Alongside CCD imaging data, PyAutoLens supports the modeling of interferometer data from submillimeter and radio observatories.
The dataset is fitted directly in the uv-plane, circumventing issues that arise when fitting a ‘dirty image’ such as correlated noise.
Real Space Mask#
To begin, we define a real-space mask. Although interferometer lens modeling is performed in the uv-plane and therefore Fourier space, we still need to define the grid of coordinates in real-space from which the lensed source’s images are computed. It is this image that is mapped to Fourier space to compare to the uv-plane data.
real_space_mask_2d = ag.Mask2D.circular( shape_native=(400, 400), pixel_scales=0.025, radius=3.0 )
We next load an
Interferometer dataset from fits files, which follows the same API that we have seen
dataset_path = "/path/to/dataset/folder" interferometer = al.Interferometer.from_fits( visibilities_path=path.join(dataset_path, "visibilities.fits"), noise_map_path=path.join(dataset_path, "noise_map.fits"), uv_wavelengths_path=path.join(dataset_path, "uv_wavelengths.fits"), real_space_mask=real_space_mask_2d ) interferometer_plotter = aplt.InterferometerPlotter(interferometer=interferometer) interferometer_plotter.figures_2d(visibilities=True, uv_wavelengths=True)
Here is what the interferometer visibilities and uv wavelength (which represent the interferometer’s baselines):
The data used in this overview contains only ~300 visibilities and is representative of a low resolution Square-Mile Array (SMA) dataset.
We discuss below how PyAutoLens can scale up to large visibilities datasets from an instrument like ALMA.
This can also plot the dataset in real-space, using the fast Fourier transforms described below.
interferometer_plotter = aplt.InterferometerPlotter(interferometer=interferometer) interferometer_plotter.figures_2d(dirty_image=True, dirty_signal_to_noise_map=True)
Here is what the image and signal-to-noise map look like in real space:
To perform uv-plane modeling, PyAutoLens Fourier transforms the lensed image (computed via a
real-space to the uv-plane.
This operation uses a
Transformer object, of which there are multiple available
in PyAutoLens. This includes a direct Fourier transform which performs the exact Fourier transform without approximation.
transformer_class = al.TransformerDFT
However, the direct Fourier transform is inefficient. For ~10 million visibilities, it requires thousands of seconds to perform a single transform. This approach is therefore unfeasible for high quality ALMA and radio datasets.
For this reason, PyAutoLens supports the non-uniform fast fourier transform algorithm PyNUFFT (https://github.com/jyhmiinlin/pynufft), which is significantly faster, being able too perform a Fourier transform of ~10 million in less than a second!
transformer_class = al.TransformerNUFFT
To perform a fit, we follow the same process we did for imaging. We do not need to mask an interferometer dataset, but we will apply the settings above:
interferometer = interferometer.apply_settings( settings=al.SettingsInterferometer(transformer_class=transformer_class) )
The interferometer can now be passed to a
FitInterferometer object to fit it to a data-set:
fit = al.FitInterferometer( interferometer=interferometer, tracer=tracer ) fit_interferometer_plotter = aplt.FitInterferometerPlotter(fit=fit) fit_interferometer_plotter.subplot_fit_interferometer() fit_interferometer_plotter.subplot_fit_real_space()
Here is what the image of the tracer looks like before it is Fourier transformed to the uv-plane:
And here is what the Fourier transformed model visibilities look like:
Here is what the fit of the galaxy looks like in real space (which is computed via a FFT from the uv-plane):
Interferometer data can also be modeled using pixelized source’s, which again perform the source reconstruction by directly fitting the visibilities in the uv-plane.
The source reconstruction is visualized in real space:
Computing this source reconstruction would be extremely inefficient if PyAutoLens used a traditional approach to linear algebra which explicitly stored in memory the values required to solve for the source fluxes. In fact, for an interferometer dataset of ~10 million visibilities this would require hundreds of GB of memory!
PyAutoLens uses the library PyLops (https://pylops.readthedocs.io/en/latest/) to represent this calculation as a sequence of memory-light linear operators.
The combination of PyNUFFT and PyLops makes the analysis of ~10 million visibilities from observatories such as ALMA and JVLA feasible in PyAutoLens.
It is straight forward to fit a lens model to an interferometer dataset, using the same API that we saw for imaging data in the modeling overview example.
Whereas we previously used an
AnalysisImaging object, we instead use an
AnalysisInterferometer object which fits
the lens model in the correct way for an interferometer dataset. This includes mapping the lens model from real-space
to the uv-plane via the Fourier transform discussed above:
lens_galaxy_model = af.Model(al.Galaxy, redshift=0.5, mass=al.mp.EllIsothermal) source_galaxy_model = af.Model(al.Galaxy, redshift=1.0, disk=al.lp.EllExponential) model = af.Collection(lens=lens_galaxy_model, source=source_galaxy_model) search = af.DynestyStatic(name="overview_interferometer") analysis = al.AnalysisInterferometer(dataset=interferometer) result = search.fit(model=model, analysis=analysis)
Simulated interferometer datasets can be generated using the
SimulatorInterferometer object, which includes adding
Gaussian noise to the visibilities:
real_space_grid_2d = ag.Grid2D.uniform( shape_native=real_space_mask.shape_native, pixel_scales=real_space_mask.pixel_scales ) simulator = al.SimulatorInterferometer( uv_wavelengths=uv_wavelengths, exposure_time=300.0, background_sky_level=1.0, noise_sigma=0.01, ) interferometer = simulator.via_tracer_from(tracer=tracer, grid=real_space_grid)
interferometer package of the
autolens_workspace contains numerous example scripts for performing
interferometer modeling and simulating strong lens interferometer datasets.