NNLSImageSkyModel.h

Classes

NNLSImageSkyModel -- Briggs' Non-Negative Least Squares deconvolution algorithm (full description)

class NNLSImageSkyModel : public CleanImageSkyModel

Interface

Public Members
virtual Bool solve (SkyEquation& me)
Private Members
Bool maskedZeroI()

Description

Prerequisite

Etymology

NNLSImageSkyModel implements the NNLS deconvolution algorithm. It is derived from ImageSkyModel.

Synopsis

NNLS is implemented using the NNLSMatrixSolver class. Since the brightness must be non-negative, only the Stokes I is solved for. Any other polarizations are not changed.

Two masks are needed: the fluxmask determining the free pixels in the output image, and the datamask determining the constraint pixels in the dirty image.

In the solution, a matrix is required having elements connecting all free pixels with all constraint pixels. This is constructed from the point spread function. The total size of this matrix dominates the memory usage.

Example

See the example for SkyModel.

Motivation

To Do

Member Description

virtual Bool solve (SkyEquation& me)

Solve for this SkyModel

Bool maskedZeroI()