Framework turns neural quantum wave functions into interpretable particle pictures

bravo_abad · x · 2026-08-26

Zichen Wang and coauthors introduce a framework that transforms neural-network quantum wave functions, such as FermiNet and DeepSolid, into an intuitive, particle-based picture of electronic structure. By combining neural representations with Periodic Dynamic Voronoi Metropolis Sampling (PDVMS), the method reconstructs individual electron positions by sampling the probability distribution while partitioning the configuration space.

Original post →

More from Embodied

Embodied channel →