Rigorously Certified Neural-Network Computation of Nonlinear Water Waves at Near-Machine Precision
Organizations: Courant Institute of Mathematical Sciences, New York University, NY, 10012, USA.
Abstract
We present a computational framework that combines machine-learning-based discovery with rigorous computer-assisted numerical certification and formal verification for strongly nonlinear partial differential equations. We demonstrate the approach for periodic gravity-capillary traveling water waves. Physics-informed neural networks are used as a search mechanism, with the wave profile, wave speed and, in continuation experiments, surface tension learned directly from the governing equations without training data. Through a staged training and refinement procedure, the neural networks achieve near-machine-precision residuals of the governing equations, even in strongly nonlinear regimes. A single formulation recovers high-amplitude Stokes waves, Crapper waves and Wilton ripples. Continuation from a high-amplitude Wilton ripple in surface tension reveals new high-amplitude candidate solutions. We freeze one candidate as an explicit finite representation and rigorously certify a uniform bound on its governing-equation residual using interval arithmetic, accounting for floating-point roundoff, Fourier truncation, aliasing and grid-to-continuum errors. This independently validates the neural-network accuracy. We formalize the certification argument in Lean, which machine-checks the deductions from certified numerical bounds to the residual theorem. Both the interval certification and formalization were developed with substantial assistance from a large language model, while the validity of the result remains independent of the language model: the numerical bounds are rigorously enclosed by interval arithmetic and the subsequent mathematical deductions are checked by the Lean kernel. Together, these steps provide a pipeline from neural discovery to rigorous numerical certification and machine-checked verification with auditable mathematical guarantees.
Figures & tables
| File(s) | Contents | Role |
|---|---|---|
| Definitions and reusable mathematical library | ||
| Definitions.lean | Defines the Fourier-multiplier Hilbert transform, finite profile, derivative, denominator , capillary expression , and residual . | Definitions |
| HilbertOnTrigPoly.lean | Proves the Hilbert-transform identities for finite trigonometric profiles, identifying with the explicit cosine sum Hyu . | Analysis |
| FourierTailHilbert.lean | Fourier-mode and tail estimates, including the Hilbert-transform bound. | Analysis |
| HilbertSmoothness.lean ; HilbertSharper.lean ; HilbertCommutesDeriv.lean | Regularity and derivative commutation for the Hilbert transform under the stated smoothness and periodicity hypotheses. | Analysis |
| AliasingErrorBound.lean | Sampled DFT aliasing estimate and equivalence with the cotangent-kernel convolution. | Analysis |
| Stage | Threads | Wall time | Peak RSS (GB) |
|---|---|---|---|
| Worker 1 | 4 | 1 h 57 min 25 s | 5.019 |
| Worker 2 | 4 | 2 h 14 min 58 s | 9.242 |
| Worker 3 | 4 | 1 h 56 min 25 s | 4.972 |
| Worker 4 | 4 | 2 h 07 min 07 s | 9.287 |
| Worker 5 | 4 | 2 h 13 min 04 s | 9.324 |
| Worker 6 | 4 | 1 h 58 min 34 s | 5.088 |
| Log entry | Reported value |
|---|---|
| Bound midpoint | |
| Bound radius | |
| Bound interval: lower endpoint | |
| Bound interval: upper endpoint | |
| Residual upper bound | |
| Denominator lower bound |
Neural PDE Solvers
Neural PDE solvers use trainable models to approximate solutions or solution operators for partial differential equations, often with physical constraints.
See all 299 papers