Zoo catalogue¶
Every pre-trained model and dataset of pp.zoo / pp.datasets, generated from the zoo manifest by pp.zoo.catalogue(). Each item passed the zoo's gate; the candidates that did not, and why, are in the inventory. How to use them: Pre-trained models and datasets.
Models¶
| Model | Version | Task | Held-out metrics (reference) | Trust report | Size | Hugging Face |
|---|---|---|---|---|---|---|
airfoil_cst_rans_mlp |
1.0.0 | Airfoil aerodynamic coefficients (Cl, Cd, Cm) from shape and angle of attack, 2-D steady RANS | test/Cl_mae 0.00285, test/Cd_mape_pct 0.545, test/Cm_mae 0.000454 | APPROVED | 0.42 MB | PINNeAPPle-Labs/airfoil-cst-rans-mlp |
heatsink_forced_mlp |
1.0.0 | Plate-fin heat sink in ducted forced convection: base-to-air thermal resistance and pressure drop | test/R_mean_abs_pct 0.712, test/dP_mean_abs_pct 1.04 | APPROVED | 0.22 MB | PINNeAPPle-Labs/heatsink-forced-mlp |
heatsink_natural_mlp |
1.0.0 | Plate-fin heat sink in natural convection (vertical fins, free air): base-to-air thermal resistance | test/R_mean_abs_pct 0.469 | APPROVED | 0.22 MB | PINNeAPPle-Labs/heatsink-natural-mlp |
rover_contact_grc1_viper |
1.1.0 | Wheel-soil contact forces of a rigid grousered wheel on lunar regolith simulant (terramechanics) | test/Fx_rmse_N 0.22, test/Fz_rmse_N 0.47, test/My_rmse_Nm 0.0707 | APPROVED | 0.33 MB | PINNeAPPle-Labs/rover-contact-grc1-viper |
Metric names are <split>/<metric>; the full set, the validity range and the physics checks are in each model card (pp.zoo.info(name)).
Datasets¶
| Dataset | Version | Content | Splits (rows) | Licence | Size | Hugging Face |
|---|---|---|---|---|---|---|
airfoil_rans_cst_re4e6 |
1.0.0 | Airfoil RANS campaign, CST shapes, Re 4e6 (OpenFOAM, k-omega SST) | train 340, test 60, naca_reference 36, xfoil_naca2412 9, grid_study 3 | cc-by-4.0 | 29 kB (bundled) | PINNeAPPle-Labs/airfoil-rans-cst-re4e6 |
heatsink_forced_convection |
1.0.0 | Plate-fin heat sinks in ducted forced convection (6000 physics-engine solves) | test 900, cal 600, val 600, train 3900 | cc-by-4.0 | 0.75 MB (bundled) | PINNeAPPle-Labs/heatsink-forced-convection |
heatsink_natural_convection |
1.0.0 | Plate-fin heat sinks in natural convection (5000 physics-engine solves) | test 750, cal 500, val 500, train 3250 | cc-by-4.0 | 0.55 MB (bundled) | PINNeAPPle-Labs/heatsink-natural-convection |
rover_contact_grc1_viper |
1.1.0 | Wong-Reece wheel-soil contact forces, GRC-1 simulant, VIPER wheel | train 3600, test 2000 | cc-by-4.0 | 0.19 MB (bundled) | PINNeAPPle-Labs/rover-contact-grc1-viper |
single_wheel_tests |
1.0.0 | Published single-wheel testbed points (drawbar pull, sinkage, torque vs slip) | all 27 | mit | 3 kB (bundled) | PINNeAPPle-Labs/single-wheel-tests |
burgers_1d_li2021 |
1.0.0 | 1-D viscous Burgers, Li et al. (2021) operator-learning benchmark (1200 trajectories, 1024 points) | train 1000, test 200 | cc-by-4.0 | 55 MB (Hub-only) | PINNeAPPle-Labs/burgers-1d-li2021 |
darcy_2d_li2021 |
1.0.0 | 2-D Darcy flow, Li et al. (2021) operator-learning benchmark (1200 coefficient / pressure pairs, 128 x 128) | train 1000, test 200 | cc-by-4.0 | 71 MB (Hub-only) | PINNeAPPle-Labs/darcy-2d-li2021 |
navier_stokes_2d_li2021_nu1e-3 |
1.0.0 | 2-D incompressible Navier-Stokes (vorticity), nu = 1e-3, Li et al. (2021) benchmark (1200 runs, 64 x 64, t 0-50) | train 1000, test 200 | cc-by-4.0 | 919 MB (Hub-only) | PINNeAPPle-Labs/navier-stokes-2d-li2021-nu1e-3 |
airfoil_rans_cst_re4e6_fields |
1.0.0 | Airfoil RANS flow fields, CST shapes, Re 4e6 (436 OpenFOAM runs, 20064-cell O-grid) | train 340, test 60, naca_reference 36 | cc-by-4.0 | 186 MB (Hub-only) | PINNeAPPle-Labs/airfoil-rans-cst-re4e6-fields |
thermal_channel_cfd |
1.0.0 | Actuated heated-channel CFD, 2-D incompressible flow + heat (82 trajectories, 128 x 32, 101 snapshots) | train 48, val 8, test_same_family 12, test_unseen_sinusoid 8, test_unseen_edge_steps 6 | cc-by-4.0 | 360 MB (Hub-only) | PINNeAPPle-Labs/thermal-channel-cfd |
cylinder_wake_lbm |
1.0.0 | Confined cylinder wake, lattice-Boltzmann, Re 20-100 (9 runs, 720 vorticity snapshots, 240 x 100) | train 480, test 240 | cc-by-4.0 | 64 MB (Hub-only) | PINNeAPPle-Labs/cylinder-wake-lbm |
blackhole_riaf_density_256 |
1.0.0 | Accretion onto a Schwarzschild black hole: 2.5-D viscous hydro density frames (256 x 128) | convergence 12, grid_r 192, grid_theta 96, ood_alpha0.3 401, test 401, train 1604, val 401 | cc-by-4.0 | 91 MB (Hub-only) | PINNeAPPle-Labs/blackhole-riaf-density-256 |
4 models and 12 datasets, 1748.7 MB in total (2.7 MB bundled with the package; the rest is downloaded on first use and checked against its SHA-256).
Cite¶
Each Hub card carries its own citation. For the software, cite PINNeAPPle (Zenodo DOI 10.5281/zenodo.23343062, CITATION.cff).