def test__transformer(
self,
visibilities_7,
visibilities_noise_map_7,
uv_wavelengths_7x2,
sub_mask_2d_7x7,
):
interferometer_7 = aa.Interferometer(
visibilities=visibilities_7,
noise_map=visibilities_noise_map_7,
uv_wavelengths=uv_wavelengths_7x2,
real_space_mask=sub_mask_2d_7x7,
settings=aa.SettingsInterferometer(
transformer_class=transformer.TransformerDFT
),
)
assert type(interferometer_7.transformer) == transformer.TransformerDFT
interferometer_7 = aa.Interferometer(
visibilities=visibilities_7,
noise_map=visibilities_noise_map_7,
uv_wavelengths=uv_wavelengths_7x2,
real_space_mask=sub_mask_2d_7x7,
settings=aa.SettingsInterferometer(
transformer_class=transformer.TransformerNUFFT
),
)
assert type(interferometer_7.transformer) == transformer.TransformerNUFFT
def test__visibilities_and_model_are_different__no_masking__check_values_are_correct(
self, ):
real_space_mask = aa.Mask2D.manual(mask=[[False, False],
[False, False]],
sub_size=1,
pixel_scales=(1.0, 1.0))
data = aa.Visibilities.manual_slim(
visibilities=[1.0 + 2.0j, 3.0 + 4.0j])
noise_map = aa.VisibilitiesNoiseMap.manual_slim(
visibilities=[2.0 + 2.0j, 2.0 + 2.0j])
interferometer = aa.Interferometer(
visibilities=data,
noise_map=noise_map,
uv_wavelengths=np.ones(shape=(2, 2)),
real_space_mask=real_space_mask,
)
model_visibilities = aa.Visibilities.manual_slim(
visibilities=[1.0 + 2.0j, 3.0 + 3.0j])
fit = aa.FitInterferometer(
interferometer=interferometer,
model_visibilities=model_visibilities,
use_mask_in_fit=False,
)
assert (fit.visibilities.slim == np.array([1.0 + 2.0j,
3.0 + 4.0j])).all()
assert (fit.noise_map.slim == np.array([2.0 + 2.0j, 2.0 + 2.0j])).all()
assert (fit.signal_to_noise_map.slim == np.array(
[0.5 + 1.0j, 1.5 + 2.0j])).all()
assert (fit.model_visibilities.slim == np.array(
[1.0 + 2.0j, 3.0 + 3.0j])).all()
assert (fit.residual_map.slim == np.array([0.0 + 0.0j,
0.0 + 1.0j])).all()
assert (fit.normalized_residual_map.slim == np.array(
[0.0 + 0.0j, 0.0 + 0.5j])).all()
assert (fit.chi_squared_map.slim == np.array([0.0 + 0.0j,
0.0 + 0.25j])).all()
assert fit.chi_squared == 0.25
assert fit.reduced_chi_squared == 0.25 / 2.0
assert fit.noise_normalization == pytest.approx(
4.0 * np.log(2 * np.pi * 2.0**2.0), 1.0e-4)
assert fit.log_likelihood == -0.5 * (fit.chi_squared +
fit.noise_normalization)
def test__interferometer(self):
visibilities_mask = np.full(fill_value=False, shape=(7, ))
real_space_mask = aa.Mask2D.unmasked(shape_native=(7, 7),
pixel_scales=0.1,
sub_size=1)
grid = aa.Grid2D.from_mask(mask=real_space_mask)
pix = aa.pix.VoronoiMagnification(shape=(7, 7))
sparse_grid = pix.sparse_grid_from_grid(grid=grid)
mapper = pix.mapper_from_grid_and_sparse_grid(
grid=grid,
sparse_grid=sparse_grid,
settings=aa.SettingsPixelization(use_border=False),
)
reg = aa.reg.Constant(coefficient=0.0)
visibilities = aa.Visibilities.manual_slim(visibilities=[
1.0 + 0.0j,
1.0 + 0.0j,
1.0 + 0.0j,
1.0 + 0.0j,
1.0 + 0.0j,
1.0 + 0.0j,
1.0 + 0.0j,
])
noise_map = aa.VisibilitiesNoiseMap.ones(shape_slim=(7, ))
uv_wavelengths = np.ones(shape=(7, 2))
interferometer = aa.Interferometer(
visibilities=visibilities,
noise_map=noise_map,
uv_wavelengths=uv_wavelengths,
real_space_mask=real_space_mask,
)
inversion = aa.Inversion(
dataset=interferometer,
mapper=mapper,
regularization=reg,
settings=aa.SettingsInversion(check_solution=False),
)
assert inversion.mapped_reconstructed_visibilities == pytest.approx(
1.0 + 0.0j * np.ones(shape=(7, )), 1.0e-4)
assert (np.imag(inversion.mapped_reconstructed_visibilities) <
0.0001).all()
assert (np.imag(inversion.mapped_reconstructed_visibilities) >
0.0).all()
def test__interferometer_linear_operator(self):
real_space_mask = aa.Mask2D.unmasked(shape_native=(7, 7),
pixel_scales=0.1,
sub_size=1)
grid = aa.Grid2D.from_mask(mask=real_space_mask)
pix = aa.pix.Rectangular(shape=(7, 7))
mapper = pix.mapper_from_grid_and_sparse_grid(
grid=grid,
sparse_grid=None,
settings=aa.SettingsPixelization(use_border=False),
)
reg = aa.reg.Constant(coefficient=0.0)
visibilities = aa.Visibilities.manual_slim(visibilities=[
1.0 + 0.0j,
1.0 + 0.0j,
1.0 + 0.0j,
1.0 + 0.0j,
1.0 + 0.0j,
1.0 + 0.0j,
1.0 + 0.0j,
])
noise_map = aa.VisibilitiesNoiseMap.ones(shape_slim=(7, ))
uv_wavelengths = np.ones(shape=(7, 2))
interferometer = aa.Interferometer(
visibilities=visibilities,
noise_map=noise_map,
uv_wavelengths=uv_wavelengths,
real_space_mask=real_space_mask,
settings=aa.SettingsInterferometer(
transformer_class=aa.TransformerNUFFT),
)
inversion = aa.Inversion(
dataset=interferometer,
mapper=mapper,
regularization=reg,
settings=aa.SettingsInversion(use_linear_operators=True,
check_solution=False),
)
assert inversion.mapped_reconstructed_visibilities == pytest.approx(
1.0 + 0.0j * np.ones(shape=(7, )), 1.0e-4)
assert (np.imag(inversion.mapped_reconstructed_visibilities) <
0.0001).all()
assert (np.imag(inversion.mapped_reconstructed_visibilities) >
0.0).all()
def test__visibilities_and_model_are_identical__inversion_included__changes_certain_properties(
self, ):
real_space_mask = aa.Mask2D.manual(mask=[[False, False],
[False, False]],
sub_size=1,
pixel_scales=(1.0, 1.0))
data = aa.Visibilities.manual_slim(
visibilities=[1.0 + 2.0j, 3.0 + 4.0j])
noise_map = aa.VisibilitiesNoiseMap.manual_slim(
visibilities=[2.0 + 2.0j, 2.0 + 2.0j])
interferometer = aa.Interferometer(
visibilities=data,
noise_map=noise_map,
uv_wavelengths=np.ones(shape=(2, 2)),
real_space_mask=real_space_mask,
)
model_visibilities = aa.Visibilities.manual_slim(
visibilities=[1.0 + 2.0j, 3.0 + 4.0j])
inversion = mock.MockFitInversion(
regularization_term=2.0,
log_det_curvature_reg_matrix_term=3.0,
log_det_regularization_matrix_term=4.0,
)
fit = aa.FitInterferometer(
interferometer=interferometer,
model_visibilities=model_visibilities,
inversion=inversion,
use_mask_in_fit=False,
)
assert fit.chi_squared == 0.0
assert fit.reduced_chi_squared == 0.0
assert fit.noise_normalization == pytest.approx(
4.0 * np.log(2 * np.pi * 2.0**2.0), 1.0e-4)
assert fit.log_likelihood == -0.5 * (fit.chi_squared +
fit.noise_normalization)
assert fit.log_likelihood_with_regularization == -0.5 * (
fit.chi_squared + 2.0 + fit.noise_normalization)
assert fit.log_evidence == -0.5 * (fit.chi_squared + 2.0 + 3.0 - 4.0 +
fit.noise_normalization)
assert fit.figure_of_merit == fit.log_evidence
def test__new_interferometer_with_with_modified_visibilities(
self, sub_mask_2d_7x7, uv_wavelengths_7x2
):
interferometer = aa.Interferometer(
visibilities=np.array([[1, 1]]),
noise_map=1,
uv_wavelengths=uv_wavelengths_7x2,
real_space_mask=sub_mask_2d_7x7,
)
interferometer = interferometer.modified_visibilities_from_visibilities(
visibilities=np.array([2 + 2j])
)
assert (interferometer.visibilities == np.array([[2 + 2j]])).all()
assert interferometer.noise_map == 1
assert (interferometer.uv_wavelengths == uv_wavelengths_7x2).all()
def test__different_interferometer_without_mock_objects__customize_constructor_inputs(
self, sub_mask_2d_7x7
):
interferometer = aa.Interferometer(
visibilities=aa.Visibilities.ones(shape_slim=(19,)),
noise_map=2.0 * aa.Visibilities.ones(shape_slim=(19,)),
uv_wavelengths=3.0 * np.ones((19, 2)),
real_space_mask=sub_mask_2d_7x7,
)
real_space_mask = aa.Mask2D.unmasked(
shape_native=(19, 19), pixel_scales=1.0, invert=True, sub_size=8
)
real_space_mask[9, 9] = False
assert (interferometer.visibilities == 1.0 + 1.0j * np.ones((19,))).all()
assert (interferometer.noise_map == 2.0 + 2.0j * np.ones((19,))).all()
assert (interferometer.uv_wavelengths == 3.0 * np.ones((19, 2))).all()
def test__dirty_properties(
self,
visibilities_7,
visibilities_noise_map_7,
uv_wavelengths_7x2,
sub_mask_2d_7x7,
):
interferometer = aa.Interferometer(
visibilities=visibilities_7,
noise_map=visibilities_noise_map_7,
uv_wavelengths=uv_wavelengths_7x2,
real_space_mask=sub_mask_2d_7x7,
)
assert interferometer.dirty_image.shape_native == (7, 7)
assert (
interferometer.transformer.image_from_visibilities(
visibilities=interferometer.visibilities
)
).all()
assert interferometer.dirty_noise_map.shape_native == (7, 7)
assert (
interferometer.transformer.image_from_visibilities(
visibilities=interferometer.noise_map
)
).all()
assert interferometer.dirty_signal_to_noise_map.shape_native == (7, 7)
assert (
interferometer.transformer.image_from_visibilities(
visibilities=interferometer.signal_to_noise_map
)
).all()
assert interferometer.dirty_inverse_noise_map.shape_native == (7, 7)
assert (
interferometer.transformer.image_from_visibilities(
visibilities=interferometer.inverse_noise_map
)
).all()
def _interferometer_from(
fit: af.Fit,
real_space_mask: Optional[aa.Mask2D] = None,
settings_interferometer: Optional[aa.SettingsInterferometer] = None,
):
"""
Returns a `Interferometer` object from an aggregator's `SearchOutput` class, which we call an 'agg_obj' to
describe that it acts as the aggregator object for one result in the `Aggregator`. This uses the aggregator's
generator outputs such that the function can use the `Aggregator`'s map function to to create a
`Interferometer` generator.
The `Interferometer` is created following the same method as the PyAutoGalaxy `Search` classes, including
using the `SettingsInterferometer` instance output by the Search to load inputs of the `Interferometer`
(e.g. psf_shape_2d).
Parameters
----------
fit : ImaginSearchOutput
A PyAutoFit aggregator's SearchOutput object containing the generators of the results of PyAutoGalaxy
model-fits.
"""
data = fit.value(name="data")
noise_map = fit.value(name="noise_map")
uv_wavelengths = fit.value(name="uv_wavelengths")
real_space_mask = real_space_mask or fit.value(name="real_space_mask")
settings_interferometer = settings_interferometer or fit.value(
name="settings_dataset"
)
interferometer = aa.Interferometer(
visibilities=data,
noise_map=noise_map,
uv_wavelengths=uv_wavelengths,
real_space_mask=real_space_mask,
)
interferometer = interferometer.apply_settings(settings=settings_interferometer)
return interferometer
def test__signal_to_noise_limit_below_max_signal_to_noise__signal_to_noise_map_capped_to_limit(
self, sub_mask_2d_7x7, uv_wavelengths_7x2
):
interferometer = aa.Interferometer(
real_space_mask=sub_mask_2d_7x7,
visibilities=aa.Visibilities(visibilities=np.array([1 + 1j, 1 + 1j])),
noise_map=aa.VisibilitiesNoiseMap(
visibilities=np.array([1 + 0.25j, 1 + 0.25j])
),
uv_wavelengths=uv_wavelengths_7x2,
)
interferometer_capped = interferometer.signal_to_noise_limited_from(
signal_to_noise_limit=2.0
)
assert (
interferometer_capped.visibilities == np.array([1.0 + 1.0j, 1.0 + 1.0j])
).all()
assert (
interferometer_capped.noise_map == np.array([1.0 + 0.5j, 1.0 + 0.5j])
).all()
assert (
interferometer_capped.signal_to_noise_map == np.array([1.0 + 2.0j])
).all()
interferometer_capped = interferometer.signal_to_noise_limited_from(
signal_to_noise_limit=0.25
)
assert (
interferometer_capped.visibilities == np.array([1.0 + 1.0j, 1.0 + 1.0j])
).all()
assert (
interferometer_capped.noise_map == np.array([4.0 + 4.0j, 4.0 + 4.0j])
).all()
assert (
interferometer_capped.signal_to_noise_map == np.array([0.25 + 0.25j])
).all()