def process(self, obs):
"""Estimate background for one observation."""
all_events = obs.events.select_circular_region(self.on_region)
self.on_phase = self._check_intervals(self.on_phase)
self.off_phase = self._check_intervals(self.off_phase)
# Loop over all ON- and OFF- phase intervals to filter the ON- and OFF- events
list_on_events = [
self.filter_events(all_events, each_on_phase)
for each_on_phase in self.on_phase
]
list_off_events = [
self.filter_events(all_events, each_off_phase)
for each_off_phase in self.off_phase
]
# Loop over all ON- and OFF- phase intervals to compute the normalization factors a_on and a_off
a_on = np.sum([_[1] - _[0] for _ in self.on_phase])
a_off = np.sum([_[1] - _[0] for _ in self.off_phase])
on_events = EventList.stack(list_on_events)
off_events = EventList.stack(list_off_events)
return BackgroundEstimate(
on_region=self.on_region,
on_events=on_events,
off_region=None,
off_events=off_events,
a_on=a_on,
a_off=a_off,
method="Phase Bkg Estimator",
)
def run(self, dataset, observation=None):
"""Run the event sampler, applying IRF corrections.
Parameters
----------
dataset : `~gammapy.cube.MapDataset`
Map dataset
observation : `~gammapy.data.Observation`
In memory observation.
edisp : Bool
It allows to include or exclude the Edisp in the simulation.
Returns
-------
events : `~gammapy.data.EventList`
Event list.
"""
if len(dataset.models) > 1:
events_src = self.sample_sources(dataset)
if len(events_src.table) > 0:
if dataset.psf:
events_src = self.sample_psf(dataset.psf, events_src)
else:
events_src.table["RA"] = events_src.table["RA_TRUE"]
events_src.table["DEC"] = events_src.table["DEC_TRUE"]
if dataset.edisp:
events_src = self.sample_edisp(dataset.edisp, events_src)
else:
events_src.table["ENERGY"] = events_src.table[
"ENERGY_TRUE"]
if dataset.background_model:
events_bkg = self.sample_background(dataset)
events = EventList.stack([events_bkg, events_src])
else:
events = events_src
if len(dataset.models) == 1 and dataset.background_model is not None:
events_bkg = self.sample_background(dataset)
events = EventList.stack([events_bkg])
events = self.event_det_coords(observation, events)
events.table["EVENT_ID"] = np.arange(len(events.table))
events.table.meta = self.event_list_meta(dataset, observation)
geom = dataset._geom
selection = geom.contains(events.map_coord(geom))
return events.select_row_subset(selection)
def sample_sources(self, dataset):
"""Sample source model components.
Parameters
----------
dataset : `~gammapy.cube.MapDataset`
Map dataset.
Returns
-------
events : `~gammapy.data.EventList`
Event list
"""
events_all = []
for idx, evaluator in enumerate(dataset._evaluators):
evaluator = copy.deepcopy(evaluator)
evaluator.edisp = None
evaluator.psf = None
npred = evaluator.compute_npred()
temporal_model = ConstantTemporalModel()
table = self._sample_coord_time(npred, temporal_model, dataset.gti)
table["MC_ID"] = idx + 1
events_all.append(EventList(table))
return EventList.stack(events_all)
def sample_sources(self, dataset):
"""Sample source model components.
Parameters
----------
dataset : `~gammapy.cube.MapDataset`
Map dataset.
Returns
-------
events : `~gammapy.data.EventList`
Event list
"""
events_all = []
for idx, model in enumerate(dataset.models):
if isinstance(model, BackgroundModel):
continue
evaluator = dataset.evaluators.get(model.name)
evaluator = copy.deepcopy(evaluator)
evaluator.model.apply_irf["psf"] = False
evaluator.model.apply_irf["edisp"] = False
npred = evaluator.compute_npred()
temporal_model = ConstantTemporalModel()
table = self._sample_coord_time(npred, temporal_model, dataset.gti)
table["MC_ID"] = idx + 1
events_all.append(EventList(table))
return EventList.stack(events_all)
def _make_counts(dataset, observation, phases):
event_lists = []
for interval in phases:
events = observation.events.select_parameter(parameter="PHASE", band=interval)
event_lists.append(events)
events = EventList.stack(event_lists)
counts = RegionNDMap.from_geom(dataset.counts.geom)
counts.fill_events(events)
return counts
def _make_counts(dataset, observation, phases):
events = observation.events.select_region(dataset.counts.region)
event_lists = []
for interval in phases:
events = events.select_parameter(parameter="PHASE", band=interval)
event_lists.append(events)
events_off = EventList.stack(event_lists)
edges = dataset.counts.energy.edges
counts = CountsSpectrum(energy_hi=edges[1:], energy_lo=edges[:-1])
counts.fill_events(events_off)
return counts
def sample_sources(self, dataset):
"""Sample source model components.
Parameters
----------
dataset : `~gammapy.cube.MapDataset`
Map dataset.
Returns
-------
events : `~gammapy.data.EventList`
Event list
"""
events_all = []
for idx, model in enumerate(dataset.models):
if isinstance(model, BackgroundModel):
continue
evaluator = dataset.evaluators.get(model.name)
evaluator = copy.deepcopy(evaluator)
evaluator.model.apply_irf["psf"] = False
evaluator.model.apply_irf["edisp"] = False
npred = evaluator.compute_npred()
if hasattr(model, "temporal_model"):
if getattr(model, "temporal_model") is None:
temporal_model = ConstantTemporalModel()
else:
temporal_model = model.temporal_model
else:
temporal_model = ConstantTemporalModel()
table = self._sample_coord_time(npred, temporal_model, dataset.gti)
if len(table) > 0:
table["MC_ID"] = idx + 1
else:
mcid = table.Column(name="MC_ID", length=0, dtype=int)
table.add_column(mcid)
events_all.append(EventList(table))
return EventList.stack(events_all)
def run(self, dataset, observation=None):
"""Run the event sampler, applying IRF corrections.
Parameters
----------
dataset : `~gammapy.cube.MapDataset`
Map dataset
observation : `~gammapy.data.Observation`
In memory observation.
edisp : Bool
It allows to include or exclude the Edisp in the simulation.
Returns
-------
events : `~gammapy.data.EventList`
Event list.
"""
events_src = self.sample_sources(dataset)
if dataset.psf:
events_src = self.sample_psf(dataset.psf, events_src)
else:
events_src.table["RA"] = events_src.table["RA_TRUE"]
events_src.table["DEC"] = events_src.table["DEC_TRUE"]
if dataset.edisp:
events_src = self.sample_edisp(dataset.edisp, events_src)
else:
events_src.table["ENERGY"] = events_src.table["ENERGY_TRUE"]
if dataset.background_model:
events_bkg = self.sample_background(dataset)
events = EventList.stack([events_bkg, events_src])
else:
events = events_src
events.table["EVENT_ID"] = np.arange(len(events.table))
events.table.meta = self.event_list_meta(dataset, observation)
return events
def test_stack(self):
event_lists = [self.events] * 3
stacked_list = EventList.stack(event_lists)
assert len(stacked_list.table) == 49 * 3
