Research map / Image generation / Autoregressive and latent generation
Crystal Structure Prediction by Joint Equivariant Diffusion
Rui Jiao, Wenbing Huang, Peijia Lin, Jiaqi Han, Pin Chen, Yutong Lu, Yang Liu
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| jointly diffuse lattice and fractional coordinates instead of lattice first then coordinates (ablation) | DiffCSP · MP-20 CSP | Match rate (%): 50.03 -> 51.49 | Improved |
| jointly diffuse lattice and fractional coordinates instead of coordinates first then lattice (ablation) | DiffCSP · MP-20 CSP | Match rate (%): 36.73 -> 51.49 | Improved |
| add O(3) equivariance via inner product L^T L (ablation) | DiffCSP · MP-20 CSP | Match rate (%): 1.66 -> 51.49 | Improved |
| add chirality (determinant sign) as input (ablation) | DiffCSP · MP-20 CSP | Match rate (%): 51.49 -> 49.68 | No clear effect |
| use wrapped normal score matching for F instead of Gaussian DDPM (ablation) | DiffCSP · MP-20 CSP | Match rate (%): 34.09 -> 51.49 | Improved |
| add Fourier transform of relative fractional coordinates (ablation) | DiffCSP · MP-20 CSP | Match rate (%): 29.15 -> 51.49 | Improved |
| use fully connected graph instead of multi-graph edges (ablation) | DiffCSP · MP-20 CSP | Match rate (%): 25.85 -> 51.49 | Improved |
| use DiffCSP instead of CDVAE (joint diffusion, fractional coords) | CSP baselines · MP-20 CSP, 20 samples | Match rate (%): 66.95 -> 77.93 | Improved |
| use DiffCSP instead of USPEX DFT search | 15 MP-20 compounds · match rate and time | Match rate (%): 53.33 -> 73.33; 12.5h -> 10s | Improved |
| use DiffCSP instead of CDVAE for ab initio generation | MP-20 ab initio · Composition validity | Composition validity (%): 86.70 -> 83.25 | Hurt |
| use DiffCSP instead of CDVAE for ab initio generation | MP-20 ab initio · Property statistics | d_E: 0.2778 -> 0.1247 | Improved |
| use DiffCSP instead of CDVAE (diversity of 20 samples) | diversity · Perov-5 vs MP-20 mean fingerprint distance | Diversity mean: Perov-5 0.3249 -> 0.3860; MP-20 0.3129 -> 0.2030 | Mixed |
Closest papers by meaning
- Equivariant Diffusion for Crystal Structure Prediction (ICML 2024 · similarity 0.93)
- WyckoffDiff -- A Generative Diffusion Model for Crystal Symmetry (ICML 2025 · similarity 0.91)
- Space Group Constrained Crystal Generation (ICLR 2024 · similarity 0.91)
- Vector Field Oriented Diffusion Model for Crystal Material Generation (AAAI 2024 · similarity 0.91)
- Kinetic Langevin Diffusion for Crystalline Materials Generation (ICML 2025 · similarity 0.90)
- SymmCD: Symmetry-Preserving Crystal Generation with Diffusion Models (ICLR 2025 · similarity 0.90)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
