Hi all,
We will have our BICEP2 / Keck CMB analysis telecon tomorrow at the
usual time:
Tues, 20 Aug 2013 - 12:00 Eastern, 11:00 Central, 9:00 am Pacific
1-866-890-3820 (toll: 1-334-323-7229) Passcode: 59702175#
Starting this week, we will begin promptly after the hour and cap the
telecon at 2:30 max.
Please send an email when you put up any posting giving a brief summary
and noting if it needs to be discussed on the telecon.
New postings must be up by 7am eastern to be discussed this week.
Here's tomorrow's agenda in brief:
1) Papers
- Paper I: Instrument paper, updates since last week? [ Walt ]
- Paper II: Beams paper, detailed status report [ Abby ]
http://bmode.caltech.edu/~bicep2/papers/2013_beams/
- timeline on assigned tasks, status of unassigned tasks
2) Computing and simsets
- odyssey / general computing report: [ Walt ]
- odyssey changes
- panlfs usage
- sim 1450 progress report [ Stefan ]
- timelines to: reduc_final pager, 50 rlz
3) Postings [ Clem out so John to lead this week... ]
4) Other initial-result-critical analysis tasks [ John to lead ]
- What tasks need effort? Can we assign it?
- “steady” crosstalk -- quantify from beammaps and then sim
- “variable” crosstalk -- what’s the path to constraining?
- sidelobe sim / spherical convolution implementation [ Sergi ]
- composite sidelobe maps
- B1 x B2
Full agenda is in the telecon notes here http://goo.gl/LNvpx
Add any more items you want discussed there before the telecon.
John
Hi everyone,
I have posted a note that extracts crosstalk levels between arbitrary
detectors using cosmic ray hits in the three-season BICEP2 data set.
This gives much better signal-to-noise than the CMB map estimate. There
is evidence for symmetric, inductive crosstalk at the level of 0.4%.
http://bicep0.caltech.edu/~spuder/analysis_logbook/analysis/20130830_cosray…
- Walt
Hi all,
I have a second posting on the crosstalk estimates from CMB data. I
have repeated the earlier analysis but without the [ADU/airmass]
factor, so that a 1% crosstalk signal will now be simply 1% in units of
CMB temperature. The results match the previous CMB analysis, but with
a few additional RGL channels included. The A/B difference and the
outliers still remain.
I think I can explain these features in terms of deviation from the
mean B_l beam profile used in the sims. If a detector has a different
beam shape, the signal subtraction won't work well, and the residual
may correlate with the offset crosstalk beam from the crosstalk sims.
Fig. 4 of the note shows a correlation between beem width and estimated
crosstalk, and I take this as support for the idea that beam shape is
having an effect here.
http://bicep0.caltech.edu/~spuder/analysis_logbook/analysis/20130830_cmb_xt…
- Walt
Hi Sarah,
Please add some notes to your Keck 2013 map posting about the 90 deg jack
testing for uniform monopole effects and demonstration that it works in rx2.
And also that rx2 "needs" relgain deproj a lot less for jack=none in 2012 than
it does for 2013. (This last point is hampered a bit by changing color scales
in speclin but I think it's valid.)
It's really neat to see these things working in practice.
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,
We've had some email threads going about prelim results seen in Stefan
and Sarah's latest pager and I wanted to share a few of my thoughts with
the whole group. Click on
http://bmode.caltech.edu/~spuder/analysis_logbook/analysis/20130823_sim1450…
I find it very encouraging that the anomalously large high ell TB and BB
appears to be relieved by "all four" deproj (specifically diff ellipt.)
We should be able to confirm that the shifts in these spectra are
consistent with measured diff ellipticity--the mode that makes the
difference. Obviously this is also (another) caution to us regarding
our systematic error budget for these spectra at high ell, and the need
to carefully quantify the effects of the undeprojected residuals.
EB in B2 remains unresolved--it still looks highly significant, and its
negative amplitude roughly traces the EE spectrum. Keck doesn't see it.
Perhaps there is some error in B2's chi calibration that explains
this. Looking at the form of B2's EB (highly significant) and TB (still
non-zero, but less significant now), and comparing to the EE and TE and
the expected leakage resulting from an overall pol rotation angle
Delta-alpha:
TB = TE sin(2 Delta-alpha)
EB = 1/2 EE sin(4 Delta-alpha)
(see e.g. arXiv:0811.0618 eq 3-7), it appears to me that a rotation of
Delta-alpha = -1 deg would give a good fit to both the EB and TB
spectra. Jon Kaufman (or anyone else), are you planning to do this?
But by far the most serious issue for publication of an r limit remains
the tension between B2 and Keck in the BB spectra at ell 50-150. (Click
on lmax=200).
Clem correctly emphasized this yesterday, I think. He also
characterized the B2 BB spectra as being equivalent to r=0.3 to 0.5
yesterday. Looking at this plot myself, accounting for the known
lensing signal, I think the excess BB at ell=70 looks more like r=0.2 to
0.25. Accounting also for foregrounds (Jamie T's latest post makes dust
in that bin at r~0.05 seem not unlikely) the "unexplained" excess would
be less. Other plots may make this more clear. For now I think we
should be careful not to overstate the amplitude of the issue here, but
it is the significance of the tension between B2 and Keck that we should
continue to focus on.
I think Clem's posts from late May / early June are the right path here.
I know he is planning to repeat them using the new sims, and I think a
lot hinges on the detailed results.
For the main science goal of a limit on r, it seems necessary to me to
quantify the ell = 30-150 BB tension between B2 and Keck before we can
consider greenlighting a set of results for release. In the meantime,
we will continue to push hard on quantifying the T->P leakage from the
beam sims and uncertainties in those estimates at all ell, the impact of
crosstalk, and other threads we discussed.
Looking at the pager (lmax=500) it also seems likely we have a detection
of BB lensing. Chao-Lin has previously raised the possibility of
pushing a result out on that. I am somewhat skeptical this is any
easier than an r limit until we've worked through the systematic error
budget (we have not been focusing on high ell with jacks, etc) but I do
think this goal is a good one and motivates including the higher ell in
our effort to wrap up systematics.
John
--
___________________________________________________________________
John Kovac jmkovac(a)cfa.harvard.edu
Associate Professor, Harvard University Astronomy Department
160 Concord Ave rm 310, Cambridge MA 02138, 617-496-0611
> Clem - the more I stare at those maps, the more correlated they look.
> The big, blue blobs at ra~-5 seem very much the same.
Well there are clearly correlation with Keck2012 also (as we already know from
"experiment jack" and B2xKeck2012 spectra). I am still bothered by the lack of
power in Keck2013 on the left side - but perhaps it is just chance.
Here is a puzzle I just noticed which I expect you can clear up: Go to B2,
real, deproj=none, type1=Q, type=map, jack=dk - we see enormous diff point
induced TtoPol leakage in the Q/U maps (which cleans up in jack=none due to
coadd of 180 deg complement angles, and also clean up if we switch to
deproj=dp).
Now select Keck2013 and rx=3 - which is B2 isn't it? Now there is no obvious
TtoPol. But the only difference observing wise is the presence of 90 deg
complements added to each half of the jack - but those shouldn't cancel diff
point leakage (at least not much). What is going on? Have the differential
pointing gone way down with transplant of B2 focal plane to Keck cryostat?
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
**********************************************************************
>> For B2 and Keck2012 we had no jack to test
>> for uniform monopole. Now we do and it appears to pass.
> For certain receivers (rx2, rx4) this jack appears very sensitive to
> relgain deproj, seemingly evidence of a real monopole effect. I think
> for rx=all we must benefit from a fair bit of averaging down. Do you
> agree with this interpretation?
I think so. In as much as the effect is uniform across rx it doesn't care if
the 90 deg complement data comes from the same receiver or a different one -
so you are right that the rx=all test is a weak one. However as you point out
we in fact see a much stronger change when turning on relgain deproj for rx2
and rx4 than for the others perhaps arguing that such effect is not fully
common across rx. There is a change for rx=all so I am not saying there is
anything which doesn't make sense here.
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 Sarah,
Thanks for this very interesting pager.
I think you have defined the alt dk jack correctly - the two angle groups are
separated by 90 deg. Under 90 deg rotation a uniform monopole effect switches
from T->Q to T->-Q. Therefore a uniform monopole cancels under 90 deg
coaddition and doubles in the jack. For B2 and Keck2012 we had no jack to test
for uniform monopole. Now we do and it appears to pass.
> There is a surprising amount of correlation between BICEP2 and Keck2013
I am not sure I see what you mean? Looking at B(pure) apmap and flicking
between B2, Keck2012 and Keck2013 the most striking thing is that the pattern
for Keck2013 looks rather blank in the left half. Looking at the 1/varmap
there seems no reason to expect that...
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
**********************************************************************