Research map / Graph learning / GNN architectures
Equivariant message passing for the prediction of tensorial properties and molecular spectra
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| Use PaiNN rather than distance-based or angular invariant message passing | QM9 dipole magnitude prediction · QM9 | MAE (D): 0.030 -> 0.012 | Helped |
| Use PaiNN rather than the strongest listed baseline | QM9 electronic spatial extent · QM9 | MAE (a₀²): 0.073 -> 0.066 | Helped |
| Use PaiNN rather than the strongest listed baseline | QM9 U₀ prediction · QM9 | MAE (meV): 6.32 -> 5.85 | Helped |
| Use PaiNN rather than the strongest listed baseline | QM9 HOMO-LUMO gap prediction · QM9 | MAE (meV): 32.6 -> 45.7 | Hurt |
| Use PaiNN rather than the strongest listed baseline | QM9 isotropic polarizability prediction · QM9 | MAE (a₀³): 0.044 -> 0.045 | No clear effect |
| Add scalar products of vector features to the update (ablation) | PaiNN · MD17 aspirin | energy MAE (kcal/mol): 0.173 -> 0.159 | Helped |
| Add scalar products of vector features to the update (ablation) | PaiNN · MD17 aspirin | force MAE (kcal/mol/Å): 0.420 -> 0.371 | Helped |
| Add vector-feature convolution to propagate directional information (ablation) | PaiNN · MD17 aspirin | energy MAE (kcal/mol): 0.183 -> 0.159 | Helped |
| Add vector-feature convolution to propagate directional information (ablation) | PaiNN · MD17 aspirin | force MAE (kcal/mol/Å): 0.441 -> 0.371 | Helped |
| Add vector features and their interactions (ablation) | PaiNN · MD17 aspirin | force MAE (kcal/mol/Å): 1.194 -> 0.371 | Helped |
| Predict dipoles using local vector features as well as latent charges (ablation) | PaiNN · Molecular dipoles | MAE (D): 0.020 -> 0.009 | Helped |
| Train PaiNN on forces only rather than energies and forces | PaiNN · MD17 aspirin | force MAE (kcal/mol/Å): 0.371 -> 0.338 | Helped |
Closest papers by meaning
- Higher-Rank Irreducible Cartesian Tensors for Equivariant Message Passing (NeurIPS 2024 · similarity 0.92)
- Efficient and Scalable Density Functional Theory Hamiltonian Prediction through Adaptive Sparsity (ICML 2025 · similarity 0.90)
- Learning the Electronic Hamiltonian of Large Atomic Structures (ICML 2025 · similarity 0.89)
- Learning Equivariant Non-Local Electron Density Functionals (ICLR 2025 · similarity 0.89)
- Equivariant Masked Position Prediction for Efficient Molecular Representation (ICLR 2025 · similarity 0.88)
- ELoRA: Low-Rank Adaptation for Equivariant GNNs (ICML 2025 · similarity 0.88)
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.