[daip] problem with VBCAL

[Eric Greisen]

Lynn D. Matthews wrote:
> I am working with some data that require applying a scaling factor to 
> the gains, and I have been attempting to do this using VBCAL.
> 
> However, when I look at resulting visibilities using, e.g., PRTUV, I see 
> that the program is turning zeros into infinities, e.g.,
> 
> BEFORE VBCAL:
> 
> TEST_32T  0.201599994 TB      1 RR          1 LL          1 RL          1 
> LR
>   Vis #     IAT      Ant    Amp Phas Wt   Amp Phas Wt   Amp Phas Wt   Amp 
> Phas Wt
> 
>      30  0/02:57:14  1-30    0.0 -61 20    0.0  87 20    0.0-146 20    0.0 
> -84 20
>      31  0/02:57:14  1-31    0.0-103 20    0.0  95 20    0.0  92 20 
> 0.0-106 20
> 
> AFTER VBCAL:
> 
> TEST_32T  0.201599994 TB      1 RR          1 LL          1 RL          1 
> LR
>   Vis #     IAT      Ant    Amp Phas Wt   Amp Phas Wt   Amp Phas Wt   Amp 
> Phas Wt
> 
>      30  0/02:57:14  1-30    0.0 -21 20    Inf  87 20    Inf-146 20    0.0 
> -31 20
>      31  0/02:57:14  1-31    Inf-103 20    Inf  95 20    Inf  92 20 
> Inf-106 20
> 
> 
> I found that this does not occur with a version of 31DEC08 aips, but this 
> was also a different machine architecture.

VBCAL has not been changed in any serious way in > 8 years.

The fluxes above are not 0.0 - they are just < 0.05 - pure 0 would 
require phase to be 0 also to avoid an abort on ATAN2(0,0).

Which are the 2 architectures?

Thanks

Eric Greisen