Reproduce “Space Group Constrained Crystal Generation” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Rui Jiao, Wenbing Huang, Yu Liu, Deli Zhao, Yang Liu
DiffCSP++, a diffusion model from Tsinghua University and Alibaba Group, integrates space group symmetry into crystal structure generation by formulating constraints for lattice matrices and fractional coordinates. This model achieves high accuracy in crystal structure prediction tasks and generates materials with physical properties that closely align with real-world distributions.
Key insights
- Explicitly incorporating space group symmetries into generative models drastically improves the accuracy and physical validity of generated crystal structures.
- Complex crystallographic symmetries can be systematically integrated into a diffusion framework by formulating them as tractable constraints on lattice parameters and atomic positions.
- The method enables controllable crystal generation, allowing for targeted materials design by conditioning the model on specific, user-defined space groups.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.