A physics-inspired neural network to solve partial differential equations – application in diffusion-induced stress

Abstract

Analyzing and predicting diffusion-induced stress is of paramount importance in understanding structural durability of lithium- and sodium-ion batteries, which generally requires to solve initial-boundary value problems, involving the partial differential...

Leave No Photon Behind: Artificial Intelligence in Multiscale Physics of Photocatalyst and Photoreactor Design

“…This concept of physics‐informed neural networks was popularized by Raissi et al, [ 108 , 109 ] and it has been increasingly applied across studies over the past several years. [ 110 , 111 , 112 , 113 , 114 ] By combining a neural network with physically meaningful constraints, one can obtain a surrogate model for describing complex behavior. Importantly, these physics‐informed models can be trained on small data sets, are compatible with noisy and high‐dimensional data, and are quite capable at solving inverse problems, all of which are relevant to gas‐phase CO 2 heterogeneous photocatalysis.…”