Source code for gammapy.makers.background.phase

# Licensed under a 3-clause BSD style license - see LICENSE.rst
import numpy as np
from regions import PointSkyRegion
from gammapy.data import EventList
from gammapy.datasets import MapDatasetOnOff, SpectrumDataset
from gammapy.makers.utils import make_counts_rad_max
from gammapy.maps import Map
from ..core import Maker

__all__ = ["PhaseBackgroundMaker"]


[docs]class PhaseBackgroundMaker(Maker): """Background estimation with on and off phases. Parameters ---------- on_phase : `tuple` or list of tuples On-phase defined by the two edges of each interval (edges are excluded). off_phase : `tuple` or list of tuples Off-phase defined by the two edges of each interval (edges are excluded). phase_column_name : `str`, optional The name of the column in the event file from which the phase information are extracted. Default is 'PHASE'. """ tag = "PhaseBackgroundMaker" def __init__(self, on_phase, off_phase, phase_column_name="PHASE"): self.on_phase = self._check_intervals(on_phase) self.off_phase = self._check_intervals(off_phase) self.phase_column_name = phase_column_name def __str__(self): s = self.__class__.__name__ s += f"\nOn phase interval : {self.on_phase}" s += f"\nOff phase interval : {self.off_phase}" s += f"\nPhase column name : {self.phase_column_name}" return s @staticmethod def _make_counts(dataset, observation, phases, phase_column_name): event_lists = [] for interval in phases: events = observation.events.select_parameter( parameter=phase_column_name, band=interval ) event_lists.append(events) events = EventList.from_stack(event_lists) geom = dataset.counts.geom if geom.is_region and isinstance(geom.region, PointSkyRegion): counts = make_counts_rad_max(geom, observation.rad_max, events) else: counts = Map.from_geom(geom) counts.fill_events(events) return counts
[docs] def make_counts_off(self, dataset, observation): """Make off counts. Parameters ---------- dataset : `SpectrumDataset` Input dataset. observation : `DatastoreObservation` Data store observation. Returns ------- counts_off : `RegionNDMap` Off counts. """ return self._make_counts( dataset, observation, self.off_phase, self.phase_column_name )
[docs] def make_counts(self, dataset, observation): """Make on counts. Parameters ---------- dataset : `SpectrumDataset` Input dataset. observation : `DatastoreObservation` Data store observation. Returns ------- counts : `RegionNDMap` On counts. """ return self._make_counts( dataset, observation, self.on_phase, self.phase_column_name )
[docs] def run(self, dataset, observation): """Make on off dataset. Parameters ---------- dataset : `SpectrumDataset` or `MapDataset` Input dataset. observation : `Observation` Data store observation. Returns ------- dataset_on_off : `SpectrumDatasetOnOff` or `MapDatasetOnOff` On off dataset. """ counts_off = self.make_counts_off(dataset, observation) counts = self.make_counts(dataset, observation) acceptance = Map.from_geom(geom=dataset.counts.geom) acceptance.data = np.sum([_[1] - _[0] for _ in self.on_phase]) acceptance_off = Map.from_geom(geom=dataset.counts.geom) acceptance_off.data = np.sum([_[1] - _[0] for _ in self.off_phase]) dataset_on_off = MapDatasetOnOff.from_map_dataset( dataset=dataset, counts_off=counts_off, acceptance=acceptance, acceptance_off=acceptance_off, ) dataset_on_off.counts = counts if isinstance(dataset, SpectrumDataset): dataset_on_off = dataset_on_off.to_spectrum_dataset(dataset._geom.region) return dataset_on_off
@staticmethod def _check_intervals(intervals): """Split phase intervals that go below phase 0 and above phase 1. Parameters ---------- intervals: `tuple`or list of `tuple` Phase interval or list of phase intervals to check. Returns ------- intervals: list of `tuple` Phase interval checked. """ if isinstance(intervals, tuple): intervals = [intervals] for phase_interval in intervals: if phase_interval[0] % 1 > phase_interval[1] % 1: intervals.remove(phase_interval) intervals.append((phase_interval[0] % 1, 1)) if phase_interval[1] % 1 != 0: intervals.append((0, phase_interval[1] % 1)) return intervals