Imaging algorithms in radio interferometry
The paper reviews progress in imaging in radio interferometry for the period 1993-1996. Unlike an optical telescope, the basic measurements of a radio interferometer (correlations between antennas) are indirectly related to a sky brightness image. In a real sense, algorithms and computers are the lenses of a radio interferometer. In the last 20 years, whereas interferometer hardware advances have resulted in improvements of a factor of a few, algorithm and computer advances have resulted in orders of magnitude improvement in image quality. Developing these algorithms has been a fruitful and comparatively inexpensive method of improving the performance of existing telescopes, and has made some newer telescopes possible. In this paper, we review recent developments in the algorithms used in the imaging part of the reduction process. What constitutes an `imaging algorithm'? Whereas once there was a steady `forward' progression in the reduction process of editing, calibrating, transforming and, finally, deconvolving, this is no longer true. The introduction of techniques such as self-calibration, and algorithms that go directly from visibilities to final images, have made the dividing lines less clear. Although we briefly consider self-calibration, for the purposes of this paper calibration issues are generally excluded. Most attention will be directed to the steps which form final images from the calibrated visibilities.