# Licensed under a 3-clause BSD style license - see LICENSE.rst
"""FoV background estimation."""
import logging
from gammapy.maps import Map
from gammapy.modeling import Datasets, Fit
__all__ = ["FoVBackgroundMaker"]
log = logging.getLogger(__name__)
[docs]class FoVBackgroundMaker:
"""Normalize template background on the whole field-of-view.
The dataset background model can be simply scaled (method="scale") or fitted (method="fit")
on the dataset counts.
The normalization is performed outside the exclusion mask that is passed on init.
If a SkyModel is set on the input dataset and method is 'fit', its are frozen during
the fov normalization fit.
Parameters
----------
method : str in ['fit', 'scale']
the normalization method to be applied. Default 'scale'.
exclusion_mask : `~gammapy.maps.WcsNDMap`
Exclusion mask
"""
def __init__(self, method="scale", exclusion_mask=None):
if method in ["fit", "scale"]:
self.method = method
else:
raise ValueError(f"Incorrect method for FoVBackgroundMaker: {method}.")
self.exclusion_mask = exclusion_mask
[docs] def run(self, dataset):
"""Run FoV background maker.
Fit the background model norm
Parameters
----------
dataset : `~gammapy.cube.fit.MapDataset`
Input map dataset.
"""
mask_fit = dataset.mask_fit
dataset.mask_fit = self._reproject_exclusion_mask(dataset)
if self.method is "fit":
self._fit_bkg(dataset)
else:
self._scale_bkg(dataset)
dataset.mask_fit = mask_fit
return dataset
def _reproject_exclusion_mask(self, dataset):
"""Reproject the exclusion on the dataset geometry"""
mask_map = Map.from_geom(dataset.counts.geom)
if self.exclusion_mask is not None:
coords = dataset.counts.geom.get_coord()
vals = self.exclusion_mask.get_by_coord(coords)
mask_map.data += vals
return mask_map.data.astype("bool")
def _fit_bkg(self, dataset):
"""Fit the FoV background model on the dataset counts data"""
# freeze all model components not related to background model
datasets = Datasets([dataset])
parameters_frozen = []
for par in datasets.parameters:
parameters_frozen.append(par.frozen)
if par not in dataset.background_model.parameters:
par.frozen = True
fit = Fit(datasets)
fit_result = fit.run()
if fit_result.success is False:
log.info(
f"FoVBackgroundMaker failed. No fit convergence for {dataset.name}."
)
# Unfreeze parameters
for i, par in enumerate(datasets.parameters):
par.frozen = parameters_frozen[i]
def _scale_bkg(self, dataset):
"""Fit the FoV background model on the dataset counts data"""
mask = dataset.mask
count_tot = dataset.counts.data[mask].sum()
bkg_tot = dataset.background_model.map.data[mask].sum()
if count_tot <= 0.0:
log.info(
f"FoVBackgroundMaker failed. No counts found outside exclusion mask for {dataset.name}."
)
elif bkg_tot <= 0.0:
log.info(
f"FoVBackgroundMaker failed. No positive background found outside exclusion mask for {dataset.name}."
)
else:
scale = count_tot / bkg_tot
dataset.background_model.norm.value = scale