Recent Changes - Search:
Main /

T 09240

Test with Agilent and SMY on 3c273

Correlation notes

512 chan delay fix pols

  1. on hex0:
     cd ~/correlations
 mkdir t09240
  2. copy machines, run.sh2 and .threads file from a previous experiment. note the threads file should now be named "threads", not "<expt_code>.threads"
  3. copy .skd file across from hobart 
     vex2config.pl -cross -evlbi -new t09240.skd
 -cross indicates create cross-products
 -evlbi indicates data will be streamed using vsib_send and vsib_recv
 -new for the new version of DiFX

This creates t09240.input

edit the input file and check the following: 

 EXECUTE TIME (SEC): 
 START MJD:          
 START SECONDS:

First time with data to both telescopes is 05:04:20. Run time2sec.pl to convert that time to seconds: 

 observer@hex0:~/correlations/t09240$ time2sec.pl 05:04:20
  05:04:20 -> 18260

so in the input file: 

 START SECONDS:      18260

For an initial fringe check, set 

 EXECUTE TIME (SEC): 300
 OUTPUT FILENAME: /home/observer/correlations/t09240/t09240_1.rpf

Lots of channels for initial search 

 NUM CHANNELS:       512

NOTE: 1024 chans doesn't seem to work!

It turned out that default clocks were way out. Try these for a start: 

 # TELESCOPE TABLE ##!
 TELESCOPE ENTRIES:  2
 TELESCOPE NAME 0:   HOB
 CLOCK DELAY (us) 0: -3
 CLOCK RATE(us/s) 0: 0.0
 TELESCOPE NAME 1:   CED
 CLOCK DELAY (us) 1: 1.05
 CLOCK RATE(us/s) 1: 0.0

Polarisations are swapped at Ceduna, so in the DATASTREAM TABLE, swap pols for second antenna (Ceduna): 

 TELESCOPE INDEX:    1
 ...
 INPUT BAND 0 POL:   L
 INPUT BAND 0 INDEX: 0
 INPUT BAND 1 POL:   L
 INPUT BAND 1 INDEX: 1
 INPUT BAND 2 POL:   R
 INPUT BAND 2 INDEX: 0
 INPUT BAND 3 POL:   R
 INPUT BAND 3 INDEX: 1

Also in the baseline table: 

 # BASELINE TABLE ###!
 ...
 POL PRODUCTS 0/1:   4
 D/STREAM A BAND 0:  1
 D/STREAM B BAND 0:  3
 D/STREAM A BAND 1:  3
 D/STREAM B BAND 1:  1
 D/STREAM A BAND 2:  1
 D/STREAM B BAND 2:  1
 D/STREAM A BAND 3:  3
 D/STREAM B BAND 3:  3

Lastly, the port numbers for data 

 # NETWORK TABLE ####!
 PORT NUM 0:         52100
 TCP WINDOW (KB) 0:  512
 PORT NUM 1:         52101
 TCP WINDOW (KB) 1:  512

Now make the model 

 vex2model_updated.pl t09240.skd

Edit the machines file. We're not using hex1, so it looks like this: 

 hex0
 hex5
 hex4
 hex2
 hex3
 hex4
 hex5

Hobart data on hex5, Ceduna on hex4

edit the threads file. Note number of spaces between text is important: 

 NUMBER OF CORES:    4
 8
 8
 7
 7

edit run.sh 

 -np 7 (where 7 is # processes, = wc -l machines)

Then start correlation:

  1. Open 3 windows, one on hex0 (head node), one on hex5 (Hobart data), one on hex4 (Ceduna data)
    1. hex5: cd /data/glast_agn/t09240
    2. hex4: cd /data_fermi/t09240/cd
    3. in one of the above (if necessary), delete any initial data files where there's no overlap
    4. hex5: vsib_send -H hex5 -p 52100 *.lba but don't press [Enter] yet!
    5. hex4: vsib_send -H hex4 -p 52101 *.lba but don't press [Enter] yet!
    6. hex0: cd ~/correlations/t09240
    7. hex0: ./run.sh
    8. When you see this:

[ -1] INFO MPI Process 0 is running on host hex0.phys.utas.edu.au Press [Enter] in the hex5 and hex4 windows. Data should start moving and the RPF file should appear and grow.

Press Ctrl-C to stop correlation. Before re-starting, change the input file to give a different output RPF file, or delete the existing RPF file, otherwise DiFX won't start.

When the correlation is finished

Setup aips on ares: 

 source /usr/local/pkg/AIPS_31DEC06/LOGIN.CSH
 $CDTST

Run aips 

 aips tv=local