Inheritance diagram for FeasibleSegmentPairFilterTrainingRun:

Public Member Functions
def	identifier (self)

def	create_path (self)

Public Attributes
	task
	Process each event according to the user's desired task (train, eval, explore)

Static Public Attributes
int	n_events = 10000
	number of events to generate

str	generator_module = "generic"
	use the generic event generator

os	bkg_files = os.path.join(os.environ["VO_BELLE2_SW_DIR"], "bkg")
	overlay background hits from the events in these files

str	truth = "truth_positive"
	degree of MC truth-matching

Detailed Description

Run to record segment pairs encountered at the SegmentPairCreator and retrain its mva method

Definition at line 19 of file trainFeasibleSegmentPairFilter.py.

Member Function Documentation

◆ create_path()

def create_path ( self )

Setup the recording path after the simulation

Reimplemented from ReadOrGenerateEventsRun.

Definition at line 37 of file trainFeasibleSegmentPairFilter.py.

    def create_path(self):
        """Setup the recording path after the simulation"""
        path = super().create_path()
 
        # In contrast to other training use only the first *half* loop for more aggressive training
        path.add_module("TFCDC_WireHitPreparer",
                        flightTimeEstimation="outwards")
 
        path.add_module('TFCDC_ClusterPreparer',
                        SuperClusterDegree=3,
                        SuperClusterExpandOverApogeeGap=True)
 
        # Also fix the segment orientation to outwards to make training additionally aggressive
        path.add_module("TFCDC_SegmentFinderFacetAutomaton")
 
        
        if self.task == "train":
            varSets = [
                "feasible",
                "filter(truth)",
            ]
 
        elif self.task == "eval":
            varSets = [
                "filter(feasible)",
                "filter(truth)",
            ]
 
        elif self.task == "explore":
            varSets = [
                "basic",
                # "feasible",
                "fitless",
                "pre_fit",
                # "fit",
                "truth",  # for weighting
                # "old_fit",
                # "filter(fitless)",
                # "filter(simple)",
                # "filter(feasible)",
                "filter(truth)",
            ]
 
        else:
            raise ValueError("Unknown task " + self.task)
 
        path.add_module("TFCDC_TrackFinderSegmentPairAutomaton",
                        SegmentPairFilter="unionrecording",
                        SegmentPairFilterParameters={
                            "rootFileName": self.sample_file_name,
                            "varSets": varSets,
                        })
 
        return path
 
 

◆ identifier()

def identifier ( self )

Database identifier of the filter being trained

Definition at line 33 of file trainFeasibleSegmentPairFilter.py.

    def identifier(self):
        """Database identifier of the filter being trained"""
        return "trackfindingcdc_FeasibleSegmentPairFilter.xml"
 