Flow matching for accelerated simulation of atomic transport in crystalline materials

Juno Nam, Sulin Liu, Gavin Winter, KyuJung Jun, Soojung Yang, Rafael Gómez-Bombarelli

Abstract: Atomic transport underpins the performance of materials in technologies such as energy storage and electronics, yet its simulation remains computationally demanding. In particular, modelling ionic diffusion in solid-state electrolytes requires methods that can overcome the scale limitations of traditional ab initio molecular dynamics. We introduce LiFlow, a generative framework to accelerate MD simulations for crystalline materials that formulates the task as the conditional generation of atomic displacements. The model uses flow matching, with a Propagator submodel to generate atomic displacements and a Corrector to locally correct unphysical geometries, and incorporates an adaptive prior based on the Maxwell–Boltzmann distribution to account for chemical and thermal conditions. We benchmark LiFlow on a dataset comprising 25-ps trajectories of lithium diffusion across 4,186 solid-state electrolyte candidates at four temperatures. The model obtains a consistent Spearman rank correlation of 0.7–0.8 for lithium mean squared displacement predictions on unseen compositions. Furthermore, LiFlow generalizes from short training trajectories to larger supercells and longer simulations and maintains high accuracy. With speed-ups of up to 600,000× compared with first-principles methods, LiFlow enables scalable simulations at significantly larger length scales and timescales. A generative framework that accelerates the simulations of atomic transport in crystalline solids is developed, enabling large-scale screening and extending simulations to larger spatiotemporal scales for energy storage materials.

External IDs:doi:10.1038/s42256-025-01125-4