Generating Symmetric Materials using Latent Flow Matching
2026-05-11 • Machine Learning
Machine Learning
AI summaryⓘ
The authors improved a materials generation model called ADiT by making a new version named SymADiT that understands crystal symmetry better. They used a special way of representing materials based on Wyckoff positions, which describe where atoms sit in a crystal. Their model generates materials in a simplified space but ensures the output respects the crystal’s symmetry rules. Tests showed their method creates stable materials that follow symmetry more accurately using a straightforward Transformer design.
materials generationAll-atom Diffusion Transformer (ADiT)SymADiTWyckoff positionscrystal symmetryspace grouplatent spaceTransformer architecture
Authors
Anmar Karmush, Cedric Mathieu Brandenburg, Soheil Ershadrad, Johanna Rosén, Michael Felsberg, Filip Ekström Kelvinius
Abstract
Tackling the task of materials generation, we aim to enhance the previously proposed All-atom Diffusion Transformer (ADiT) by introducing SymADiT, a symmetry-aware variant. To do so, we use a representation of materials based on Wyckoff positions. We follow ADiT and perform generative modelling in latent space, adapted to our symmetry-aware representation. By forcing the output of the generative model to adhere to the symmetry restrictions imposed by the generated crystal's space group and each atom's Wyckoff-position, the generated materials exhibit more realistic symmetry properties. We benchmark our method against both symmetry-aware and symmetry-agnostic models for materials generation and show competitive performance, generating stable, symmetric materials with a simple Transformer architecture.