def test___all_fit_quantities__hyper_background_noise(self, interferometer_7):
hyper_background_noise = ag.hyper_data.HyperBackgroundNoise(noise_scale=1.0)
hyper_noise_map = hyper_background_noise.hyper_noise_map_from_complex_noise_map(
noise_map=interferometer_7.noise_map
)
g0 = ag.Galaxy(
redshift=0.5,
light_profile=ag.lp.EllSersic(intensity=1.0),
mass_profile=ag.mp.SphIsothermal(einstein_radius=1.0),
)
g1 = ag.Galaxy(redshift=1.0, light_profile=ag.lp.EllSersic(intensity=1.0))
plane = ag.Plane(redshift=0.75, galaxies=[g0, g1])
fit = ag.FitInterferometer(
interferometer=interferometer_7,
plane=plane,
hyper_background_noise=hyper_background_noise,
)
assert hyper_noise_map.slim == pytest.approx(fit.noise_map.slim)
fit = ag.FitInterferometer(
interferometer=interferometer_7,
plane=plane,
hyper_background_noise=hyper_background_noise,
use_hyper_scalings=False,
)
assert fit.noise_map == pytest.approx(interferometer_7.noise_map, 1.0e-4)
assert fit.noise_map != pytest.approx(hyper_noise_map.slim, 1.0e-4)
def test__fit_figure_of_merit(interferometer_7):
g0 = ag.Galaxy(
redshift=0.5,
light_profile=ag.lp.EllSersic(intensity=1.0),
mass_profile=ag.mp.SphIsothermal(einstein_radius=1.0),
)
g1 = ag.Galaxy(redshift=1.0, light_profile=ag.lp.EllSersic(intensity=1.0))
plane = ag.Plane(redshift=0.75, galaxies=[g0, g1])
fit = ag.FitInterferometer(dataset=interferometer_7, plane=plane)
assert (fit.noise_map.slim == np.full(fill_value=2.0 + 2.0j,
shape=(7, ))).all()
assert fit.log_likelihood == pytest.approx(-2398107.3849, 1e-4)
assert fit.figure_of_merit == pytest.approx(-2398107.3849, 1.0e-4)
pix = ag.pix.Rectangular(shape=(3, 3))
reg = ag.reg.Constant(coefficient=0.01)
g0 = ag.Galaxy(redshift=0.5, pixelization=pix, regularization=reg)
plane = ag.Plane(galaxies=[ag.Galaxy(redshift=0.5), g0])
fit = ag.FitInterferometer(dataset=interferometer_7, plane=plane)
assert (fit.noise_map.slim == np.full(fill_value=2.0 + 2.0j,
shape=(7, ))).all()
assert fit.log_evidence == pytest.approx(-66.90612, 1e-4)
assert fit.figure_of_merit == pytest.approx(-66.90612, 1.0e-4)
galaxy_light = ag.Galaxy(redshift=0.5,
light_profile=ag.lp.EllSersic(intensity=1.0))
pix = ag.pix.Rectangular(shape=(3, 3))
reg = ag.reg.Constant(coefficient=1.0)
galaxy_pix = ag.Galaxy(redshift=1.0, pixelization=pix, regularization=reg)
plane = ag.Plane(redshift=0.75, galaxies=[galaxy_light, galaxy_pix])
fit = ag.FitInterferometer(dataset=interferometer_7, plane=plane)
assert (fit.noise_map.slim == np.full(fill_value=2.0 + 2.0j,
shape=(7, ))).all()
assert fit.log_evidence == pytest.approx(-283424.48941, 1e-4)
assert fit.figure_of_merit == pytest.approx(-283424.48941, 1.0e-4)
def test___fit_galaxy_model_image_dict__images_and_inversion_mapped_reconstructed_image(
self, interferometer_7
):
pix = ag.pix.Rectangular(shape=(3, 3))
reg = ag.reg.Constant(coefficient=1.0)
g0 = ag.Galaxy(redshift=0.5)
g1 = ag.Galaxy(redshift=1.0, pixelization=pix, regularization=reg)
plane = ag.Plane(redshift=0.75, galaxies=[g0, g1])
fit = ag.FitInterferometer(interferometer=interferometer_7, plane=plane)
mapper = pix.mapper_from_grid_and_sparse_grid(
grid=interferometer_7.grid, sparse_grid=None
)
inversion = inversions.InversionInterferometerMatrix.from_data_mapper_and_regularization(
mapper=mapper,
regularization=reg,
visibilities=interferometer_7.visibilities,
noise_map=interferometer_7.noise_map,
transformer=interferometer_7.transformer,
)
assert (fit.galaxy_model_image_dict[g0].native == np.zeros((7, 7))).all()
assert fit.galaxy_model_image_dict[g1].slim == pytest.approx(
inversion.mapped_reconstructed_image.slim, 1.0e-4
)
def test___fit_galaxy_visibilities_dict__corresponds_to_galaxy_visibilities(
self, interferometer_7
):
g0 = ag.Galaxy(
redshift=0.5,
light_profile=ag.lp.EllSersic(intensity=1.0),
mass_profile=ag.mp.SphIsothermal(einstein_radius=1.0),
)
g1 = ag.Galaxy(redshift=1.0, light_profile=ag.lp.EllSersic(intensity=1.0))
plane = ag.Plane(redshift=0.75, galaxies=[g0, g1])
fit = ag.FitInterferometer(interferometer=interferometer_7, plane=plane)
g0_profile_visibilities = g0.profile_visibilities_from_grid_and_transformer(
grid=interferometer_7.grid, transformer=interferometer_7.transformer
)
g1_profile_visibilities = g1.profile_visibilities_from_grid_and_transformer(
grid=interferometer_7.grid, transformer=interferometer_7.transformer
)
assert fit.galaxy_model_visibilities_dict[g0].slim == pytest.approx(
g0_profile_visibilities, 1.0e-4
)
assert fit.galaxy_model_visibilities_dict[g1].slim == pytest.approx(
g1_profile_visibilities, 1.0e-4
)
assert fit.model_visibilities.slim == pytest.approx(
fit.galaxy_model_visibilities_dict[g0].slim
+ fit.galaxy_model_visibilities_dict[g1].slim,
1.0e-4,
)
def test__fit_figure_of_merit__includes_hyper_image_and_noise__matches_fit(
self, interferometer_7
):
hyper_background_noise = ag.hyper_data.HyperBackgroundNoise(noise_scale=1.0)
galaxy = ag.Galaxy(redshift=0.5, light=ag.lp.EllSersic(intensity=0.1))
model = af.Collection(
hyper_background_noise=hyper_background_noise,
galaxies=af.Collection(galaxy=galaxy),
)
analysis = ag.AnalysisInterferometer(dataset=interferometer_7)
instance = model.instance_from_unit_vector([])
fit_figure_of_merit = analysis.log_likelihood_function(instance=instance)
plane = analysis.plane_for_instance(instance=instance)
fit = ag.FitInterferometer(
interferometer=interferometer_7,
plane=plane,
hyper_background_noise=hyper_background_noise,
)
assert fit.log_likelihood == fit_figure_of_merit
def test___all_fit_quantities__hyper_background_noise(self, interferometer_7):
hyper_background_noise = ag.hyper_data.HyperBackgroundNoise(noise_scale=1.0)
hyper_noise_map = hyper_background_noise.hyper_noise_map_from_complex_noise_map(
noise_map=interferometer_7.noise_map
)
galaxy_light = ag.Galaxy(
redshift=0.5, light_profile=ag.lp.EllSersic(intensity=1.0)
)
pix = ag.pix.Rectangular(shape=(3, 3))
reg = ag.reg.Constant(coefficient=1.0)
galaxy_pix = ag.Galaxy(redshift=1.0, pixelization=pix, regularization=reg)
plane = ag.Plane(redshift=0.75, galaxies=[galaxy_light, galaxy_pix])
fit = ag.FitInterferometer(
interferometer=interferometer_7,
plane=plane,
hyper_background_noise=hyper_background_noise,
)
assert hyper_noise_map.slim == pytest.approx(fit.inversion.noise_map, 1.0e-4)
assert hyper_noise_map.slim == pytest.approx(fit.noise_map.slim)
def test___all_fit_quantities__no_hyper_methods(self, interferometer_7):
g0 = ag.Galaxy(
redshift=0.5,
light_profile=ag.lp.EllSersic(intensity=1.0),
mass_profile=ag.mp.SphIsothermal(einstein_radius=1.0),
)
g1 = ag.Galaxy(redshift=1.0, light_profile=ag.lp.EllSersic(intensity=1.0))
plane = ag.Plane(redshift=0.75, galaxies=[g0, g1])
fit = ag.FitInterferometer(interferometer=interferometer_7, plane=plane)
assert interferometer_7.noise_map == pytest.approx(fit.noise_map)
model_visibilities = plane.profile_visibilities_from_grid_and_transformer(
grid=interferometer_7.grid, transformer=interferometer_7.transformer
)
assert model_visibilities == pytest.approx(fit.model_visibilities, 1e-4)
residual_map = ag.util.fit.residual_map_from(
data=interferometer_7.visibilities, model_data=model_visibilities
)
assert residual_map == pytest.approx(fit.residual_map, 1e-4)
normalized_residual_map = ag.util.fit.normalized_residual_map_complex_from(
residual_map=residual_map, noise_map=interferometer_7.noise_map
)
assert normalized_residual_map == pytest.approx(
fit.normalized_residual_map, 1e-4
)
chi_squared_map = ag.util.fit.chi_squared_map_complex_from(
residual_map=residual_map, noise_map=interferometer_7.noise_map
)
assert chi_squared_map == pytest.approx(fit.chi_squared_map, 1e-4)
chi_squared = ag.util.fit.chi_squared_complex_from(
chi_squared_map=fit.chi_squared_map
)
noise_normalization = ag.util.fit.noise_normalization_complex_from(
noise_map=interferometer_7.noise_map
)
log_likelihood = ag.util.fit.log_likelihood_from(
chi_squared=chi_squared, noise_normalization=noise_normalization
)
assert log_likelihood == pytest.approx(fit.log_likelihood, 1e-4)
assert log_likelihood == fit.figure_of_merit
def test__model_visibilities(interferometer_7):
g0 = ag.Galaxy(redshift=0.5,
light_profile=MockLightProfile(image_2d=np.ones(2)))
plane = ag.Plane(galaxies=[g0])
fit = ag.FitInterferometer(dataset=interferometer_7, plane=plane)
assert fit.model_visibilities.slim[0] == pytest.approx(
np.array([1.2933 + 0.2829j]), 1.0e-4)
assert fit.log_likelihood == pytest.approx(-27.06284, 1.0e-4)
def test___fit_galaxy_model_visibilities_dict__has_image_and_inversion_mapped_reconstructed_image(
self, interferometer_7
):
g0 = ag.Galaxy(redshift=0.5, light_profile=ag.lp.EllSersic(intensity=1.0))
g1 = ag.Galaxy(redshift=0.5, light_profile=ag.lp.EllSersic(intensity=2.0))
pix = ag.pix.Rectangular(shape=(3, 3))
reg = ag.reg.Constant(coefficient=1.0)
galaxy_pix = ag.Galaxy(redshift=1.0, pixelization=pix, regularization=reg)
plane = ag.Plane(redshift=0.75, galaxies=[g0, g1, galaxy_pix])
fit = ag.FitInterferometer(interferometer=interferometer_7, plane=plane)
g0_visibilities = g0.profile_visibilities_from_grid_and_transformer(
grid=interferometer_7.grid, transformer=interferometer_7.transformer
)
g1_visibilities = g1.profile_visibilities_from_grid_and_transformer(
grid=interferometer_7.grid, transformer=interferometer_7.transformer
)
profile_visibilities = g0_visibilities + g1_visibilities
profile_subtracted_visibilities = (
interferometer_7.visibilities - profile_visibilities
)
mapper = pix.mapper_from_grid_and_sparse_grid(
grid=interferometer_7.grid,
settings=ag.SettingsPixelization(use_border=False),
)
inversion = inversions.InversionInterferometerMatrix.from_data_mapper_and_regularization(
visibilities=profile_subtracted_visibilities,
noise_map=interferometer_7.noise_map,
transformer=interferometer_7.transformer,
mapper=mapper,
regularization=reg,
)
g0_image = g0.image_2d_from_grid(grid=interferometer_7.grid)
g1_image = g1.image_2d_from_grid(grid=interferometer_7.grid)
assert fit.galaxy_model_image_dict[g0].slim == pytest.approx(
g0_image.slim, 1.0e-4
)
assert fit.galaxy_model_image_dict[g1].slim == pytest.approx(
g1_image.slim, 1.0e-4
)
assert fit.galaxy_model_image_dict[galaxy_pix].slim == pytest.approx(
inversion.mapped_reconstructed_image.slim, 1.0e-4
)
def test__noise_map__with_and_without_hyper_background(interferometer_7):
g0 = ag.Galaxy(redshift=0.5,
light_profile=MockLightProfile(image_2d=np.ones(2)))
plane = ag.Plane(galaxies=[g0])
fit = ag.FitInterferometer(dataset=interferometer_7, plane=plane)
assert (fit.noise_map.slim == np.full(fill_value=2.0 + 2.0j,
shape=(7, ))).all()
hyper_background_noise = ag.hyper_data.HyperBackgroundNoise(
noise_scale=1.0)
fit = ag.FitInterferometer(
dataset=interferometer_7,
plane=plane,
hyper_background_noise=hyper_background_noise,
)
assert (fit.noise_map.slim == np.full(fill_value=3.0 + 3.0j,
shape=(7, ))).all()
assert fit.log_likelihood == pytest.approx(-30.24288, 1.0e-4)
def test__1x2_image__1x2_visibilities__simple_fourier_transform(self):
# The image plane image generated by the galaxy is [1.0, 1.0]
# Thus the chi squared is 4.0**2.0 + 3.0**2.0 = 25.0
real_space_mask = ag.Mask2D.manual(
mask=[
[True, True, True, True],
[True, False, False, True],
[True, True, True, True],
],
pixel_scales=1.0,
)
interferometer = ag.Interferometer(
visibilities=ag.Visibilities.full(fill_value=5.0, shape_slim=(1,)),
noise_map=ag.Visibilities.ones(shape_slim=(1,)),
uv_wavelengths=np.array([[0.0, 0.0]]),
real_space_mask=real_space_mask,
settings=ag.SettingsInterferometer(
grid_class=ag.Grid2D, sub_size=1, transformer_class=ag.TransformerDFT
),
)
interferometer.visibilities[0] = 5.0 + 4.0j
# Setup as a ray trace instance, using a light profile for the galaxy
g0 = ag.Galaxy(redshift=0.5, light_profile=MockLightProfile(value=1.0, size=2))
plane = ag.Plane(galaxies=[g0])
fit = ag.FitInterferometer(interferometer=interferometer, plane=plane)
assert (fit.visibilities.slim == np.array([5.0 + 4.0j])).all()
assert (fit.noise_map.slim == np.array([1.0 + 1.0j])).all()
assert (fit.model_visibilities.slim == np.array([2.0 + 0.0j])).all()
assert (fit.residual_map.slim == np.array([3.0 + 4.0j])).all()
assert (fit.normalized_residual_map.slim == np.array([3.0 + 4.0j])).all()
assert (fit.chi_squared_map.slim == np.array([9.0 + 16.0j])).all()
assert fit.chi_squared == 25.0
assert fit.noise_normalization == pytest.approx(
2.0 * np.log(2 * np.pi * 1.0 ** 2.0), 1.0e-4
)
assert fit.log_likelihood == pytest.approx(
-0.5 * (25.0 + 2.0 * np.log(2 * np.pi * 1.0 ** 2.0)), 1.0e-4
)
def test__fit_figure_of_merit__matches_correct_fit_given_galaxy_profiles(
self, interferometer_7
):
galaxy = ag.Galaxy(redshift=0.5, light=ag.lp.EllSersic(intensity=0.1))
model = af.Collection(galaxies=af.Collection(galaxy=galaxy))
analysis = ag.AnalysisInterferometer(dataset=interferometer_7)
instance = model.instance_from_unit_vector([])
fit_figure_of_merit = analysis.log_likelihood_function(instance=instance)
plane = analysis.plane_for_instance(instance=instance)
fit = ag.FitInterferometer(interferometer=interferometer_7, plane=plane)
assert fit.log_likelihood == fit_figure_of_merit
def test___fit_figure_of_merit__linear_operator(interferometer_7_lop):
pix = ag.pix.Rectangular(shape=(3, 3))
reg = ag.reg.Constant(coefficient=0.01)
g0 = ag.Galaxy(redshift=0.5, pixelization=pix, regularization=reg)
plane = ag.Plane(galaxies=[ag.Galaxy(redshift=0.5), g0])
fit = ag.FitInterferometer(
dataset=interferometer_7_lop,
plane=plane,
settings_inversion=ag.SettingsInversion(use_linear_operators=True),
)
assert (fit.noise_map.slim == np.full(fill_value=2.0 + 2.0j,
shape=(7, ))).all()
assert fit.log_evidence == pytest.approx(-71.5177, 1e-4)
assert fit.figure_of_merit == pytest.approx(-71.5177, 1.0e-4)
def test___fit_galaxy_model_image_dict__corresponds_to_profile_galaxy_images(
self, interferometer_7
):
g0 = ag.Galaxy(
redshift=0.5,
light_profile=ag.lp.EllSersic(intensity=1.0),
mass_profile=ag.mp.SphIsothermal(einstein_radius=1.0),
)
g1 = ag.Galaxy(redshift=1.0, light_profile=ag.lp.EllSersic(intensity=1.0))
plane = ag.Plane(redshift=0.75, galaxies=[g0, g1])
fit = ag.FitInterferometer(interferometer=interferometer_7, plane=plane)
g0_image = g0.image_2d_from_grid(grid=interferometer_7.grid)
g1_image = g1.image_2d_from_grid(grid=interferometer_7.grid)
assert fit.galaxy_model_image_dict[g0].slim == pytest.approx(g0_image, 1.0e-4)
assert fit.galaxy_model_image_dict[g1].slim == pytest.approx(g1_image, 1.0e-4)
def test__hyper_background_changes_background_sky__reflected_in_likelihood(self,):
uv_wavelengths = np.array([[1.0, 0.0], [1.0, 1.0], [2.0, 2.0]])
real_space_mask = ag.Mask2D.manual(
mask=[
[True, True, True, True, True],
[True, False, False, False, True],
[True, True, True, True, True],
],
pixel_scales=1.0,
)
interferometer = ag.Interferometer(
visibilities=ag.Visibilities.full(fill_value=5.0, shape_slim=(3,)),
noise_map=ag.Visibilities.full(fill_value=2.0, shape_slim=(3,)),
uv_wavelengths=uv_wavelengths,
real_space_mask=real_space_mask,
settings=ag.SettingsInterferometer(grid_class=ag.Grid2D, sub_size=1),
)
# Setup as a ray trace instance, using a light profile for the galaxy
g0 = ag.Galaxy(redshift=0.5, light_profile=MockLightProfile(value=1.0, size=2))
plane = ag.Plane(galaxies=[g0])
hyper_background_noise = ag.hyper_data.HyperBackgroundNoise(noise_scale=1.0)
fit = ag.FitInterferometer(
interferometer=interferometer,
plane=plane,
hyper_background_noise=hyper_background_noise,
)
assert (
fit.visibilities.slim == np.array([5.0 + 5.0j, 5.0 + 5.0j, 5.0 + 5.0j])
).all()
assert (
fit.noise_map.slim == np.array([3.0 + 3.0j, 3.0 + 3.0j, 3.0 + 3.0j])
).all()
def fit_interferometer_from_agg_obj(agg_obj):
"""
Returns a generator of `FitInterferometer` objects from an input aggregator, which generates a list of the
`FitInterferometer` objects for every set of results loaded in the aggregator.
This is performed by mapping the `fit_interferometer_from_agg_obj` with the aggregator, which sets up each fit
using only generators ensuring that manipulating the fits of large sets of results is done in a memory efficient
way.
Parameters
----------
aggregator : af.Aggregator
A PyAutoFit aggregator object containing the results of PyAutoGalaxy model-fits."""
interferometer = interferometer_from_agg_obj(agg_obj=agg_obj)
plane = plane_from_agg_obj(agg_obj=agg_obj)
return ag.FitInterferometer(
interferometer=interferometer,
plane=plane,
settings_pixelization=agg_obj.settings_pixelization,
settings_inversion=agg_obj.settings_inversion,
)
def fit_interferometer_via_database_from(
fit: Fit,
real_space_mask: Optional[ag.Mask2D] = None,
settings_interferometer: Optional[ag.SettingsInterferometer] = None,
settings_pixelization: Optional[ag.SettingsPixelization] = None,
settings_inversion: Optional[ag.SettingsInversion] = None,
):
"""
Returns a generator of `FitInterferometer` objects from an input aggregator, which generates a list of the
`FitInterferometer` objects for every set of results loaded in the aggregator.
This is performed by mapping the `fit_interferometer_from_agg_obj` with the aggregator, which sets up each fit
using only generators ensuring that manipulating the fits of large sets of results is done in a memory efficient
way.
Parameters
----------
aggregator : af.Aggregator
A PyAutoFit aggregator object containing the results of PyAutoGalaxy model-fits.
"""
settings_pixelization = settings_pixelization or fit.value(
name="settings_pixelization"
)
settings_inversion = settings_inversion or fit.value(name="settings_inversion")
interferometer = interferometer_via_database_from(
fit=fit,
real_space_mask=real_space_mask,
settings_interferometer=settings_interferometer,
)
plane = plane_via_database_from(fit=fit)
return ag.FitInterferometer(
interferometer=interferometer,
plane=plane,
settings_pixelization=settings_pixelization,
settings_inversion=settings_inversion,
)
def test__fit_figure_of_merit__include_hyper_methods(interferometer_7):
hyper_background_noise = ag.hyper_data.HyperBackgroundNoise(
noise_scale=1.0)
g0 = ag.Galaxy(
redshift=0.5,
light_profile=ag.lp.EllSersic(intensity=1.0),
mass_profile=ag.mp.SphIsothermal(einstein_radius=1.0),
)
g1 = ag.Galaxy(redshift=1.0, light_profile=ag.lp.EllSersic(intensity=1.0))
plane = ag.Plane(redshift=0.75, galaxies=[g0, g1])
fit = ag.FitInterferometer(
dataset=interferometer_7,
plane=plane,
hyper_background_noise=hyper_background_noise,
)
assert (fit.noise_map.slim == np.full(fill_value=3.0 + 3.0j,
shape=(7, ))).all()
assert fit.log_likelihood == pytest.approx(-1065843.7193, 1e-4)
assert fit.figure_of_merit == pytest.approx(-1065843.7193, 1.0e-4)
fit = ag.FitInterferometer(
dataset=interferometer_7,
plane=plane,
hyper_background_noise=hyper_background_noise,
use_hyper_scalings=False,
)
assert fit.noise_map == pytest.approx(interferometer_7.noise_map, 1.0e-4)
pix = ag.pix.Rectangular(shape=(3, 3))
reg = ag.reg.Constant(coefficient=0.01)
g0 = ag.Galaxy(redshift=0.5, pixelization=pix, regularization=reg)
plane = ag.Plane(galaxies=[ag.Galaxy(redshift=0.5), g0])
fit = ag.FitInterferometer(
dataset=interferometer_7,
plane=plane,
hyper_background_noise=hyper_background_noise,
)
assert (fit.noise_map.slim == np.full(fill_value=3.0 + 3.0j,
shape=(7, ))).all()
assert fit.log_evidence == pytest.approx(-68.63380, 1e-4)
assert fit.figure_of_merit == pytest.approx(-68.63380, 1.0e-4)
galaxy_light = ag.Galaxy(redshift=0.5,
light_profile=ag.lp.EllSersic(intensity=1.0))
pix = ag.pix.Rectangular(shape=(3, 3))
reg = ag.reg.Constant(coefficient=1.0)
galaxy_pix = ag.Galaxy(redshift=1.0, pixelization=pix, regularization=reg)
plane = ag.Plane(redshift=0.75, galaxies=[galaxy_light, galaxy_pix])
fit = ag.FitInterferometer(
dataset=interferometer_7,
plane=plane,
hyper_background_noise=hyper_background_noise,
)
assert (fit.noise_map.slim == np.full(fill_value=3.0 + 3.0j,
shape=(7, ))).all()
assert fit.log_evidence == pytest.approx(-161108.8377, 1e-4)
assert fit.figure_of_merit == pytest.approx(-161108.8377, 1.0e-4)
def test___all_fit_quantities__uses_linear_operator_inversion(
self, interferometer_7_lop
):
# Ensures the inversion grid is used, as this would cause the test to fail.
interferometer_7_lop.grid[0, 0] = -100.0
pix = ag.pix.Rectangular(shape=(3, 3))
reg = ag.reg.Constant(coefficient=0.01)
g0 = ag.Galaxy(redshift=0.5, pixelization=pix, regularization=reg)
plane = ag.Plane(galaxies=[ag.Galaxy(redshift=0.5), g0])
fit = ag.FitInterferometer(
interferometer=interferometer_7_lop,
plane=plane,
settings_inversion=ag.SettingsInversion(use_linear_operators=True),
)
mapper = pix.mapper_from_grid_and_sparse_grid(
grid=interferometer_7_lop.grid_inversion, sparse_grid=None
)
inversion = inversions.InversionInterferometerLinearOperator.from_data_mapper_and_regularization(
mapper=mapper,
regularization=reg,
visibilities=interferometer_7_lop.visibilities,
noise_map=interferometer_7_lop.noise_map,
transformer=interferometer_7_lop.transformer,
settings=ag.SettingsInversion(use_linear_operators=True),
)
assert inversion.mapped_reconstructed_visibilities == pytest.approx(
fit.model_visibilities, 1.0e-4
)
residual_map = ag.util.fit.residual_map_from(
data=interferometer_7_lop.visibilities,
model_data=inversion.mapped_reconstructed_visibilities,
)
assert residual_map.slim == pytest.approx(fit.residual_map.slim, 1.0e-4)
normalized_residual_map = ag.util.fit.normalized_residual_map_complex_from(
residual_map=residual_map, noise_map=interferometer_7_lop.noise_map
)
assert normalized_residual_map.slim == pytest.approx(
fit.normalized_residual_map.slim, 1.0e-4
)
chi_squared_map = ag.util.fit.chi_squared_map_complex_from(
residual_map=residual_map, noise_map=interferometer_7_lop.noise_map
)
assert chi_squared_map.slim == pytest.approx(fit.chi_squared_map.slim, 1.0e-4)
chi_squared = ag.util.fit.chi_squared_complex_from(
chi_squared_map=chi_squared_map
)
noise_normalization = ag.util.fit.noise_normalization_complex_from(
noise_map=interferometer_7_lop.noise_map
)
log_likelihood = ag.util.fit.log_likelihood_from(
chi_squared=chi_squared, noise_normalization=noise_normalization
)
assert log_likelihood == pytest.approx(fit.log_likelihood, 1e-4)
log_likelihood_with_regularization = ag.util.fit.log_likelihood_with_regularization_from(
chi_squared=chi_squared,
regularization_term=inversion.regularization_term,
noise_normalization=noise_normalization,
)
assert log_likelihood_with_regularization == pytest.approx(
fit.log_likelihood_with_regularization, 1e-4
)
log_evidence = ag.util.fit.log_evidence_from(
chi_squared=chi_squared,
regularization_term=inversion.regularization_term,
log_curvature_regularization_term=inversion.log_det_curvature_reg_matrix_term,
log_regularization_term=inversion.log_det_regularization_matrix_term,
noise_normalization=noise_normalization,
)
assert log_evidence == fit.log_evidence
assert log_evidence == fit.figure_of_merit
mapped_reconstructed_image = ag.util.inversion.mapped_reconstructed_data_from(
mapping_matrix=fit.inversion.mapper.mapping_matrix,
reconstruction=fit.inversion.reconstruction,
)
assert (
fit.inversion.mapped_reconstructed_image.slim == mapped_reconstructed_image
).all()
def test___galaxy_model_visibilities_dict(interferometer_7):
g0 = ag.Galaxy(
redshift=0.5,
light_profile=ag.lp.EllSersic(intensity=1.0),
mass_profile=ag.mp.SphIsothermal(einstein_radius=1.0),
)
g1 = ag.Galaxy(redshift=1.0, light_profile=ag.lp.EllSersic(intensity=1.0))
plane = ag.Plane(redshift=0.75, galaxies=[g0, g1])
fit = ag.FitInterferometer(dataset=interferometer_7, plane=plane)
g0_visibilities = g0.visibilities_via_transformer_from(
grid=interferometer_7.grid, transformer=interferometer_7.transformer)
g1_visibilities = g1.visibilities_via_transformer_from(
grid=interferometer_7.grid, transformer=interferometer_7.transformer)
assert fit.galaxy_model_visibilities_dict[g0].slim == pytest.approx(
g0_visibilities, 1.0e-4)
assert fit.galaxy_model_visibilities_dict[g1].slim == pytest.approx(
g1_visibilities, 1.0e-4)
assert fit.model_visibilities.slim == pytest.approx(
fit.galaxy_model_visibilities_dict[g0].slim +
fit.galaxy_model_visibilities_dict[g1].slim,
1.0e-4,
)
pix = ag.pix.Rectangular(shape=(3, 3))
reg = ag.reg.Constant(coefficient=1.0)
g0 = ag.Galaxy(redshift=0.5)
g1 = ag.Galaxy(redshift=1.0, pixelization=pix, regularization=reg)
plane = ag.Plane(redshift=0.75, galaxies=[g0, g1])
fit = ag.FitInterferometer(dataset=interferometer_7, plane=plane)
mapper = pix.mapper_from(source_grid_slim=interferometer_7.grid,
source_pixelization_grid=None)
inversion = ag.Inversion(dataset=interferometer_7,
linear_obj_list=[mapper],
regularization_list=[reg])
assert (fit.galaxy_model_visibilities_dict[g0] == 0.0 + 0.0j * np.zeros(
(7, ))).all()
assert fit.galaxy_model_visibilities_dict[g1].slim == pytest.approx(
inversion.mapped_reconstructed_data.slim, 1.0e-4)
assert fit.model_visibilities.slim == pytest.approx(
fit.galaxy_model_visibilities_dict[g1].slim, 1.0e-4)
g0 = ag.Galaxy(redshift=0.5, light_profile=ag.lp.EllSersic(intensity=1.0))
g1 = ag.Galaxy(redshift=0.5, light_profile=ag.lp.EllSersic(intensity=2.0))
g2 = ag.Galaxy(redshift=0.5)
galaxy_pix = ag.Galaxy(redshift=1.0, pixelization=pix, regularization=reg)
plane = ag.Plane(redshift=0.75, galaxies=[g0, g1, g2, galaxy_pix])
fit = ag.FitInterferometer(dataset=interferometer_7, plane=plane)
g0_visibilities = g0.visibilities_via_transformer_from(
grid=interferometer_7.grid, transformer=interferometer_7.transformer)
g1_visibilities = g1.visibilities_via_transformer_from(
grid=interferometer_7.grid, transformer=interferometer_7.transformer)
profile_visibilities = g0_visibilities + g1_visibilities
profile_subtracted_visibilities = (interferometer_7.visibilities -
profile_visibilities)
mapper = pix.mapper_from(
source_grid_slim=interferometer_7.grid,
settings=ag.SettingsPixelization(use_border=False),
)
inversion = ag.InversionInterferometer(
visibilities=profile_subtracted_visibilities,
noise_map=interferometer_7.noise_map,
transformer=interferometer_7.transformer,
linear_obj_list=[mapper],
regularization_list=[reg],
)
assert (fit.galaxy_model_visibilities_dict[g2] == 0.0 + 0.0j * np.zeros(
(7, ))).all()
assert fit.galaxy_model_visibilities_dict[g0].slim == pytest.approx(
g0_visibilities.slim, 1.0e-4)
assert fit.galaxy_model_visibilities_dict[g1].slim == pytest.approx(
g1_visibilities.slim, 1.0e-4)
assert fit.galaxy_model_visibilities_dict[
galaxy_pix].slim == pytest.approx(
inversion.mapped_reconstructed_data.slim, 1.0e-4)
assert fit.model_visibilities.slim == pytest.approx(
fit.galaxy_model_visibilities_dict[g0].slim +
fit.galaxy_model_visibilities_dict[g1].slim +
inversion.mapped_reconstructed_data.slim,
1.0e-4,
)
def test___all_fit_quantities__no_hyper_methods(self, interferometer_7):
galaxy_light = ag.Galaxy(
redshift=0.5, light_profile=ag.lp.EllSersic(intensity=1.0)
)
pix = ag.pix.Rectangular(shape=(3, 3))
reg = ag.reg.Constant(coefficient=1.0)
galaxy_pix = ag.Galaxy(redshift=1.0, pixelization=pix, regularization=reg)
plane = ag.Plane(redshift=0.75, galaxies=[galaxy_light, galaxy_pix])
fit = ag.FitInterferometer(interferometer=interferometer_7, plane=plane)
profile_visibilities = plane.profile_visibilities_from_grid_and_transformer(
grid=interferometer_7.grid, transformer=interferometer_7.transformer
)
assert profile_visibilities.slim == pytest.approx(fit.profile_visibilities.slim)
profile_subtracted_visibilities = (
interferometer_7.visibilities - profile_visibilities
)
assert profile_subtracted_visibilities.slim == pytest.approx(
fit.profile_subtracted_visibilities.slim
)
mapper = pix.mapper_from_grid_and_sparse_grid(
grid=interferometer_7.grid,
settings=ag.SettingsPixelization(use_border=False),
)
inversion = inversions.InversionInterferometerMatrix.from_data_mapper_and_regularization(
visibilities=profile_subtracted_visibilities,
noise_map=interferometer_7.noise_map,
transformer=interferometer_7.transformer,
mapper=mapper,
regularization=reg,
)
model_visibilities = (
profile_visibilities + inversion.mapped_reconstructed_visibilities
)
assert model_visibilities.slim == pytest.approx(fit.model_visibilities.slim)
residual_map = ag.util.fit.residual_map_from(
data=interferometer_7.visibilities, model_data=model_visibilities
)
assert residual_map.slim == pytest.approx(fit.residual_map.slim)
normalized_residual_map = ag.util.fit.normalized_residual_map_complex_from(
residual_map=residual_map, noise_map=interferometer_7.noise_map
)
assert normalized_residual_map.slim == pytest.approx(
fit.normalized_residual_map.slim
)
chi_squared_map = ag.util.fit.chi_squared_map_complex_from(
residual_map=residual_map, noise_map=interferometer_7.noise_map
)
assert chi_squared_map.slim == pytest.approx(fit.chi_squared_map.slim)
chi_squared = ag.util.fit.chi_squared_complex_from(
chi_squared_map=chi_squared_map
)
noise_normalization = ag.util.fit.noise_normalization_complex_from(
noise_map=interferometer_7.noise_map
)
log_likelihood = ag.util.fit.log_likelihood_from(
chi_squared=chi_squared, noise_normalization=noise_normalization
)
assert log_likelihood == pytest.approx(fit.log_likelihood, 1e-4)
log_likelihood_with_regularization = ag.util.fit.log_likelihood_with_regularization_from(
chi_squared=chi_squared,
regularization_term=inversion.regularization_term,
noise_normalization=noise_normalization,
)
assert log_likelihood_with_regularization == pytest.approx(
fit.log_likelihood_with_regularization, 1e-4
)
log_evidence = ag.util.fit.log_evidence_from(
chi_squared=chi_squared,
regularization_term=inversion.regularization_term,
log_curvature_regularization_term=inversion.log_det_curvature_reg_matrix_term,
log_regularization_term=inversion.log_det_regularization_matrix_term,
noise_normalization=noise_normalization,
)
assert log_evidence == fit.log_evidence
assert log_evidence == fit.figure_of_merit
mapped_reconstructed_image = ag.util.inversion.mapped_reconstructed_data_from(
mapping_matrix=fit.inversion.mapper.mapping_matrix,
reconstruction=fit.inversion.reconstruction,
)
assert (
fit.inversion.mapped_reconstructed_image.slim == mapped_reconstructed_image
).all()
def make_fit_interferometer_x2_galaxy_inversion_7x7():
return ag.FitInterferometer(
interferometer=make_interferometer_7(),
plane=make_plane_x2_galaxy_inversion_7x7(),
)
def test__simulate_interferometer_data_and_fit__chi_squared_is_0__noise_normalization_correct(
):
grid = ag.Grid2D.uniform(shape_native=(51, 51),
pixel_scales=0.1,
sub_size=2)
lens_galaxy = ag.Galaxy(
redshift=0.5,
light=ag.lp.EllSersic(centre=(0.1, 0.1), intensity=0.1),
mass=ag.mp.EllIsothermal(centre=(0.1, 0.1), einstein_radius=1.0),
)
source_galaxy = ag.Galaxy(redshift=1.0,
light=ag.lp.EllExponential(centre=(0.1, 0.1),
intensity=0.5))
plane = ag.Plane(galaxies=[lens_galaxy, source_galaxy])
simulator = ag.SimulatorInterferometer(
uv_wavelengths=np.ones(shape=(7, 2)),
transformer_class=ag.TransformerDFT,
exposure_time=300.0,
noise_if_add_noise_false=1.0,
noise_sigma=None,
)
interferometer = simulator.from_plane_and_grid(plane=plane, grid=grid)
file_path = path.join(
"{}".format(path.dirname(path.realpath(__file__))),
"data_temp",
"simulate_and_fit",
)
try:
shutil.rmtree(file_path)
except FileNotFoundError:
pass
if path.exists(file_path) is False:
os.makedirs(file_path)
interferometer.output_to_fits(
visibilities_path=path.join(file_path, "visibilities.fits"),
noise_map_path=path.join(file_path, "noise_map.fits"),
uv_wavelengths_path=path.join(file_path, "uv_wavelengths.fits"),
)
real_space_mask = ag.Mask2D.unmasked(shape_native=(51, 51),
pixel_scales=0.1,
sub_size=2)
interferometer = ag.Interferometer.from_fits(
visibilities_path=path.join(file_path, "visibilities.fits"),
noise_map_path=path.join(file_path, "noise_map.fits"),
uv_wavelengths_path=path.join(file_path, "uv_wavelengths.fits"),
real_space_mask=real_space_mask,
settings=ag.SettingsInterferometer(grid_class=ag.Grid2D,
transformer_class=ag.TransformerDFT,
sub_size=2),
)
plane = ag.Plane(galaxies=[lens_galaxy, source_galaxy])
fit = ag.FitInterferometer(
interferometer=interferometer,
plane=plane,
settings_pixelization=ag.SettingsPixelization(use_border=False),
)
assert fit.chi_squared == pytest.approx(0.0)
pix = ag.pix.Rectangular(shape=(7, 7))
reg = ag.reg.Constant(coefficient=0.0001)
lens_galaxy = ag.Galaxy(
redshift=0.5,
light=ag.lp.EllSersic(centre=(0.1, 0.1), intensity=0.1),
mass=ag.mp.EllIsothermal(centre=(0.1, 0.1), einstein_radius=1.0),
)
source_galaxy = ag.Galaxy(redshift=1.0,
pixelization=pix,
regularization=reg)
plane = ag.Plane(galaxies=[lens_galaxy, source_galaxy])
fit = ag.FitInterferometer(
interferometer=interferometer,
plane=plane,
settings_pixelization=ag.SettingsPixelization(use_border=False),
)
assert abs(fit.chi_squared) < 1.0e-4
file_path = path.join("{}".format(path.dirname(path.realpath(__file__))),
"data_temp")
shutil.rmtree(file_path, ignore_errors=True)
def make_fit_interferometer_7x7():
return ag.FitInterferometer(interferometer=make_interferometer_7(),
plane=make_plane_7x7())
