This package is a work-in-progress for generating inferred signals in dead regions of a DUNE-ND-like liquid argon time projection chamber (LArTPC).
It is a generative Sparse 3D Convolutional Neural Network.
View the plotting notebook with outputs on nbviewer
Examples and preprint coming soon...
- Data Preparation
Place the simulation .npz files for muon-like tracks in the /data directory. A link to download these files is provided in a text file located in /data. Instructions on generating these simulation files will be added here in the future.
-
Your environment needs to have the packages in
requirements.txt -
Instructions to install MinkowksiEngine is coming soon ...
-
Running LArDRIP
To run the model,
First, modify model_twoinactive_new.py to point to the inputs above and place the outputs where you want them to exist.
Then use the following command:
bash python3 /ME_model/model_twoinactive_new.py
If you need to resume training from a specific epoch (for instance, after an interruption), use the following command:
bash python3 /ME_model/model_twoinactive_new.py --resume --checkpoint_path /path/to/checkpoint/checkpoint_epoch_{epoch_number}.pth --wandb_run_id {wandb_id}
{epoch_number}: The epoch from which to resume training. The model will start from this epoch + 1.
{wandb_id}: The Weights and Biases ID for tracking the training session.