Continuum Imaging Pipeline python class

class src.imaging_prototype.ContinuumImagingPipeline(imaging_context, use_dask=False)

Prototype Continuum Imaging Pipeline. Doesn’t use calibration.

_init_model_images_list(n_pixel, cell_size=None)

Initialize model images from input visibility data

Note: According to RASCIL Continuum Imaging Pipeline,

each model image needs to have a single frequency channel. This will make sure that when we run invert with ng, the resulting Image list will have a channel each and, hence deconvolution won’t break (which expects 1 chan / Image).

Parameters

• cell_size – image cell size [rad]; if None, it is calculated

• n_pixel – number of pixels on a side of the image

_init_psf_list()

Create Point Spread Functions (PSF)

_init_skymodel_list()

Initialize SkyModel for each model image (Note: we may need to allow it as input too, in the future)

_load_bvis_from_ms(input_ms, channels_in_data, nchan_per_vis)

Load MeasurementSet data into Visibility objects. (rascil.data_models.memory_data_models.Visibility)

Note: - Polarization of data is always converted to Stokes I after loading. - Based on rascil.apps.rascil_imager.get_vis_list and

rascil.workflows.rsexecute.visibility.visibility_rsexecute.create_visibility_from_ms_rsexecute

• these RASCIL functions take multiple arguments, most of which we hardcode here. However, in the future, we may need to allow for users to specify these.

Parameters

• input_ms – path to input MeasurementSet

• channels_in_data – how many frequency channels does the data set contain

• nchan_per_vis – how many frequency channels to load into a single Visibility

deconvolve(fit_skymodel, component_threshold, clean_threshold)

Run deconvolution.

Parameters

• fit_skymodel – True: fit the skymodel and extract sky components False: run CLEAN-based deconvolution

• component_threshold – Sources with absolute flux > this level (Jy) are fitted

• clean_threshold – Clean stopping threshold (Jy/beam)

Returns

updates self.skymodel_list in place

invert()

Run the invert step.

Returns

updates self.dirty_image_list in place

predict(processing_func_source)

Run the predict step, including subtracting the predicted model data from the input data.

Parameters

processing_func_source –

which type of processing functions to use; Options: ‘rascil’ - use RASCIL

’proc_func’ - Use the Processing Function Library

Returns

updates self.bvis_list in place

restore()

Restore images.

Uses RASCIL’s restore_skymodel_pipeline_rsexecute_workflow function, which implicitly uses rsexecute to run with Dask.