TeaNet paper compares RDF curves against ReaxFF and experiment

CatAstro_Piyush · x · 2026-07-26

A post about the paper TeaNet: Universal neural network interatomic potential inspired by iterative electronic relaxations.

The image shows radial distribution function plots comparing TeaNet, ReaxFF, and experiment, suggesting TeaNet better captures several structural features in molecular/materials simulation. The author notes they had not heard of TeaNet before and highlights the paper’s route from genetic algorithms for crystal prediction to graph-convolution-based MLIPs.

Original post →

More from Research

Research channel →