[Difx-users] vex2difx freqId list

Wed May 16 09:00:00 EDT 2018

Hi Cormac,

thanks for your reply! Yes freqId is global, I think it refers to
the frequency table in difx.

In the end I avoided vex hacking by experimenting more with the
freqId list. Started with 0 and added new items one by one until
things broke.

And it was a user mistake...

The freqId takes frequency ids, but I was looking at the subbands
reported by vexpeek and vex2difx. There were two bands per frequency
so "freqIds = 0,1,2,3,4,5,6,7,10,11" (vexpeek/vex2difx subband IDs)
had to be reduced to "freqIds = 0,1,2,3,5" (freq IDs).

Found later that the list of acceptable frequency IDs is actually in
the .input file FREQ TABLE. The indices <n> of the first few "FREQ
(MHZ) <n>" entries, just before zoom freqs in the same table, are
indices usable in 'freqIds'. I updated the vex2difx wiki with this info.

cheers,
Jan

Am 16.05.2018 um 14:14 schrieb Cormac.Reynolds at csiro.au:
> Hi Jan,
>
> My understanding of this is that freqId is a global parameter, not
> station specific, while the frequency lists are station specific,
> not global. This poses an obvious  ambiguity where different
> stations have different frequency structures.
>
> This has been discussed at a previous difx meeting, but IIRC the
> fix was not quite trivial and the demand for a generalisation was
> insufficient to force a change.
>
> In your scenario, selecting just freqId=2,3 (or perhaps
> freqId=11,12) might do what you want,  as I'm not quite sure how
> the selection logic works in detail, but have often found that it
> does the 'sensible' thing in poorly defined scenarios.
>
> If not, a cheat I have used in the past is to set the frequencies
> I wish to discard to some nonsense value such that they do not
> form a correlatable baseline. This vex hacking can be a bit messy,
> but has a very predictable result.
>
> Regards,
> Cormac
>
> Sent from Blue <http://www.bluemail.me/r?b=12976>
> On 16 May 2018, at 19:27, Jan Wagner <jwagner105 at googlemail.com
> <mailto:jwagner105 at googlemail.com>> wrote:
>
>     Hi again,
>
>     got another potential user or interpretation mistake related to v2d.
>     This cropped up while trying to fix something in v2d+vex for
>     RadioAstron correlation. Basically we have stations with 16 MHz
>     recorded bands that overlap only a 1 x 64 MHz band out of VLBA 4 x
>     64 MHz. The other 3 x 64 MHz should supposedly be thrown away.
>
>     We tried to create 16 MHz zooms into just that 64 MHz VLBA band, and
>     discard the other 64 MHz bands via including only desired bands in
>     the v2d SETUP "freqId" parameter ("select which bands to
>     correlate"). However, vex2difx then complains about mismatched
>     bandwidth correlation. Despite the correct (I think) freqId band
>     selection.
>
>     With 'vexpeek' the original list of bands is:
>
>     # 16 MHz stations
>     # 22236 MHz +16M = 22236 MHz to 22252 MHz
>     Subband[0]=[2.2236e+10 Hz, 1.6e+07 Hz, sb=U, pol=R]
>     Subband[1]=[2.2236e+10 Hz, 1.6e+07 Hz, sb=U, pol=L]
>     # 22236 MHz -16M = 22220 MHz to 22236 MHz
>     Subband[2]=[2.2236e+10 Hz, 1.6e+07 Hz, sb=L, pol=R]
>     Subband[3]=[2.2236e+10 Hz, 1.6e+07 Hz, sb=L, pol=L]
>     # 22268 MHz +16M = 22268 MHz to 22284 MHz
>     Subband[4]=[2.2268e+10 Hz, 1.6e+07 Hz, sb=U, pol=R]
>     Subband[5]=[2.2268e+10 Hz, 1.6e+07 Hz, sb=U, pol=L]
>     # 22268 MHz -16M = 22252 MHz to 22268 MHz
>     Subband[6]=[2.2268e+10 Hz, 1.6e+07 Hz, sb=L, pol=R]
>     Subband[7]=[2.2268e+10 Hz, 1.6e+07 Hz, sb=L, pol=L]
>
>     # 64 MHz VLBA stations
>     # 22156 MHz +64M = 22156 to 22220 MHz ; *discard*
>     Subband[8]=[2.2156e+10 Hz, 6.4e+07 Hz, sb=U, pol=R]
>     Subband[9]=[2.2156e+10 Hz, 6.4e+07 Hz, sb=U, pol=L]
>     # 22220 MHz +64M = 22220 to 22284 MHz, keep
>     Subband[10]=[2.222e+10 Hz, 6.4e+07 Hz, sb=U, pol=R]
>     Subband[11]=[2.222e+10 Hz, 6.4e+07 Hz, sb=U, pol=L]
>     # 22396 MHz +64M = 22396 to 22460 MHz ; *discard*
>     Subband[12]=[2.2396e+10 Hz, 6.4e+07 Hz, sb=U, pol=R]
>     Subband[13]=[2.2396e+10 Hz, 6.4e+07 Hz, sb=U, pol=L]
>     # 22460 MHz +64M = 22460 to 22524 MHz ; *discard*
>     Subband[14]=[2.246e+10 Hz, 6.4e+07 Hz, sb=U, pol=R]
>     Subband[15]=[2.246e+10 Hz, 6.4e+07 Hz, sb=U, pol=L]
>
>     The ZOOM section has zooms for only the desired one band out of 4 x
>     64 MHz VLBA:
>
>     addZoomFreq = freq at 22220.00/bw at 16.0/noparent at true
>     addZoomFreq = freq at 22236.00/bw at 16.0/noparent at true
>     addZoomFreq = freq at 22252.00/bw at 16.0/noparent at true
>     addZoomFreq = freq at 22268.00/bw at 16.0/noparent at true
>
>     The SETUP section has selects only the desired 16 MHz bands and the
>     one 64 MHz band:
>
>     freqIds = 0,1,2,3,4,5,6,7,10,11
>
>     However, vex2difx still complains about mismatched bandwidths. The
>     complaint disappears only when I add three more 16 MHz zooms so that
>     a piece of each three undesired 64 MHz VLBA bands is included.
>
>     Apparently the "freqIds" list is either ignored, or not parsed
>     correctly, or is checked too late?
>
>     The wiki describes the SETUP parameter "freqId" as "a comma
>     separated list of integers that are freq table indexes to select
>     which bands to correlate; default is to correlate all. Note: this
>     should be used to select parent bands for zoom frequencies if
>     difx2fits is to be run."
>
>     From the actual behaviour though it seems that "freqId" does not
>     actually "select which bands to correlate"?
>
>     cheers,
>     Jan
>
>
>
>     --------------------------------------------------------------------
>
>     Difx-users mailing list
>     Difx-users at listmgr.nrao.edu
>     https://listmgr.nrao.edu/mailman/listinfo/difx-users
>