PulseAugur
LIVE 09:35:51
research · [2 sources] ·
0
research

AI methods tackle complex nonlinear PDEs with sparse identification

Researchers have developed a novel framework using sparse radial basis function networks to solve nonlinear partial differential equations (PDEs). This approach incorporates sparsity-promoting regularization to manage over-parameterization and reduce redundant features, aiming to improve upon existing methods like physics-informed neural networks and Gaussian processes. The method is grounded in the theory of Reproducing Kernel Banach Spaces and employs a three-phase algorithm for computational efficiency, including adaptive feature selection and pruning. Numerical experiments indicate its effectiveness, particularly in scenarios where it outperforms Gaussian process approaches. AI

Summary written by gemini-2.5-flash-lite from 2 sources. How we write summaries →

IMPACT New methods for solving PDEs could accelerate scientific discovery and engineering simulations by improving computational efficiency and accuracy.

RANK_REASON This cluster contains two academic papers detailing new methods for solving nonlinear PDEs using machine learning techniques.

Read on Hugging Face Daily Papers →

AI methods tackle complex nonlinear PDEs with sparse identification

COVERAGE [2]

  1. arXiv cs.LG TIER_1 · Zihan Shao, Konstantin Pieper, Xiaochuan Tian ·

    Solving Nonlinear PDEs with Sparse Radial Basis Function Networks

    arXiv:2505.07765v3 Announce Type: replace-cross Abstract: We propose a novel framework for solving nonlinear PDEs using sparse radial basis function (RBF) networks. Sparsity-promoting regularization is employed to prevent over-parameterization and reduce redundant features. This …

  2. Hugging Face Daily Papers TIER_1 ·

    Balance-Guided Sparse Identification of Multiscale Nonlinear PDEs with Small-coefficient Terms

    Data-driven discovery of governing equations has advanced significantly in recent years; however, existing methods often struggle in multiscale systems where dynamically significant terms may have small coefficients. Therefore, we propose Balance-Guided SINDy (BG-SINDy) inspired …