Research map / Probabilistic and scientific modeling / Molecules and proteins
Machine learning for physical simulation
Builds datasets, benchmarks and graph models for fluid and multiphysics simulation and design.
96 accepted papers at ICML, NeurIPS, ICLR, CVPR and AAAI (2016–2026) · an approach in Molecules and proteins, keyword cluster “decision · machine · statistical”
State History Tried Checklist Preprints Who Export
Where it stands
Not enough benchmark data Fewer than four papers report the same benchmark, so there is no record to follow.
32 papers in 2023–24 and 45 in 2025–26 (at ICML, NeurIPS, ICLR, CVPR and AAAI); its share of the topic shrank, from 6.5% to 4.9%.
How this is decided
A benchmark is tracked when at least four papers of this approach report their own method on the same dataset and metric in at least two different years. Its record is the best number reported up to 2024 compared with the best reported in 2025–26: still improving means at least 1% better, flat means at most 0.2% better with at least two papers from 2025–26 trying. Numbers on a different scale from the rest (a fraction among percentages) and far outliers are left out. Results come from different papers with different setups, so they are signals, not head-to-head comparisons. The share of papers is counted at ICML, NeurIPS, ICLR, CVPR and AAAI.
How it got here
Most cited paper of each year
- 2016 · On the Analysis of Complex Backup Strategies in Monte Carlo Tree Search (ICML 2016 · 14 citations)
- 2018 · Scalable End-to-End Autonomous Vehicle Testing via Rare-event Simulation (NeurIPS 2018 · 39 citations)
- 2020 · Learning Composable Energy Surrogates for PDE Order Reduction (NeurIPS 2020 · 9 citations)
- 2021 · Reinforcement learning for optimization of variational quantum circuit architectures (NeurIPS 2021 · 51 citations)
- 2022 · Qubit Routing Using Graph Neural Network Aided Monte Carlo Tree Search (AAAI 2022 · 24 citations)
- 2023 · ClimaX: A foundation model for weather and climate (ICML 2023 · 167 citations)
- 2024 · Learning to Predict Structural Vibrations (NeurIPS 2024 · 4 citations)
- 2025 · Physics-Guided Foundation Model for Scientific Discovery: An Application to Aquatic Science (AAAI 2025 · 9 citations)
- 2026 · Principle-Evolvable Scientific Discovery via Uncertainty Minimization (ICML 2026 · 0 citations)
Earliest papers here
- On the Analysis of Complex Backup Strategies in Monte Carlo Tree Search (ICML 2016 · 14 citations)
- Scalable End-to-End Autonomous Vehicle Testing via Rare-event Simulation (NeurIPS 2018 · 39 citations)
- Inference in Higher Order MRF-MAP Problems With Small and Large Cliques (CVPR 2018 · 5 citations)
What has been tried
70 papers here report 707 measured changes (a component added, removed or swapped, compared with a baseline): helped 67%, no clear effect 5%, hurt 21%, mixed 7%. Papers mostly publish changes that helped, so "helped" is over-represented.
Ablations in recent papers
| What was changed | Measured on | Result | Effect | Paper |
|---|---|---|---|---|
| Add pre-training | Water temperature, summer · RMSE | 3.578 (0.938) -> 1.953 (0.126) | Helped | Physics-Guided Foundation Model for Scientific Discovery: An Application to Aquatic Science (AAAI 2025 · 9 citations) |
| Increase the objective-loss weight w1 | Validation set · average weight | Helped | Deep Generative Model for Mechanical System Configuration Design (AAAI 2025 · 5 citations) | |
| Use a directional autoencoder instead of a traditional autoencoder | Nanophotonics · correlation of latent variables with physical quantities | Helped | AutoSciLab: A Self-Driving Laboratory for Interpretable Scientific Discovery (AAAI 2025 · 4 citations) | |
| Use frequency queries instead of predicting all frequencies on a fixed grid | V-5000 · MSE | 0.19 -> 0.08 | Helped | Learning to Predict Structural Vibrations (NeurIPS 2024 · 4 citations) |
| warm up with PtO MSE training, then fine-tune with Blackbox PnO | 7 benchmark datasets · regret | Mixed | Benchmarking PtO and PnO Methods in the Predictive Combinatorial Optimization Regime (NeurIPS 2024 · 3 citations) | |
| Use y(r) features instead of x(r) | W4-17 · RMSE | 2.35 -> 1.95 | Helped | Learning Generalized Residual Exchange-Correlation-Uncertain Functional for Density Functional Theory (AAAI 2025 · 2 citations) |
| increase the number of training samples | 5G-Urban · nMSE | Helped | Physics-Informed Generative Modeling of Wireless Channels (ICML 2025 · 1 citation) | |
| Use selection-only uncertainty adaptation | Space Invaders, offline · average reward | Helped | Monte Carlo Tree Search in the Presence of Transition Uncertainty (AAAI 2024 · 1 citation) | |
| Add GTO and HFM to PW-only prediction | Benzene · NMAE (%) | 6.92 -> 2.45 | Helped | Gaussian Plane-Wave Neural Operator for Electron Density Estimation (ICML 2024 · 1 citation) |
| Add neural Fourier layers instead of standard CNN layers | ERA5 data assimilation, 1.40625° observations · MSE | 0.0230 -> 0.0202 | Helped | FNP: Fourier Neural Processes for Arbitrary-Resolution Data Assimilation (NeurIPS 2024 · 1 citation) |
| Add stochastic depth to the interpolator | Interpolated timesteps 1 to 5 · spread-skill ratio | around 0.26 to 0.35 | Helped | Probablistic Emulation of a Global Climate Model with Spherical DYffusion (NeurIPS 2024 · 1 citation) |
| Remove the overlap matrix input | Hamiltonian prediction (ablation; figure) · Hamiltonian MAE | Hurt | Infusing Self-Consistency into Density Functional Theory Hamiltonian Prediction via Deep Equilibrium Models (NeurIPS 2024 · 1 citation) |
What reviewers will expect
Datasets and baselines named in the 34 papers of this approach from 2025–2026 that have a full-text summary. Leaving out the common ones is a frequent review question.
Baselines
- Diffusion Forcing · 3 of 34 papers
- MeshGraphNet · 2 of 34 papers
- FNO · 2 of 34 papers
- GNOT · 2 of 34 papers
- Diffuser · 2 of 34 papers
- GINO · 2 of 34 papers
- Transolver · 2 of 34 papers
- Diffuser-Replan · 2 of 34 papers
Recent preprints
Loading recent arXiv preprints…
Who publishes on it
Institutions on the most papers of this approach since 2023 (of 77 papers; from OpenAlex and the papers' affiliations).
- Microsoft · 5 papers
- University of California, Irvine · 4 papers
- Massachusetts Institute of Technology · 4 papers
- University of Science and Technology of China · 3 papers
- University of California, Los Angeles · 3 papers
- NVIDIA · 3 papers
- Princeton University · 3 papers
- Korea Advanced Institute of Science and Technology · 3 papers
- SAP Labs Korea · 3 papers
- Shanghai Jiao Tong University · 3 papers
Use it
Download BibTeX of 38 key papersOpen the topic in the interactive map
Get alerts. Describe your idea in scime atlas and save it (free scime account): you get a weekly email when new preprints land close to it.
Related approaches
Same keyword cluster
- Prediction-powered and conformal inference (129 papers)
- Statistical analysis of ML practice (122 papers)
- Counterfactual and statistical explanations (81 papers)
- Sparse concepts and hypothesis testing (78 papers)
- Cold-start recommendation models (80 papers)
Elsewhere in the topic
- Generative protein structure design (142 papers)
- Neuro-symbolic probabilistic layers (120 papers)
- Structure-based 3D molecule generation (107 papers)
- Sparse graphical and tensor models (97 papers)
- Flow matching generative modeling (97 papers)
- Multimodal trajectory prediction (95 papers)
- Probabilistic embeddings and relational inference (93 papers)
- Equivariant and disentangled latent representations (90 papers)
Benchmark numbers, ablations and baseline numbers are extracted by AI from each paper's full text and can be wrong; the paper is the authority. Data: results index of 2026-11-02, citation counts from OpenAlex (2026-10-09).