[Issues] Problems with Global Astrometry

Fedor I. Getman tig at na.astro.it
Thu Jul 19 18:08:13 CEST 2007

Hi John,

We updated the python source tree with your modifications and repeated 
GAS, but  unfortunately  it does not work yet.
In order to figure out what happens, we checked the intermediate results 
running GAS manually for the same dataset, but only for 1 ccd.
It worked apparently well up to the creation of ccd1.cat1 from 
GAS*.slist.out files.

Command "asctoldac -i asctoldac.txt -o dither.pairs -c 
tmp1184771271.14.asctoldac.conf -t PAIRS"
produced dither.pairs catalog, which can't be read back to ascii format 
by ldactoasc unless  the "-s" key is used. (Is this normal?)

Next command "astrom -i ccd1.cat1 -o ccd1.cat2 -p dither.pairs -c 
tmp1184771271.3.astrom.conf -r dither.residuals"
gives the completely wrong Ra and Dec positions of objects in 
dither.residuals catalog with following warnings:

WARNING(prepare_correl): Serious problem in covariance matrix: Diagonal 
element 1 is negativeWARNING(prepare_correl): Covariance matrix is non 
semi-definite positive!! ABS(503586.031250) >1!!! for 0,0
WARNING(prepare_correl): Covariance matrix is non semi-definite 
positive!! 1 1 1.000000 0.000000 0.000000

Finally command "make_distort -i ccd1.cat2 -o ccd1.cat3 -c 
tmp1184771271.52.make_distort.conf" produces catalog ccd1.cat3 with 
wrong coordinates of objects.

So in our case the astrometric solution for each dithering is still wrong.

We appreciate  a lot your help and any advice you may give us to solve 
the problem.

Anna, Mario and Fedor.

John P. McFarland wrote:
> Hi Anna and Mario,
> It appears that my modifications to LDAC's Python wrapper was the cause of 
> the errors.  The configuration file for asctoldac was being written 
> incorrectly without any immediate errors.  That problem is now fixed and all 
> appears well after running the example commands Mario provided below. 
> Please test things on your side to make sure everything is working correctly 
> and the results are within the appropriate limits.
> Cheers,
> -=John

More information about the Issues mailing list