Saved in:
Bibliographic Details
Main Authors: Baba, Zakaria, Bayen, Alexandre M., Canesse, Alexi, Monache, Maria Laura Delle, Drieux, Martin, Fu, Zhe, Lichtlé, Nathan, Liu, Zihe, Matin, Hossein Nick Zinat, Piccoli, Benedetto
Format: Preprint
Published: 2026
Subjects:
Online Access:https://arxiv.org/abs/2601.06388
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866915719963213824
author Baba, Zakaria
Bayen, Alexandre M.
Canesse, Alexi
Monache, Maria Laura Delle
Drieux, Martin
Fu, Zhe
Lichtlé, Nathan
Liu, Zihe
Matin, Hossein Nick Zinat
Piccoli, Benedetto
author_facet Baba, Zakaria
Bayen, Alexandre M.
Canesse, Alexi
Monache, Maria Laura Delle
Drieux, Martin
Fu, Zhe
Lichtlé, Nathan
Liu, Zihe
Matin, Hossein Nick Zinat
Piccoli, Benedetto
contents We present a neural network-based method for learning scalar hyperbolic conservation laws. Our method replaces the traditional numerical flux in finite volume schemes with a trainable neural network while preserving the conservative structure of the scheme. The model can be trained both in a supervised setting with efficiently generated synthetic data or in an unsupervised manner, leveraging the weak formulation of the partial differential equation. We provide theoretical results that our model can perform arbitrarily well, and provide associated upper bounds on neural network size. Extensive experiments demonstrate that our method often outperforms efficient schemes such as Godunov's scheme, WENO, and Discontinuous Galerkin for comparable computational budgets. Finally, we demonstrate the effectiveness of our method on a traffic prediction task, leveraging field experimental highway data from the Berkeley DeepDrive drone dataset.
format Preprint
id arxiv_https___arxiv_org_abs_2601_06388
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Supervised and Unsupervised Neural Network Solver for First Order Hyperbolic Nonlinear PDEs
Baba, Zakaria
Bayen, Alexandre M.
Canesse, Alexi
Monache, Maria Laura Delle
Drieux, Martin
Fu, Zhe
Lichtlé, Nathan
Liu, Zihe
Matin, Hossein Nick Zinat
Piccoli, Benedetto
Numerical Analysis
Machine Learning
We present a neural network-based method for learning scalar hyperbolic conservation laws. Our method replaces the traditional numerical flux in finite volume schemes with a trainable neural network while preserving the conservative structure of the scheme. The model can be trained both in a supervised setting with efficiently generated synthetic data or in an unsupervised manner, leveraging the weak formulation of the partial differential equation. We provide theoretical results that our model can perform arbitrarily well, and provide associated upper bounds on neural network size. Extensive experiments demonstrate that our method often outperforms efficient schemes such as Godunov's scheme, WENO, and Discontinuous Galerkin for comparable computational budgets. Finally, we demonstrate the effectiveness of our method on a traffic prediction task, leveraging field experimental highway data from the Berkeley DeepDrive drone dataset.
title Supervised and Unsupervised Neural Network Solver for First Order Hyperbolic Nonlinear PDEs
topic Numerical Analysis
Machine Learning
url https://arxiv.org/abs/2601.06388