trainFacetRelationFilter.py

#!/usr/bin/env python3
 

 
# This is for training a mva classifier for relations between hit triplets
# It performs a tip better than the current default chi2 filter
# However run time is quite a bit slower which is why it won't be used in the standard chain.
# Nevertheless this script can be used to generate variables to consider
# for improvements or cross checks with --task explore
 
import sys
 
from tracking.run.event_generation import StandardEventGenerationRun
from trackfindingcdc.run.training import TrainingRunMixin
 
 
class FacetRelationFilterTrainingRun(TrainingRunMixin, StandardEventGenerationRun):
    """Run for recording facets encountered at the filter"""
    ## Suggested number for this analysis
    n_events = 1000
 
    ## Suggested generator module
    generator_module = "generic"
    # detector_setup = "TrackingDetector"
 
    ## Default task set to explore
    task = "explore"
 
    ## Name of the truth variables
    truth = "truth_positive"
 
    @property
    def identifier(self):
        """Database identifier of the filter being trained"""
        return "trackfindingcdc_FacetRelationFilter.xml"
 
    def create_path(self):
        """Setup the recording path after the simulation"""
        path = super().create_path()
 
        ## Post-process events according to the user's desired task (train, eval, explore)
        if self.task == "train":
            var_sets = [
                "mva",
                "filter(truth)",
            ]
 
        elif self.task == "eval":
            var_sets = [
                "basic",
                "filter(chi2)",
                "filter(chi2_old)",
                "filter(simple)",
                "filter(truth)",
            ]
 
            self.variables = [
                "chi2_weight",
                "chi2_accept",
                "chi2_old_weight",
                "chi2_old_accept",
                "simple_accept",
            ]
 
            self.groupby = ["", "superlayer_id"]
            self.auxiliaries = [
                "superlayer_id",
            ]
 
        elif self.task == "explore":
            var_sets = [
                "basic",
                "bend",
                "fit",
                "filter(chi2)",
                "filter(simple)",
                "filter(truth)",
            ]
 
            ## Signal some variables to select in the classification analysis
            self.variables = [
                # "delta_phi",
                # "delta_phi_pull",
                # "delta_phi_pull_per_r",
                # "delta_curv",
                # "delta_curv_pull",
                # "delta_curv_pull_per_r",
 
                # "cos_delta",
                # "from_middle_cos_delta",
                # "to_middle_cos_delta",
 
                "chi2_0",
                # "chi2_0_per_s",
                # "erf_0",
                # "fit_0_phi0",
                # "fit_0_cos_delta",
 
                # "chi2_1",
                # "chi2_1_per_s",
                # "fit_1_phi0",
                # "fit_1_cos_delta",
 
                # "chi2",
                # "chi2_per_s",
                # "fit_phi0",
                # "fit_cos_delta",
 
                # "phi0_from_sigma",
                # "phi0_to_sigma",
 
                # "phi0_ref_pull",
                # "phi0_ref_diff",
                # "phi0_ref_sigma",
 
                # "chi2_comb",
                # "phi0_comb_pull",
                # "phi0_comb_diff",
                # "phi0_comb_sigma",
 
                # "chi2_kari_unit",
                # "abs_curv_unit",
 
                # "chi2_kari_l",
                # "abs_curv_l",
 
                # "chi2_kari_pseudo",
                # "abs_curv_pseudo",
 
                # "chi2_kari_proper",
                # "abs_curv_proper",
            ]
 
            ## group output by superlayer_id
            self.groupby = ["", "superlayer_id"]
            ## auxiliary histograms
            self.auxiliaries = [
                "superlayer_id",
            ]
 
        path.add_module("TFCDC_WireHitPreparer",
                        flightTimeEstimation="outwards",
                        UseNLoops=1.0)
 
        path.add_module("TFCDC_ClusterPreparer")
 
        path.add_module("TFCDC_SegmentFinderFacetAutomaton",
                        FacetRelationFilter="unionrecording",
                        FacetRelationFilterParameters={
                            "rootFileName": self.sample_file_name,
                            "varSets": var_sets,
                        })
 
        return path
 
 
def main():
    """Execute the facet relation recording"""
    run = FacetRelationFilterTrainingRun()
    run.configure_and_execute_from_commandline()
 
 
if __name__ == "__main__":
    import logging
    logging.basicConfig(stream=sys.stdout, level=logging.INFO, format='%(levelname)s:%(message)s')
    main()