Hi John,
I think shifting the nominal ell of that first bin is a no brainer. I have
slight reservation about moving them all about by epsilon amounts. But as you
say it doesn't really matter.
Clem
--
**********************************************************************
Clem Pryke - Associate Professor - Physics
University of Minnesota,
Room 313 Tate, 116 Church Street S.E. Minneapolis, MN, 55455
Tel: 612-624-7578 Fax: 612-624-4578 email: pryke(a)physics.umn.edu
**********************************************************************
> An alternative would be to plot our points at the band centers (bpwf
> center of mass or similar), instead of the naive band centers.
I think this is a good idea. We should update the code in reduc_final to
correct the ell value after the iteration over supfac and bpwf. (We should
only do this for bins where the shift is significant which is presumably just
the lowest one.)
Clem
--
**********************************************************************
Clem Pryke - Associate Professor - Physics
University of Minnesota,
Room 313 Tate, 116 Church Street S.E. Minneapolis, MN, 55455
Tel: 612-624-7578 Fax: 612-624-4578 email: pryke(a)physics.umn.edu
**********************************************************************
http://bicep0.caltech.edu/~spuder/keck_analysis_logbook/analysis/20131002_s…
Many dark detectors see the signal.
Dark SQUIDs don't see the signal.
This confirms that the pickup mechanism requires a live, responsive TES.
Also shows that the pickup mechanism doesn't go through the antenna
network and lumped element filter.
- Walt
Hi all,
I have put up a new Keck Array posting:
http://bicep0.caltech.edu/~spuder/keck_analysis_logbook/analysis/20131001_s…
Robert and the South Pole satcom crew have run a second test on the
pickup of GOES transmissions, following up on the previous test
(results in [20130925_satcom_goes_test]). There are a few reasons for
the second test: the weather got bad during the first test, introducing
odd features in some channels, which we wanted to make sure were not
related to the satellite comms; and this test varied the power level of
the high-power amplifier. This gives us the opportunity to find out
whether the signal in our data scales with power or field amplitude
(see [20120411_SkyNet...] for a similar test with a different
transmitter). I confirm that the bad-weather channels look fine this
time and don't seem to have been seeing the ground station. I also find
that the signal level is consistent with a linear scaling with power,
not field amplitude.
- Walt
> how about only keeping rel gain deproj and using subtraction for
> all beam systematics ?
That is of course the logical next question. The planned code changes will
allow this and we should take a look at what happens.
The absolute amount removed by the 6 main beam modes is rather comparable for
jack0. However for deck jack it is order of magnitude higher for diff points
(because the dk rot "obs strategy" cancellation is not in effect).
Clem
--
**********************************************************************
Clem Pryke - Associate Professor - Physics
University of Minnesota,
Room 313 Tate, 116 Church Street S.E. Minneapolis, MN, 55455
Tel: 612-624-7578 Fax: 612-624-4578 email: pryke(a)physics.umn.edu
**********************************************************************
> Subtraction based systematic removal might be good enough for TE or EE , but
> if our default plan is to use deprojection for BB , it is a slightly
> complicated story to tell. If we are forced to do it we will do it.
To be clear - I am proposing to switch diff ellip removal from deproj (coeff
fitted from CMB data) to subtraction (coeff taken from beam map info) for all
spectra. So there is no significant complication to the story.
I don't think there is any reason to believe this will not be good enough. As
I said the undeproj residual would be need to validated in Chin-Lin's sims
using the same scheme.
We'll give it a try and see what happens.
Clem
--
**********************************************************************
Clem Pryke - Associate Professor - Physics
University of Minnesota,
Room 313 Tate, 116 Church Street S.E. Minneapolis, MN, 55455
Tel: 612-624-7578 Fax: 612-624-4578 email: pryke(a)physics.umn.edu
**********************************************************************
I've put up a new BICEP2 posting:
http://bicep0.caltech.edu/~spuder/analysis_logbook/analysis/20130930_satcom…
In this note I take another look at the two scansets in March 2012
where the satcom signal appeared to turn on when it should turn off (at
the end of the GOES pass). In at least one of them, I find that
there's actually a weak signal there all along, and it just gets
stronger and shifts around half-scan 60.
There's a lot of variation in pair-diff signal level during GOES
passes, depending on unknown factors. It seems likely that in these two
scansets we happened to have a low but nonzero level of contamination,
and at the end of the GOES pass something happened to give us a much
larger signal. This step up could perhaps be a different transmitter,
for TDRS for example.
Another idea: perhaps the antenna stopped tracking and went to its
standby position, but the high-power amplifier was inadvertently left
powered on. This would move whatever sidelobe we're seeing, perhaps
changing the power level and the spatial pattern.
- Walt
I think there may be a much simpler way to deal with this issue.
Instead of deprojecting diff ellip we can instead just subtract it. We are
presumably very confident that the coefficients are not time varying and
hopefully we have reasonable s/n on them from the beam maps.
Adding the ability to subtract rather than deproject to the
reduc_coaddpairmaps was something I John and I had already decided should be
done. We could have something like coaddopt.deproj=[1,1,0,2] specify that for
diff ellip the coefficients should be taken from get_array_info's output
rather then regressed. We will want to take some care that for sims the "as
simulated" values are used.
We can make the real maps for this new dp1102 type and see if the high ell
TB/EB goes away as it does for dp1101. If it does we would need to re-run the
sim set for this new option. We would also need to do this new option in
Chin-Lin's sims to validate it in terms of residual false BB.
I will take a look at the code and talk with Chris about how to implement this
in the cleanest possible way - right now vector coaddopt.deproj does something
else...
Clem
--
**********************************************************************
Clem Pryke - Associate Professor - Physics
University of Minnesota,
Room 313 Tate, 116 Church Street S.E. Minneapolis, MN, 55455
Tel: 612-624-7578 Fax: 612-624-4578 email: pryke(a)physics.umn.edu
**********************************************************************
> We should discuss what we want to do for TE.
The problem is not just for TE.
Look at the supfac plot
http://bicep.caltech.edu/~spuder/analysis_logbook/analysis/20130909_sim1450…
and click between "dp+rg" and "rg+dp+diff-ellipt".
Deprojecting diff ellip causes the EE curve (red) to go up around ell=175! This is craziness - filtering modes from the map increases polarized power? It is presumably only possible because of the bias on the deprojection coefficients - they have a DC offset which is presumably a transfer of power from T to pol at these ells?
Things look fine in the final EE plots because we force this effect away by renormalizing the sim output to match input (and because the real sky is apparently LCDM like these sims). We notice a problem in TE simply because (due to the technical problem of zero crossings) we don't normally use a direct TE suppression factor. Instead we use gmean(TT,EE) and this apparently isn't correct so we notice an in/out mismatch.
We need to think about this more. Right now I am highly skeptical that the EE (and TE) points with diff ellip deproj on will ever be trustworthy enough for cosmological parameter analysis.
Clem
--
**********************************************************************
Clem Pryke - Associate Professor - Physics
University of Minnesota,
Room 313 Tate, 116 Church Street S.E. Minneapolis, MN, 55455
Tel: 612-624-7578 Fax: 612-624-4578 email: pryke(a)physics.umn.edu
**********************************************************************
Hi,
I put up a (mini) post showing the 1450/1350 data compared to WMAP.
Nothing surprising. We have very good data through the second peak. The
second bin, which we have sometimes fretted about (the last simset didn't
have any E modes as large as our data), is in complete agreement with WMAP.
http://bicep.caltech.edu/~spuder/analysis_logbook/analysis/20130930_wmappol…
-Sarah