def to_image(self, spectrum=None, keepdims=True):
"""Reduce to a 2-D map after weighing
with the associated exposure and a spectrum
Parameters
----------
spectrum : `~gammapy.modeling.models.SpectralModel`, optional
Spectral model to compute the weights.
Default is power-law with spectral index of 2.
keepdims : bool, optional
If True, the energy axis is kept with one bin.
If False, the axis is removed
Returns
-------
psf_out : `PSFMap`
`PSFMap` with the energy axis summed over
"""
from gammapy.makers.utils import _map_spectrum_weight
if spectrum is None:
spectrum = PowerLawSpectralModel(index=2.0)
exp_weighed = _map_spectrum_weight(self.exposure_map, spectrum)
exposure = exp_weighed.sum_over_axes(axes=["energy_true"],
keepdims=keepdims)
psf_data = exp_weighed.data * self.psf_map.data / exposure.data
psf_map = Map.from_geom(geom=self.psf_map.geom,
data=psf_data,
unit="sr-1")
psf = psf_map.sum_over_axes(axes=["energy_true"], keepdims=keepdims)
return self.__class__(psf_map=psf, exposure_map=exposure)
def test_map_spectrum_weight():
axis = MapAxis.from_edges([0.1, 10, 1000], unit="TeV", name="energy_true")
expo_map = WcsNDMap.create(npix=10, binsz=1, axes=[axis], unit="m2 s")
expo_map.data += 1
spectrum = ConstantSpectralModel(const="42 cm-2 s-1 TeV-1")
weighted_expo = _map_spectrum_weight(expo_map, spectrum)
assert weighted_expo.data.shape == (2, 10, 10)
assert weighted_expo.unit == "m2 s"
assert_allclose(weighted_expo.data.sum(), 100)
def run(self, dataset):
"""
Run adaptive smoothing on input MapDataset.
The latter should have
Parameters
----------
dataset : `~gammapy.cube.MapDataset` or `~gammapy.cube.MapDatasetOnOff`
the input dataset (with one bin in energy at most)
Returns
-------
images : dict of `~gammapy.maps.WcsNDMap`
Smoothed images; keys are:
* 'counts'
* 'background'
* 'flux' (optional)
* 'scales'
* 'significance'.
"""
# Check dimensionality
if len(dataset.data_shape) == 3:
if dataset.data_shape[0] != 1:
raise ValueError(
"ASmoothMapEstimator.run() requires a dataset with 1 energy bin at most."
)
counts = dataset.counts.sum_over_axes(keepdims=False)
background = dataset.npred()
if isinstance(dataset, MapDatasetOnOff):
background += dataset.background
background = background.sum_over_axes(keepdims=False)
if dataset.exposure is not None:
exposure = _map_spectrum_weight(dataset.exposure, self.spectrum)
exposure = exposure.sum_over_axes(keepdims=False)
else:
exposure = None
return self.estimate_maps(counts, background, exposure)
