PIL-Net: a physics-informed graph convolutional network for predicting atomic multipoles

2025/01/01 by Caitlin Whitter, Alex Pothen, Aurora E. Clark · 1 voice
Biochemistry, Genetics and Molecular Biology · Materials Science · Physics and Astronomy · #Advanced Chemical Physics Studies #Machine Learning in Materials Science #Protein Structure and Dynamics

paper · pdf · doi:10.1039/d5dd00228a

openalex publication_date 2025/01/01 · openalex created_date 2025/10/10 · openalex updated_date 2026/08/01

Abstract

PIL-Net is a physics-informed graph convolutional network capable of predicting atomic multipole moments quickly and with high accuracy.

Citations

• FreeSolv: a database of experimental and calculated hydration free energies, with input files
• Continuously Differentiable Exponential Linear Units
• Less is more: Sampling chemical space with active learning
• P SI4 1.4: Open-source software for high-throughput quantum chemistry
• Physics-informed machine learning
• Highly accurate protein structure prediction with AlphaFold
• E(3)-equivariant graph neural networks for data-efficient and accurate interatomic potentials
• e3nn: Euclidean Neural Networks
• Integrating Scientific Knowledge with Machine Learning for Engineering and Environmental Systems
• Directional Message Passing for Molecular Graphs
• Introduction to Electrodynamics
• Deep Graph Library: A Graph-Centric, Highly-Performant Package for Graph Neural Networks

Discussions

• New featured paper by Whitter Pothen and Clark (@chemnetworks.bsky.social): PIL-Net is a physics-informed machine learning model that predicts atomic multipole moments quickly and with low error, achi [bsky, 0 points, 0 comments]