User Tools

Site Tools


analysis:correlator-report

Correlator report generation

This is done on magilla. Below, post-processing of AUST10 is shown as an aexample

Binary files (qcodes and snratio) are in ~/bin

Make sure the directory 1234 for AUST10 is in /data/AUSTRAL/aust10

$ cd /data/AUSTRAL/aust10/1234
$ aedit -f alist.ed.out
> edit dup SNR
> psfile 1234.psfile_rf
   give full path to 1234.ovex file
$ qcodes

when prompted, enter “aust10 1234”

This will create the file 1234.qcodes. Note qcodes expects to find the psfile in /data/AUSTRAL/aust10/1234/1234.psfile_rf so check path is correct and change name of aust?? directory to lower case if needed. Alternatively, you can use “AUST10 1234” at the prompt in which case it will look in /data/AUSTRAL/AUST10/1234/1234.psfile_rf

You need the a priori SNR values which sked can generate. Currently sked is only running on a virtual machine on Jim's Mac! A-priori SNRs for AUST10 to AUST16 have been done. To do this, on the virtual machine:
$ cd ~/schedules
$ ./snrlist.sh aust11.skd 
$ ./snrlist.sh aust12.skd  
$ ./snrlist.sh aust13.skd 
$ ./snrlist.sh aust14.skd 
$ ./snrlist.sh aust15.skd 
$ ./snrlist.sh aust16.skd 


All output files copied to hex2:/data/AUSTRAL/snr_apriori

Back on hex2:

$ cd /data/AUSTRAL/aust10
$ mkdir SNR
$ cd SNR

Copy the final alist file, stations.m and aust10_snr.apriori to the SNR directory. You can get stations.m from a previous experiment. Make sure the single letter codes match the two letter codes as defined earlier in the processing. Check by looking in a scan sub-directory (e.g. ../1234/185-0700/<source>.wzpetk) and edit the stations.m file if necessary.

In the case of aust10:

Ke=i
Hb=L
Ht=g
Yg=x

To avoid memory stack problems in snratio:

$ ulimit -s 102400
$ ~/bin/snratio aust10_snr.apriori alist.ed.out

This produced an error message:

“Could not match S baseline Lg from alist”

Lg = Hb-Ht, but the apriori file has this baseline as Ht-Hb (i.e. gL), so edit line 2 of the apriori file and change both instances of Ht-Hb to Hb-Ht. OK now

Check /data/AUSTRAL/Stations_Codes_CR.txt agrees with coding used for experiment Create a file called /data/AUSTRAL/aust10/1234/aust10.txt with some header information as below

Experiment: A1401
Correlator: dummy dummy CRTN
Start: 2014-247-04:00:00 - 2014-248-03:51:15
name    dummy Ht Hb Ke Ww Yg

Put vex and v2d files from Curtin (name them aust10.vex, aust10.v2d) into /data/AUSTRAL/aust10

$ WriteCR.pl

For the correlation dates use the Curtin data release date as the end and one day previous as the start. Output file is called /data/AUSTRAL/aust10/1234/aust10.corr.perl

Edit the file and add any correlator or station notes by hand. Note the database entry is currently blank.

One thing to check for is the presence of clock breaks. This is best done using aedit with the command “plot mbd”. You may need to set the y-scale to see things clearly with “yscale -0.05 0.05” (scale in microseconds, so this gives a \pm 50 ns window)

If you are producing a tarball to send to Washington, you can considerably reduce the size of the file by excluding the baseband data when using tar. To do this, use this command from the data directory:

 tar --exclude='*..*' -cvf  aust10.tar aust10

This will reduce the size of the gzipped tarball from by a factor of ~10-20 (Expect sizes of a few 100 MBs).

If you're creating a database yourself, leave the baseband data alone. You can now proceed to the next step using calc/solve (installed on ops7).

/home/www/auscope/opswiki/data/pages/analysis/correlator-report.txt · Last modified: 2014/11/18 04:29 by Jamie McCallum