Research map / Probabilistic and scientific modeling / Molecules and proteins
Large-Scale Molecular Dynamics Simulations: Direct Interatomic Modeling with Dilated Message Passing
Haokai Hong, Wanyu Lin, KC Tan
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 |
|---|---|---|---|
| use dilated distance-ranking K-star message passing instead of baselines on large protein-ligand complexes | DKMP-R · MISATO test (filtered) | N-MSE: 5.76 (EGNO) -> 0.92 | Improved |
| use dilated radius cutoff intervals instead of other ML force fields | DKMP-C · Chignolin | force MAE: 0.187 (LSRM) -> 0.126 | Improved |
| replace dilated distance ranking with a single KNN graph with the same total edges (ablation) | DKMP-R · AdK backbone / full-atom | F-MSE: 1.900 -> 2.739 (backbone); OOM on full-atom | Hurt |
| use distance-based grouping rather than random edge partition across layers (ablation) | DKMP-R · AdK backbone / full-atom | F-MSE: 2.962 -> 1.900 (backbone); 3.253 -> 2.715 (full-atom) | Improved |
| avoid edge reuse across layers (disjoint dilated edge sets) (ablation) | DKMP-R · AdK backbone / full-atom | F-MSE: 2.383 -> 1.900 (backbone); 2.740 -> 2.715 (full-atom) | Improved |
| enlarge the radius cutoff of baseline message passing models (ablation) | PaiNN, ViSNet, LSRM, DKMP-C · Chignolin | force MAE | Hurt |
| increase MPNN depth for baselines versus DKMP-C (ablation) | ViSNet, DKMP-C · Chignolin | force MAE | Mixed |
| randomly rotate the test set (DKMP-R is not equivariant) (ablation) | DKMP-R · MISATO test (filtered) | N-MSE: 0.92 -> 0.94 | No clear effect |
| add DKMP-C dilated message passing for NVE dynamics | DKMP-C · Chignolin NVE 100 ps | energy drift | Improved |
Closest papers by meaning
- Long-Short-Range Message-Passing: A Physics-Informed Framework to Capture Non-Local Interaction for Scalable Molecular Dynamics Simulation (ICLR 2024 · similarity 0.89)
- The Importance of Being Scalable: Improving the Speed and Accuracy of Neural Network Interatomic Potentials Across Chemical Domains (NeurIPS 2024 · similarity 0.89)
- Generative Modeling of Full-Atom Protein Conformations using Latent Diffusion on Graph Embeddings (NeurIPS 2025 · similarity 0.89)
- Neural Atoms: Propagating Long-range Interaction in Molecular Graphs through Efficient Communication Channel (ICLR 2024 · similarity 0.88)
- UniSim: A Unified Simulator for Time-Coarsened Dynamics of Biomolecules (ICML 2025 · similarity 0.88)
- Deep Signature: Characterization of Large-Scale Molecular Dynamics (ICLR 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.
