PhyNet: Physics Guided Neural Networksfor Particle Drag Force Prediction in Assembly.

evaluation_and_visualization.py

all_models.py

utils.py

Note: PhyDNN is the same as the PhyNet model. It is just named differently.

notebooks/

• DNN.ipynb: This contains the feed-forward neural network model.
• DNN+Pres.ipynb: The DNN+ model with dragforce + pressure field prediction.
• DNN+Vel.ipynb: The DNN+ model with dragforce + velocity field prediction.
• DNN-MT-Pres.ipynb: The DNN multi-task model with dragforce and pressure field prediction as two separate tasks. The architecture has a set of shared layers and a few layers separate for each of the two tasks.
• DNN-MT-Vel.ipynb: The DNN multi-task model with dragforce and velocity field prediction as two separate tasks. The architecture has a set of shared layers and a few layers separate for each of the two tasks.
• PhyDNN.ipynb: This is the proposed model with physics guided architecture and statistical priors incorporated via aggregate supervision.
• PhyDNN-PxTx.ipynb: Similar to PhyDNN.ipynb except that in this model, the drag force components (pressure and shear drag) are predicted only for the x-direction instead of the x,y,z directions.