[daip] VTESS

[Eric Greisen]

Katherine Blundell wrote:
> Dear Eric,
> 
> Do hope that this finds you well.
> 
> I'm trying to use VTESS to deconvolve some optical data - and I've given 
> it a good estimate of the PSF for the beam file - but on the first pass, 
> I hit a curious error message which appears in the AIPS window that I 
> normally type in, saying:
> 
>  >go
> AIPS 1: Resumes
>  >VTESS1(3111) malloc: *** mmap(size=3206549504) failed (error code=12)
> *** error: can't allocate region
> *** set a breakpoint in malloc_error_break to debug


> AIPS 1: BLC         201         273        Bottom left corner of VTESS
> AIPS 1:               1           1           1           1
> AIPS 1:               1
> AIPS 1: TRC         971         945        Top right corner of VTESS
> AIPS 1:               1           1           1           1
> AIPS 1:               1

> The header of the GETN file is:

> AIPS 1:           1200   0.0000000E+00     600.00  1.0000000E+00    0.00
> AIPS 1:           1200   0.0000000E+00     600.00  1.0000000E+00    0.00

> 
> BLOSSO> VTESS1: Task VTESS  (release of 31DEC07) begins
> BLOSSO> VTESS1: Creating  10 scratch files
>
> BLOSSO> VTESS1: VTESS BMAJ =   3.878E+04 / Beam major axis (asec)
> BLOSSO> VTESS1: VTESS BMIN =   3.734E+04 / Beam minor axis (asec)
> BLOSSO> VTESS1: VTESS BPA  =  -6.963E+01 / Beam p.a. (degrees)
> BLOSSO> VTESS1: QINIT: did a FREE of    2821 Kwords, OFF          5441713
> BLOSSO> VTESS1: ZMEMRY: MEMORY ALLOCATION FAILS AT VTESS
> BLOSSO> VTESS1: QINIT: ERROR  3 FROM ZMEMRY DOING GET WORDS********
> BLOSSO> VTESS1: QINIT: did a GET  of       0 Kwords, OFF                1
> BLOSSO> VTESS1: ***********************************************
> BLOSSO> VTESS1: PASS1: USING MORE THAN ONE PASS!
> BLOSSO> VTESS1: ***********************************************
> BLOSSO> VTESS1: APXPOS: ERROR IN INPUT, NROW =     0, NCOL =   600
> BLOSSO> VTESS1: CORR: ERROR   8 IN DSKFFT MAP TO U,V
> BLOSSO> VTESS1: CALRES
>
> So the problem seems to be that VTESS doesn't seem to want to read the 
> bmaj and bmin from the headers.   This being optical data, the fits 
> files didn't have a beam defined, so I did this with puthead.    Is this 
> a silly way to proceed?   I'd be most grateful for any advice you can 
> offer...   deconvolving optical data is a huge faff!

The first thing it tries to do is allocate more than 3 Gbytes of RAM 
which it cannot do in a 32-bit addressing machine.  Why it continues at 
that point I cannot say.  VTESS can only clean 1/4 of the image - 
preferably the center 1/4 of the area.  Your BLC and TRC call for more 
than that.  Try a smaller area and see what happens.

Eric