[daip] IMAGR problems

[Eric Greisen]

Andy Biggs wrote:
> Hi Eric. I've tried IMAGR again and for the case of a single field it 
> works fine (I didn't try and clean, but the map and beam were made 
> okay). I've also tried a big multi-field clean (68 fields) and there 
> still seem to be problems. IMAGR makes the beams in two batches, 1-32 
> and 33-68. The beams for the first batch are fine. However, for the 
> second batch the beam forming goes wrong somehow. IMAGR reports 
> 'SOLUTION FOR RESTORING BEAM FAILED' and the beam does look very odd. I 
> stopped IMAGR after it showed me the first map and this too was 
> obviously wrong. I've attached the full set of messages.
>

The problem was caused by your having images of two sizes.  I have found 
the gridding error and several minor TV gotchas as well.  Tomorrow 
morning's MNJ should fix things.

Note one serious point.  Your BOXFILE calls for cleaning objects out of 
the very corners of images.  This is a VERY bad idea.  The corners are 
not computed well (correction factors reach 10^6 or more) and aliasing 
is still not suppressed there at all well.  Use SETFC to make the facet 
center list.

Thanks for your continued help

Eric Greisen