[daip] responses

[Lawrence Rudnick]

Thanks for comments from Namir and Rick. Here are some responses,
and a further question for Rick.


1. did we use FLATN images, which have problems with flux? and 
calibration issues?
     yes. there is no choice except for FLATN.  the problems with FLATN
     have to do with peak flux, but integrated should be ok.  integrated
     for us shows same *bias* problems.
     there is no choice because we are using multiple pointings,
     and counting on the sensitivity to be gained by putting together all
     the pointings.  and LTESS has a scaling/PBCOR problem.

2. re: the PB asymmetry - this is probably related to what we're seeing
    on the problem of getting different answers from different pointings.
    we have snapshots spread over 7 hours for each field.
RICK - WHAT DO YOU HAVE MORE QUANTITATIVELY ON ASYMMETRY, ETC. IN PBAND
PRIMARY BEAMS?


3. self-cal causing problems? - phase only, so should be ok.

4. confusing sources way out? - i forget how far i looked - i think at
    least 20 degrees - didn't see any obvious problems. had imagr work
    on fields that were 5.7 degrees across for each pointing.

5. general confusion?  - yes, this could be a problem.  i've never
    worked in/near confusion before, don't understand how clean works,
    etc. RP comments that 'large and complex objects', fidelity not
    as good as apparent dynamic range.  i guess that approaching
    confusion is in a similar category to 'large and complex'?


6. extrapolation to zero - should it?  yes.  assume there is a single
     spectral index between L and P, then definitely extrapolates to 
zero.  assume now there is a mixture, then you can think of this as
a series of lines with different slopes in the L vs. P diagram, each
of which extrapolates to zero, so what it should look like in the
absence of noise is a wedge in L vs. P with its apex at zero.
     there are two ways to get a non-zero extrap.  1) an offset,(NOT
a scale factor, so the calibration scale, etc. is not an issue); and
2) a major change in the mean spectral index below the fluxes
that are being sampled.
      you can see what would be needed if 2) were true by looking at
http://www.astro.umn.edu/~larry/LSS/index.html#fluxes
the spectral index distribution gradually becomes flatter, overall,
as flux densities decrease (not the emergence of a new population)
and a significant part of the distribution at low flux densities are
inverted spectra.  all of this disappears
    http://www.astro.umn.edu/~larry/LSS/index.html#biascorr
if you put in a bias correction.
    so, my conclusion is there is a bias.  don't know where it comes
from.

lr