Event Reconstruction

In this tutorial, we will demonstrate how to reconstruct events with a HyPER model.

Reconstruction

You might have noticed that in the configuration file shown here, there exit few prediction related settings:

predict_set:            null
predict_with:           "cpu"
predict_model:          null
predict_output:         null
topology:               ttbar_allhad

where you can provide the dataset and path to the HyPER model you want to use for the prediction. By default, the first model you trained is saved to HyPER_logs/version_0. The output of the prediction is a .pkl file.

With configuration ready, running event reconstruction via:

python -m HyPER.predict --config-name=default [options]

Similar to Hyper.train, you can overwrite any options by providing them at the end.

Topology

topology defines which topology you wish to reconstruct your events into. These are functions stored in HyPER/topology.

If you wish additional ones to be included, you can create an issue here. Or if you wish to contribute, you can create a pull request here.