    J. Nicodemus, J. Kneifl, J. Fehr, and B. Unger, “Physics-informed neural networks-based model predictive control for multi-link manip- ulators,”IFAC-PapersOnLine, vol. 55, no. 20, pp. 331–336, 2022, 10th Vienna International Conference on Mathematical Modelling MATHMOD 2022

    2022

  12. [20]

    Ramp-net: A robust adaptive mpc for quadro- tors via physics-informed neural network,

    S. Sanyal and K. Roy, “Ramp-net: A robust adaptive mpc for quadro- tors via physics-informed neural network,” in2023 IEEE International Conference on Robotics and Automation (ICRA), 2023, pp. 1019–1025

    2023

  13. [21]

    Real-time neural mpc: Deep learning model predictive control for quadrotors and agile robotic platforms,

    T. Salzmann, E. Kaufmann, J. Arrizabalaga, M. Pavone, D. Scara- muzza, and M. Ryll, “Real-time neural mpc: Deep learning model predictive control for quadrotors and agile robotic platforms,”IEEE Robotics and Automation Letters, vol. 8, no. 4, pp. 2397–2404, 2023

    2023

  14. [22]

    Fourier neural operator for parametric partial differential equations,

    Z. Li, N. Kovachki, K. Azizzadenesheli, B. Liu, K. Bhattacharya, A. Stuart, and A. Anandkumar, “Fourier neural operator for parametric partial differential equations,”arXiv preprint arXiv:2010.08895, 2020

    2010 arXiv

  15. [23]

    Universal approximation to nonlinear operators by neural networks with arbitrary activation functions and its applica- tion to dynamical systems,

    T. Chen and H. Chen, “Universal approximation to nonlinear operators by neural networks with arbitrary activation functions and its applica- tion to dynamical systems,”IEEE Transactions on Neural Networks, vol. 6, no. 4, pp. 911–917, 1995

    1995

  16. [24]

    Deep neural operators as accurate surrogates for shape optimization,

    K. Shukla, V . Oommen, A. Peyvan, M. Penwarden, N. Plewacki, L. Bravo, A. Ghoshal, R. M. Kirby, and G. E. Karniadakis, “Deep neural operators as accurate surrogates for shape optimization,”En- gineering Applications of Artificial Intelligence, vol. 129, p. 107615, 2024

    2024

  17. [25]

    Learning two-phase microstructure evolution using neural operators and autoencoder architectures,

    V . Oommen, K. Shukla, S. Goswami, R. Dingreville, and G. E. Kar- niadakis, “Learning two-phase microstructure evolution using neural operators and autoencoder architectures,”npj Computational Materi- als, vol. 8, no. 1, p. 190, 2022

    2022

  18. [26]

    Deep operator learning-based sur- rogate models for aerothermodynamic analysis of aedc hypersonic waverider,

    K. Shukla, J. Ratchford, L. Bravo, V . Oommen, N. Plewacki, A. Ghoshal, and G. Karniadakis, “Deep operator learning-based sur- rogate models for aerothermodynamic analysis of aedc hypersonic waverider,”arXiv preprint arXiv:2405.13234, 2024

    2024 arXiv

  19. [27]

    Deepm&mnet: Inferring the electroconvection multiphysics fields based on operator approximation by neural networks,

    S. Cai, Z. Wang, L. Lu, T. A. Zaki, and G. E. Karniadakis, “Deepm&mnet: Inferring the electroconvection multiphysics fields based on operator approximation by neural networks,”Journal of Computational Physics, vol. 436, p. 110296, 2021

    2021

  20. [28]

    Machine learning: Bane or boon for control?: 2023 bode prize lecture,

    M. Krstic, “Machine learning: Bane or boon for control?: 2023 bode prize lecture,”IEEE Control Systems, vol. 44, no. 4, pp. 24–37, 2024

    2023

  21. [29]

    Neural operators of backstepping controller and observer gain functions for reaction–diffusion pdes,

    M. Krstic, L. Bhan, and Y . Shi, “Neural operators of backstepping controller and observer gain functions for reaction–diffusion pdes,” Automatica, vol. 164, p. 111649, 2024

    2024

  22. [30]

    Neural operators for predictor feedback control of nonlinear delay systems,

    L. Bhan, P. Qin, M. Krstic, and Y . Shi, “Neural operators for predictor feedback control of nonlinear delay systems,”arXiv, no. 2411.18964, 2025

    2025 arXiv

  23. [31]

    Mionet: Learning multiple-input operators via tensor product,

    P. Jin, S. Meng, and L. Lu, “Mionet: Learning multiple-input operators via tensor product,”SIAM Journal on Scientific Computing, vol. 44, no. 6, pp. A3490–A3514, 2022

    2022

  24. [32]

    Deep multi-input and multi-output operator networks method for optimal control of pdes,

    J. Yong, X. Luo, and S. Sun, “Deep multi-input and multi-output operator networks method for optimal control of pdes,”Electronic Research Archive, vol. 32, no. 7, pp. 4291–4320, 2024

    2024

  25. [33]

    The no-prop algorithm: A new learning algorithm for multilayer neural networks,

    B. Widrow, A. Greenblatt, Y . Kim, and D. Park, “The no-prop algorithm: A new learning algorithm for multilayer neural networks,” Neural Networks, vol. 37, pp. 182–188, 2013, twenty-fifth Anniversay Commemorative Issue

    2013

  26. [34]

    Neural data–enabled predictive control,

    M. Lazar, “Neural data–enabled predictive control,”IFAC- PapersOnLine, vol. 58, no. 15, pp. 91–96, 2024, 20th IFAC Symposium on System Identification SYSID 2024

    2024

  27. [35]

    Pytorch: An imperative style, high-performance deep learning library,

    A. Paszke, S. Gross, F. Massa, A. Lerer, J. Bradbury, G. Chanan, T. Killeen, Z. Lin, N. Gimelshein, L. Antiga, A. Desmaison, A. K ¨opf, E. Z. Yang, Z. DeVito, M. Raison, A. Tejani, S. Chilamkurthy, B. Steiner, L. Fang, J. Bai, and S. Chintala, “Pytorch: An imperative style, hi...

    1912 arXiv

  28. [36]

    Approximation capabilities of multilayer feedforward networks,

    K. Hornik, “Approximation capabilities of multilayer feedforward networks,”Neural networks, vol. 4, no. 2, pp. 251–257, 1991

    1991

  29. [37]

    Approximation with artificial neural networks,

    B. C. Cs ´ajiet al., “Approximation with artificial neural networks,” Faculty of Sciences, Etvs Lornd University, Hungary, vol. 24, no. 48, p. 7, 2001

    2001

  30. [38]

    W. A. Sutherland,Introduction to metric and topological spaces. Oxford University Press, 2009

    2009

  31. [39]

    Basis-functions nonlinear data-enabled predictive control: Consistent and computationally efficient formulations,

    M. Lazar, “Basis-functions nonlinear data-enabled predictive control: Consistent and computationally efficient formulations,” in2024 Euro- pean Control Conference (ECC), 2024, pp. 888–893

    2024

  32. [40]

    Robust training and initialization of deep neural networks: An adap- tive basis viewpoint,

    E. C. Cyr, M. A. Gulian, R. G. Patel, M. Perego, and N. A. Trask, “Robust training and initialization of deep neural networks: An adap- tive basis viewpoint,” inProceedings of The First Mathematical and Scientific Machine Learning Conference, ser. Proceedings of Machine Learni...

    2020

  33. [41]

    Data-Enabled Predictive Control: In the Shallows of the DeePC,

    J. Coulson, J. Lygeros, and F. D ¨orfler, “Data-Enabled Predictive Control: In the Shallows of the DeePC,” in18th European Control Conference, Napoli, Italy, 2019, pp. 307–312

    2019

  34. [42]

    Four MPC implementations compared on the Quadruple Tank Process Benchmark: pros and cons of neural MPC,

    P. Cl ´ement Blaud, P. Chevrel, F. Claveau, P. Haurant, and A. Mouraud, “Four MPC implementations compared on the Quadruple Tank Process Benchmark: pros and cons of neural MPC,” IFAC-PapersOnLine, vol. 55, no. 16, pp. 344–349, 2022. [Online]. Available: https://hal.science/hal...

    2022

  35. [43]

    Swing-up and stabiliza- tion of a cart–pendulum system under restricted cart track length,

    D. Chatterjee, A. Patra, and H. K. Joglekar, “Swing-up and stabiliza- tion of a cart–pendulum system under restricted cart track length,” Systems & control letters, vol. 47, no. 4, pp. 355–364, 2002

    2002

  36. [44]

    Choose wisely: Data- enabled predictive control for nonlinear systems using online data selection,

    J. N ¨af, K. Moffat, J. Eising, and F. D ¨orfler, “Choose wisely: Data- enabled predictive control for nonlinear systems using online data selection,” 2025. [Online]. Available: https://arxiv.org/abs/2503.18845 Thomas O. de Jong(Student Member, IEEE) received the B.S. degree i...

    2025 arXiv

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.