| Directory | Purpose |
|---|---|
| eve/2d/ · eve/3d/ | EvE agent evolution framework, experiment configurations, prompts and initial guidance |
| simulation/2d/ | 2D swimmer geometry, dynamics and CFD evaluation |
| simulation/3d/experiments/ | 3D swimmer simulation, moving-window evaluation and flow rendering; experiment settings are in configs/ |
| simulation/validation/ | Foil, cylinder and moving-window validation experiments |
| generalization/ | Fixed policies and configurations for the thirteen generalization conditions |
| drl/ | Behavior cloning, PPO training and policy inference |
| reproduction/ | Extract plotting inputs from experimental records and check numerical consistency |
| plotting/ | Generate manuscript panels and tables from derived_data/ |
For extraction and plotting commands, see the main README. Simulation setup is described in the CFD validation guide; fixed-policy evaluations are described in the generalization guide. DRL entry points and dependencies are listed in drl/index.json. Each runtime includes its own dependency files.
The 2D and 3D
run archives contain policies/, guidance/ and agent_calls/.
Policy scores are grouped by dimension in raw_data/policy_scores/.