AIMCEG

Instructions

In order to train the generator to produce events, you need to provide data set. The framework currently supports csv, json, and root file formats. You can also pass in your preformatted dataframe.

Format of input data files?

There are certain rules that these different datafiles need to follow.

• The csv file should have headers / Column names.
• For features that belongs to different particles, the naming convention of the column names should be “particleName_featureName”
• The features that does not belong to any particle, their column names should not contain underscore (_)
• Features should include center of mass momentum energy named as “gamma”
• The datafile can have columns that you might not want to utilize for event generator training, you can simply ignore those column with ignoreColumns/ignoreParticles parameter (explained in Tutorials section).
• Each particle that is utilized for training should at least have momentum vector. For example if you want to train the generator for Proton (p), Pi_plus (pip), and Pi_minus (pim) particles’ events, your datafile should at least contain p_x, p_y, p_z, pip_x, pip_y, pip_z, pim_x, pim_y, pim_z, gamma. Note - you can also use px, py, pz for momentum vector features. (ex p_px, p_py, p_pz)

root file

If you are using root file, it should contain root tree(s). If it contains more than one root tree, you need to provide the name of the headNode that contains the dataset of interest. This should be a single level root tree with leaf nodes following above described constraints.

To learn how to use this toolkit, please visit Tutorial page.