Symmetry-aware Bayesian flow networks for crystal generation
- 1Institute for Anthropomatics and Robotics, Karlsruhe Institute of Technology, Karlsruhe, Germany
- 2Institute of Nanotechnology, Karlsruhe Institute of Technology, Karlsruhe, Germany
Abstract
The discovery of new crystalline materials is essential to scientific and technological progress. However, traditional trial-and-error approaches are inefficient due to the vast search space. Recent advancements in machine learning have enabled generative models to predict new stable materials by incorporating structural symmetries and to condition the generation on desired properties. In this work, we introduce SymmBFN, a novel symmetry-aware Bayesian Flow Network (BFN) for crystalline material generation that accurately reproduces the distribution of space groups found in experimentally observed crystals. SymmBFN substantially improves efficiency, generating stable structures at least one order of magnitude faster than the next-best method, at similar or even superior quality. Furthermore, we demonstrate its capability for property-conditioned generation, enabling the design of materials with tailored properties. Our findings establish BFNs as an effective tool for accelerating the discovery of crystalline materials.
Related
- Symmetry-Aware Flow Matching for End-to-end Molecular Crystal Generation
- PORMAKE2: Topology-Based Generation of Hypothetical Metal-Organic Frameworks through Single-Metal Nodes and Symmetry-Aware Linker Rotation
- Mirror Symmetry. The Mother of all Crystal Symmetries. By M. A. Wahab. Springer Series in Solid-State Sciences, Vol. 200, 2024. Pp. XIII+224. Softcover: ISBN 978-981-99-8363-6 price GBP 38.49. Hardcover: ISBN 978-981-99-8360-5 price GBP 48.99. eBook: ISBN 978-981-99-8361-2 price GBP 43.99
- Realization of a chiral photonic-crystal cavity with broken time-reversal symmetry
- Modulating the polarization-resolved second harmonic generation response of a Sulfamic acid multicomponent crystal via amino acid coformer selection
BibTeX
@article{ruple2026symmetryaware,
title = {{Symmetry-aware Bayesian flow networks for crystal generation}},
author = {Laura Ruple and Luca Torresi and Henrik Schopmans and Pascal Friederich},
journal = {npj Computational Materials},
year = {2026},
month = may,
doi = {10.1038/s41524-026-02140-8},
url = {https://doi.org/10.1038/s41524-026-02140-8},
note = {CC BY},
}
RIS
TY - JOUR
TI - Symmetry-aware Bayesian flow networks for crystal generation
AU - Laura Ruple
AU - Luca Torresi
AU - Henrik Schopmans
AU - Pascal Friederich
PY - 2026
DA - 2026/05/19
JO - npj Computational Materials
DO - 10.1038/s41524-026-02140-8
UR - https://doi.org/10.1038/s41524-026-02140-8
AB - The discovery of new crystalline materials is essential to scientific and technological progress. However, traditional trial-and-error approaches are inefficient due to the vast search space. Recent advancements in machine learning have enabled generative models to predict new stable materials by incorporating structural symmetries and to condition the generation on desired properties. In this work, we introduce SymmBFN, a novel symmetry-aware Bayesian Flow Network (BFN) for crystalline material generation that accurately reproduces the distribution of space groups found in experimentally observed crystals. SymmBFN substantially improves efficiency, generating stable structures at least one order of magnitude faster than the next-best method, at similar or even superior quality. Furthermore, we demonstrate its capability for property-conditioned generation, enabling the design of materials with tailored properties. Our findings establish BFNs as an effective tool for accelerating the discovery of crystalline materials.
SN - 2057-3960
ER -