[daip] Help running DDT/Y2K
Eric Greisen
egreisen at nrao.edu
Tue Sep 6 15:12:31 EDT 2005
Bryan Follis writes:
> Besides the fact that this is printed:
>
> PRTAC1: Appears to have ended successfully
This only says that the task PRTAC apparently ran with no
internal error (to print the accounting info).
>
> Is there any way to indicate a successful run with accurate results?
> Moreover, is there supposed to be some sort of visual rendering that
> goes along with this? In my xterm environment there is a black window
> that started out as an "AIPS 24 Bit" logo, but it never gets anything
> in it.
That is the aips TV display and it can be used to look at the
results but is not part of the test. We want the test to measure
times and so avoid using socket-connected "devices" - I usually kill
the message server so that is not used during timing tests.
In the OUTPRINT file you will find a section that looks like
2 5 31-AUG-2005 13:27:47 AIPS 'IMAGE' 12.4692392 15.5282879
2 5 31-AUG-2005 13:27:47 AIPS 'IMBEAM' 13.9143038 16.5563126
2 5 31-AUG-2005 13:27:47 AIPS 'APCLN' 8.7059097 16.1575108
2 5 31-AUG-2005 13:27:47 AIPS 'APRES' 17.8541603 24.590538
2 5 31-AUG-2005 13:27:47 AIPS 'IMCLN' 8.2134628 16.1144638
2 5 31-AUG-2005 13:27:47 AIPS 'VTESS' 15.6747179 22.9842434
2 5 31-AUG-2005 13:27:47 AIPS 'VTESSC' 16.9863358 20.6356678
2 5 31-AUG-2005 13:27:47 AIPS 'IMAGE' -0.0001221 0.00006104 0.00009441
2 5 31-AUG-2005 13:27:47 AIPS 'IMBEAM' 0 0 0.000000477
2 5 31-AUG-2005 13:27:47 AIPS 'APCLN' -0.00006104 -0.0001831 -0.0174179
2 5 31-AUG-2005 13:27:47 AIPS 'APRES' 0 0 0
2 5 31-AUG-2005 13:27:47 AIPS 'IMCLN' -0.0136108 0.0014648 -0.0187984
2 5 31-AUG-2005 13:27:47 AIPS 'VTESS' -0.00006104 0.00006104 0.000000775
2 5 31-AUG-2005 13:27:47 AIPS 'VTESSC' 0 0 -0.00001383
for the Y2K test. The two columns at right in the first half are the
"number of bits" of agreement between the locally computed image and
our standard image. The lower number is the log base 2 of the peak of
the image divided by the peak of the difference image, while the
higher number of the peak of the image divided by the rms of the
difference image. Thus bigger is better. The 8.7 and 8.2 above are
acceptable, the rest are very good. It is normal for the Cleans to be
lower. The test also runs UVDIF and one should get nothing printed by
it other than perhaps a sort order issue or two.
Will print all total int. points > 0.500000 JY
Will print all polarization points > 1000.000000 JY
Will print all uv differences > 0.100000 lambda
File1 = TY2KL .CALIB . 1 Vol = 1 UserID = 36 Channels = 1 to 1 BIF= 1
File2 = MY2KL .CALIB . 1 Vol = 1 UserID = 36 Channels = 1 to 1 BIF= 1
Source= 3C405 RA = 19 59 28.36 DEC = 40 44 2.0
Freq= 8.514899759 GHz ncor= 2 No. vis= 425155 Sort order= TB
Weights have been multiplied by 1.00E+01
Source= 3C405 Freq= 8.514899759 Sort= TB 1 RR 1 LL
Vis # IAT Ant U(Klam) V(Klam) W(Klam) Amp Phas Wt Amp Phas Wt
0 Points printed
0 Points found for flux differences
0 Points found for u,v,w differences
0 Points found for other differences
0 Points swapped for apparent sort order reversal
Visibilty Max diff/peak = 5.030E-05 RMS/peak = 9.623E-10
If you want to look at some of the results - the CYG A image of the
Y2K test is rather nice - try the verb TVLOD, perhaps with adverb
FUNCT='SQ' and then TVFIDDLE. See the CookBook chapter 6 for display
options. The CookBook can be gotten to from the web page
http://www.aoc.nrao.edu/aips/cook.html
and you have $AIPSPUBL/COOK6.PS on your system as well. The web gets
you to an html and pdf version which are cross-linked. The latter is
also on your system $AIPSPUBL/COOKBOOK.PDF.
Eric Greisen
More information about the Daip
mailing list