Research map / Probabilistic and scientific modeling / Molecules and proteins
MatRIS: Toward Reliable and Efficient Pretrained Machine Learning Interaction Potentials
Yuanchang Zhou, Siyu Hu, Xiangyu Zhang, Hongyu Wang, Guangming Tan, Weile Jia
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 |
|---|---|---|---|
| add dim-wise softmax (per-feature-dimension attention weights) (ablation) | MatRIS-S · WBM subset, MPTrj-trained | Ef MAE (meV/atom): 28.4 -> 28.0 | Improved |
| add separable (two-direction) attention (ablation) | MatRIS-S · WBM subset, MPTrj-trained | Ef MAE (meV/atom): 29.1 -> 28.4 | Improved |
| add learnable envelope function (ablation) | MatRIS-S · WBM subset, MPTrj-trained | Ef MAE (meV/atom): 31.3 -> 29.1 | Improved |
| add denoising pretraining (ablation) | MatRIS-S · WBM subset, MPTrj-trained | Ef MAE (meV/atom): 28.0 -> 27.2 | Improved |
| add magnetic moment prediction as auxiliary task (ablation) | MatRIS-S · WBM subset, MPTrj-trained | Ef MAE (meV/atom): 29.7 -> 28.0 | Improved |
| add graph-level loss reduction (ablation) | MatRIS-S · WBM subset, MPTrj-trained | Ef MAE (meV/atom): 30.2 -> 29.7 | Improved |
| use MatRIS-M/S invariant model instead of larger equivariant models | MatRIS vs eSEN-30M-MP / eqV2 S DeNS · Matbench-Discovery (compliant) | training efficiency: 13.0x and 6.4x training efficiency at comparable accuracy | Improved |
| MatRIS has more layers than MACE-L (4 and 6 vs 2) | MatRIS-S/M · inference time across system sizes | inference time | Hurt |
Closest papers by meaning
- UMA: A Family of Universal Models for Atoms (NeurIPS 2025 · similarity 0.86)
- Learning the Electronic Hamiltonian of Large Atomic Structures (ICML 2025 · similarity 0.86)
- MLIP Arena: Advancing Fairness and Transparency in Machine Learning Interatomic Potentials via an Open, Accessible Benchmark Platform (NeurIPS 2025 · similarity 0.86)
- Quadruple Attention in Many-body Systems for Accurate Molecular Property Predictions (ICML 2025 · similarity 0.86)
- Learning Smooth and Expressive Interatomic Potentials for Physical Property Prediction (ICML 2025 · similarity 0.86)
- Higher-Rank Irreducible Cartesian Tensors for Equivariant Message Passing (NeurIPS 2024 · similarity 0.85)
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.
