Verifiable Error Bounds for Physics-Informed Neural Network Solutions of Lyapunov and Hamilton-Jacobi-Bellman Equations
Pith reviewed 2026-05-21 10:13 UTC · model grok-4.3
The pith
Verifiable residual bounds for neural approximations to Lyapunov and HJB equations imply relative error bounds on the exact solutions.
A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.
Core claim
For both the Lyapunov and HJB PDEs, a verifiable residual bound yields relative error bounds with respect to the true solutions as well as computable a posteriori estimates in terms of the approximate solutions. For the HJB equation this also yields certified upper and lower bounds on the optimal value function on compact sublevel sets and quantifies the optimality gap of the induced feedback policy. We further show that one-sided residual bounds already imply that the approximation itself defines a valid Lyapunov or control Lyapunov function.
What carries the argument
Comparison principles and Gronwall-type arguments that convert a bound on the PDE residual into a bound on the solution error.
Load-bearing premise
The PDE residual of the approximate solution must admit a rigorous and verifiable upper bound, and the equations must satisfy the regularity conditions required for the comparison principles and Gronwall arguments to relate the residual to the solution error.
What would settle it
Finding a concrete approximate solution to a Lyapunov or HJB equation with a small certified residual for which the actual error to the true solution exceeds the derived relative error bound would falsify the central claim.
Figures
read the original abstract
Many core problems in nonlinear systems analysis and control can be recast as solving partial differential equations (PDEs) such as Lyapunov and Hamilton-Jacobi-Bellman (HJB) equations. Physics-informed neural networks (PINNs) have emerged as a promising mesh-free approach for approximating their solutions, but in most existing works there is no rigorous guarantee that a small PDE residual implies a small solution error. This paper develops verifiable error bounds for approximate solutions of Lyapunov and HJB equations, with particular emphasis on PINN-based approximations. For both the Lyapunov and HJB PDEs, we show that a verifiable residual bound yields relative error bounds with respect to the true solutions as well as computable a posteriori estimates in terms of the approximate solutions. For the HJB equation, this also yields certified upper and lower bounds on the optimal value function on compact sublevel sets and quantifies the optimality gap of the induced feedback policy. We further show that one-sided residual bounds already imply that the approximation itself defines a valid Lyapunov or control Lyapunov function. We illustrate the results with numerical examples.
Editorial analysis
A structured set of objections, weighed in public.
Referee Report
Summary. The manuscript develops verifiable error bounds for physics-informed neural network (PINN) solutions to Lyapunov and Hamilton-Jacobi-Bellman (HJB) equations. It demonstrates that a verifiable bound on the PDE residual implies relative error bounds with respect to the true solutions, as well as computable a posteriori estimates. For the HJB equation, this provides certified upper and lower bounds on the optimal value function on compact sublevel sets and quantifies the optimality gap of the induced feedback policy. One-sided residual bounds are shown to certify the approximation as a valid Lyapunov or control-Lyapunov function. The results are illustrated with numerical examples.
Significance. If the derivations hold, this paper makes a significant contribution to the field of nonlinear control and systems analysis by providing rigorous, verifiable guarantees for neural network approximations of key PDEs. This addresses a critical limitation in existing PINN literature for control applications, where small residuals do not necessarily imply small solution errors. The approach leverages standard PDE theory (comparison principles, Gronwall estimates) to yield practical a posteriori bounds, which could enable certified control design and analysis in safety-critical systems. The explicit handling of one-sided bounds for Lyapunov certification is particularly useful.
minor comments (3)
- [§4.1] §4.1: The statement of the comparison principle used to convert residual bounds to solution error bounds should include a precise citation to the version of the theorem invoked (e.g., the specific viscosity-solution comparison result from Crandall et al.).
- [Table 1] Table 1: The reported a posteriori error estimates would benefit from an additional column showing the ratio of the computed bound to the observed numerical error to illustrate tightness.
- [Figure 3] Figure 3 caption: The domain of the compact sublevel set on which the HJB bounds are certified is not stated explicitly; add the numerical value of the level set parameter used.
Simulated Author's Rebuttal
We thank the referee for their positive summary and significance assessment of our work on verifiable error bounds for PINN solutions of Lyapunov and HJB equations. We appreciate the recommendation for minor revision and will incorporate improvements to enhance clarity and presentation in the revised version.
Circularity Check
No significant circularity; derivation uses standard PDE comparison principles
full rationale
The paper's central derivation converts a verifiable residual bound on the PDE into solution error bounds via comparison principles and Gronwall-type estimates, with regularity conditions (Lipschitz continuity, viscosity solutions) stated explicitly before each theorem and constants derived directly from those assumptions. No steps reduce by construction to fitted inputs, self-definitions, or load-bearing self-citations; the abstract and skeptic analysis confirm the logical chain from residual to a-posteriori error and optimality gap is independent and self-contained against external PDE theory benchmarks.
Axiom & Free-Parameter Ledger
axioms (1)
- domain assumption The Lyapunov and HJB PDEs satisfy sufficient regularity and comparison-principle conditions so that residual bounds imply solution error bounds.
Lean theorems connected to this paper
-
IndisputableMonolith/Cost/FunctionalEquation.leanwashburn_uniqueness_aczel unclear?
unclearRelation between the paper passage and the cited Recognition theorem.
a verifiable residual bound yields relative error bounds... |r(x)|≤ε ω(x) ... |V̂(x)−V(x)|≤ε V(x)
What do these tags mean?
- matches
- The paper's claim is directly supported by a theorem in the formal canon.
- supports
- The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
- extends
- The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
- uses
- The paper appears to rely on the theorem as machinery.
- contradicts
- The paper's claim conflicts with a theorem or certificate in the canon.
- unclear
- Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.
Reference graph
Works this paper leans on
-
[1]
Martino Bardi, Italo Capuzzo Dolcetta, et al.Optimal control and viscosity solutions of Hamilton-Jacobi-Bellman equations, volume 12. Springer, 1997
work page 1997
-
[2]
Galerkin approximations of the generalized hamilton-jacobi-bellman equation
Randal W Beard, George N Saridis, and John T Wen. Galerkin approximations of the generalized hamilton-jacobi-bellman equation. Automatica, 33(12):2159–2177, 1997
work page 1997
-
[3]
Fabio Camilli, Lars Gr ¨une, and Fabian Wirth. A generalization of Zubov’s method to perturbed systems.SIAM Journal on Control and Optimization, 40(2):496–515, 2001
work page 2001
-
[4]
dReal: An SMT solver for nonlinear theories over the reals
Sicun Gao, Soonho Kong, and Edmund M Clarke. dReal: An SMT solver for nonlinear theories over the reals. InProc. of CADE, 2013
work page 2013
-
[5]
Peter Giesl.Construction of global Lyapunov functions using radial basis functions. Springer, 2007
work page 2007
-
[6]
Extreme learning machine: theory and applications.Neurocomputing, 70(1- 3):489–501, 2006
Guang-Bin Huang, Qin-Yu Zhu, and Chee-Kheong Siew. Extreme learning machine: theory and applications.Neurocomputing, 70(1- 3):489–501, 2006
work page 2006
-
[7]
Artificial neural networks for solving ordinary and partial differential equations
Isaac E Lagaris, Aristidis Likas, and Dimitrios I Fotiadis. Artificial neural networks for solving ordinary and partial differential equations. IEEE Transactions on Neural Networks, 9(5):987–1000, 1998
work page 1998
-
[8]
On strict verification of neural Lyapunov functions.IFAC-PapersOnLine, 59(19):304–309, 2025
Jun Liu and Maxwell Fitzsimmons. On strict verification of neural Lyapunov functions.IFAC-PapersOnLine, 59(19):304–309, 2025
work page 2025
-
[9]
Formally verified physics-informed neural control lyapunov functions
Jun Liu, Maxwell Fitzsimmons, Ruikun Zhou, and Yiming Meng. Formally verified physics-informed neural control lyapunov functions. InProc. of ACC, pages 1347–1354. IEEE, 2025
work page 2025
-
[10]
Towards learning and verifying maximal neural Lyapunov functions
Jun Liu, Yiming Meng, Maxwell Fitzsimmons, and Ruikun Zhou. Towards learning and verifying maximal neural Lyapunov functions. InProc. of CDC, 2023
work page 2023
-
[11]
Jun Liu, Yiming Meng, Maxwell Fitzsimmons, and Ruikun Zhou. LyZNet: A lightweight python tool for learning and verifying neural Lyapunov functions and regions of attraction. InProc. of HSCC, pages 1–8, 2024
work page 2024
-
[12]
Jun Liu, Yiming Meng, Maxwell Fitzsimmons, and Ruikun Zhou. Physics-informed neural network Lyapunov functions: PDE charac- terization, learning, and verification.Automatica, 175:112193, 2025
work page 2025
-
[13]
Physics-informed neural network policy iteration: Algorithms, convergence, and verification
Yiming Meng, Ruikun Zhou, Amartya Mukherjee, Maxwell Fitzsim- mons, Christopher Song, and Jun Liu. Physics-informed neural network policy iteration: Algorithms, convergence, and verification. InProc. of ICML, 2024
work page 2024
-
[14]
Ian M. Mitchell, Alexandre M. Bayen, and Claire J. Tomlin. A time-dependent Hamilton–Jacobi formulation of reachable sets for continuous dynamic games.IEEE Transactions on Automatic Control, 50(7):947–957, 2005
work page 2005
-
[15]
Maziar Raissi, Paris Perdikaris, and George E Karniadakis. Physics- informed neural networks: A deep learning framework for solving forward and inverse problems involving nonlinear partial differential equations.Journal of Computational physics, 378:686–707, 2019
work page 2019
-
[16]
Shiqi Wang, Huan Zhang, Kaidi Xu, Xue Lin, Suman Jana, Cho-Jui Hsieh, and J Zico Kolter. Beta-CROWN: Efficient bound propagation with per-neuron split constraints for complete and incomplete neural network verification.Proc. of NeurIPS, 34, 2021
work page 2021
-
[17]
Kaidi Xu, Huan Zhang, Shiqi Wang, Yihan Wang, Suman Jana, Xue Lin, and Cho-Jui Hsieh. Fast and complete: Enabling complete neural network verification with rapid and massively parallel incomplete verifiers. InProc. of ICLR, 2021
work page 2021
-
[18]
Physics-informed extreme learning machine Lyapunov functions
Ruikun Zhou, Maxwell Fitzsimmons, Yiming Meng, and Jun Liu. Physics-informed extreme learning machine Lyapunov functions. IEEE Control Systems Letters, 8:1763–1768, 2024
work page 2024
discussion (0)
Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.