Hi all,
I have a posting on crosstalk as estimated from CMB maps.
http://bicep0.caltech.edu/~spuder/analysis_logbook/analysis/20130826_cmb_xt…
Nearest neighbor crosstalk is detected at a level of around 0.4% in B
detectors and 0.1% in A detectors. This indicates that it's largely
settling-time xtalk, i.e. it preferentially comes from the previous
channel in the muxing order. There are some channels that may have
crosstalk levels of a few percent, but the signal-to-noise isn't great
and they need some more work to be sure.
This confirms that we do indeed have crosstalk in our CMB maps and that
it's at a level consistent with expectations.
- Walt
Hi All,
I have added a posting:
http://bicep.caltech.edu/~spuder/analysis_logbook/analysis/20130826_xcorr_p…
deriving the 2-point correlation maps for the newest set of maps built by
Chin Lin using Planck 143 T.
I will get some more elaborated conclusions adding a more complete set of
sims. If time allows, however, we may have a look tomorrow.
Thanks. Cheers,
Sergi
--
Dr Sergi Hildebrandt Rafels
Jet Propulsion Laboratory ----- California Institute of Technology
169-217 Cahill-383
4800 Oak Grove Drive 1200, E. California Bvd
Pasadena, CA, 91109 Pasadena, CA, 91125
MC/169-237 MC/367-17
1-818-354-0220 1-626-395-2147
Hi
I've put up a posting on a follow up to a lot of the issues that we brought
up last week on my rev2 composite beam map sim.
http://bmode.caltech.edu/~spuder/analysis_logbook/analysis/20130826_composi…
I include the dp1111 BB plots in this posting, as well as plot the
jackknife BB, TB, and EB spectra for this sim and compare it to real data
from sim1234. I also include the spectra and the jackknifes for a beam map
sim run with a Planck sky as the input map.
Thanks.
Chin Lin
Hi Chin-Lin,
Thanks for your posting - sorry I couldn't join the telecon discussion.
As I have said several times I think it is extremely important to look at the
jack spectra - the non-jack should not be taken at face value unless the
corresponding jacks are at or below the levels seen in the real data.
Otherwise it is clear that the input beam maps are not sufficiently clean.
In this post you don't show any jacks - however you do provide the map pager
and I just spent some time looking at this and comparing this to the real map
pager in 20130802 post from Stefan and Sarah.
1) In your map pager click "Q", "apmap" and "diffpoint". Then alternate
between jack "none" and "dk" - we see a radical change in the pattern - note
that dk jack color stretch is larger but the residual is concentrated in the
bottom part of the field and looks to be much more small scale power. Several
of the other jacks show patterns on equal color stretch as the non jack (tile
and fp-inner/outer).
2) Now go to "B-pure" and select the identical options in the real pager -
B-pure, apmap, diffpoint. Now compare your sim versus real for jack none -
clearly there is way more power in the real but some of this is noise and E/B
mixing. Going to dk jack the color scales differ by factor 3 but it does not
look like your sim is ruled out by this test. Switching back to jack none and
selecting the r=0.1 sim which is available in the Stefan/Sarah pager it does
look like your predicted false signal is in the ball park of r=0.1 consistent
with your 1d plots.
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 All,
When Chris does the forward sims of the crosstalk beams, it would be
instructive to run a case for the Keck configuration as well. Since we
don't have Keck beam parameters, I'm thinking just take the BICEP2
parameters for each FPU and rotate 5x around the mount. I expect the cross
talk leakage averages down in Keck due to all the rotation angles and are
highly reduced in B2xKeck. Which sounds like a systematic of interest. It
would also be useful to have this Keck signature for Sergi's analysis with
the TB cross-correlation.
Jamie
-----Original Message-----
From: bicep2-list-bounces(a)lists.fas.harvard.edu
[mailto:bicep2-list-bounces@lists.fas.harvard.edu] On Behalf Of Clem Pryke
Sent: Tuesday, August 20, 2013 10:46 AM
To: Chin Lin Wong
Cc: bicep2-list
Subject: Re: [Bicep2-list] Posting: Rev2 Composite beam maps
Hi Chin-Lin,
Thanks for your posting - sorry I couldn't join the telecon discussion.
As I have said several times I think it is extremely important to look at
the jack spectra - the non-jack should not be taken at face value unless the
corresponding jacks are at or below the levels seen in the real data.
Otherwise it is clear that the input beam maps are not sufficiently clean.
In this post you don't show any jacks - however you do provide the map pager
and I just spent some time looking at this and comparing this to the real
map pager in 20130802 post from Stefan and Sarah.
1) In your map pager click "Q", "apmap" and "diffpoint". Then alternate
between jack "none" and "dk" - we see a radical change in the pattern - note
that dk jack color stretch is larger but the residual is concentrated in the
bottom part of the field and looks to be much more small scale power.
Several of the other jacks show patterns on equal color stretch as the non
jack (tile and fp-inner/outer).
2) Now go to "B-pure" and select the identical options in the real pager -
B-pure, apmap, diffpoint. Now compare your sim versus real for jack none -
clearly there is way more power in the real but some of this is noise and
E/B mixing. Going to dk jack the color scales differ by factor 3 but it does
not look like your sim is ruled out by this test. Switching back to jack
none and selecting the r=0.1 sim which is available in the Stefan/Sarah
pager it does look like your predicted false signal is in the ball park of
r=0.1 consistent with your 1d plots.
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
**********************************************************************
_______________________________________________
Bicep2-list mailing list
Bicep2-list(a)lists.fas.harvard.edu
https://lists.fas.harvard.edu/mailman/listinfo/bicep2-list
Hi
I've put up a posting on the 2nd revision of composite beam maps, and the
first results of feeding the composite map into one sim realization.
http://bmode.caltech.edu/~spuder/analysis_logbook/analysis/20130819_composi…
I make rev2 composite beam maps from the median of constituent maps, and
look at the polarization leakage in the sim. The results are interesting!
It looks like the rev2 composite beam maps have the level of leakage that
is slightly less than the BB spectra at r=0.1 after deprojection is
applied. I also look at the per-pair leakage for each pixel.
It would be great if we could discuss it on the telecon
Chin Lin
Hi All,
you may find in:
http://bicep.caltech.edu/~spuder/analysis_logbook/analysis/20130820_xcorr/
an analysis that may be of interest given other recent work, in particular
from Chin Lin and Walt.
I think it may be worth it discussing during the telecon.
4 am limit, I think good time for this stuff. I am on vacations and will
be busy in the morning, but I think Jamie could comment on his ideas.
Sergi
--
Dr Sergi Hildebrandt Rafels
Jet Propulsion Laboratory ---------------- California Institute of
Technology
169-217 Cahill-383
4800 Oak Grove Drive 1200, E. California Bvd
Pasadena, CA, 91109 Pasadena, CA, 91125
MC/169-237 MC/367-17
1-818-354-0220 1-626-395-2147
I've put up a posting that tries to tie up a couple of loose ends from
the Run 7 data:
http://bicep0.caltech.edu/~spuder/analysis_logbook/analysis/20130819_run7_b…
The question is how much the blue leak was reduced with the island
redesign, but before the addition of the Ade filter. I find a response
with TGF in place that's about 1% of the open-aperture response.
- Walt
Hi,
I just put up a posting comparing the first 20 sims from the current round
of sim production to the sims that Eric/Stefan made in April.
The main differences are:
- noise is up by a factor of 1/0.85 due to the new channel cuts
- the beam suppression ratio is ~10% different at ell~100
Nothing unexpected. Just things to keep in mind as we start looking at the
chi2 pager...
-Sarah
Hi all,
I have put up a posting with several options for getting rid of
artifacts in the composite beam maps that are used in Chin Lin's sims.
This doesn't need a full discussion during the Tuesday telecon, since
the results have already been communicated to Chin Lin for use in sims.
http://bicep0.caltech.edu/~spuder/analysis_logbook/analysis/20130814_beamma…
1) Mean of maps -- what we've been using -- contains "acne", stripes,
and other artifacts of transient features from input maps
2) Median of maps -- gets rid of many of these features and is simple to
implement
3) Masked mean -- cuts out parts of maps that are too discrepant from
the median.
The masked mean is able to identify and remove most of the transient
features. It could be combined with inverse-variance weighting to make
a weighted, masked mean, which should in principle be able to do better
than the median. However, it is also more complicated, and would take
additional "tuning". For now, the median does a bit better, and has
the advantage of simplicity.
Therefore, I recommend using the simple median-compositing algorithm
for now, and this is what Chin Lin is doing in the current round of
sims.
- Walt