The most popular variant to the
HOGBOM algorithm is due to
Clark (1980). The iterative search for point sources involves minor and
major cycles.
- In minor cycles,
- an
HOGBOM search is performed with two
limitations: 1) Only the brightest pixels are considered in the above
step 1, and 2) the convolution of the found point sources (step 3 above)
is done with a spatially truncated dirty beam
6.
Both limitations fasten
the search but may lead to difficult convergence in cases where the
secondary side lobes are a large fraction (e.g. 40%) of the main side
lobe.
- In major cycles,
- the clean components found in the last minor cycle
are removed in a single step from the residual map in the Fourier plane.
The use of the Fourier transform enable to clean slightly more than the
inner quarter of the map.
CLARK is faster than
HOGBOM, but less stable.
