Walt,
It is disturbing that so many pairs that we are currently including in the CMB
analysis are unable to make sensible looking maps of the galaxy.
However the all pair maps look very good. Is it possible to look for the 0.6
degree overall pol angle difference we are finding in TB/EB by comparing these
to their Keck equivalents? This is just a random thought - I know it is not
what you are driving at right now.
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
**********************************************************************
It seems worth noting that for the cross-spectra TB/TE, the combos like
B2-cross = (B2(T) x B2(B)) - (B2(T) x Keck(B))
= B2(T) x (B2(B)-Keck(B))
are very nearly reproducing the "half-jacks" that we've constructed for
the latest BICEP1 results. The intention there is to form a null test
tailored to probe the consistency of the polarization modes (in which
shifts at the level of the noise can significantly affect the
bandpowers) that are specifically selected by crossing against the T
modes (which themselves are measured to such high S/N that shifts at the
noise level are irrelevant to bandpowers).
More generally, I think we can think of these "auto - cross" combos
similarly as tests of consistency of a specific subset of modes. For
example, for BB the test
B2-cross = B2(B) x (B2(B)-Keck(B))
asks whether the specific set of B modes contributing to B2's auto
spectrum is consistently seen in B2 and Keck to within noise.
John
On 9/17/13 6:20 PM, Clem Pryke wrote:
> OK - makes sense.
>
>> cross-Keck = (B2(T)-Keck(T)) x Keck(B)
>
> The fact that the cross-Keck TB shows higher fluctuation than the map based
> "experiment jack" *is* presumably due to non cancellation of the T signal due
> to differing masks.
>
> However in general you are right - these alternate spectral differences are
> useful for BB, but for most other spectra the map based "experiment jack" is
> clearly superior.
>
> Sorry for the noise,
>
> Clem
>
--
___________________________________________________________________
John Kovac jmkovac(a)cfa.harvard.edu
Associate Professor, Harvard University Astronomy Department
160 Concord Ave rm 310, Cambridge MA 02138, 617-496-0611
On 9/17/13 5:40 PM, Clem Pryke wrote:
> John,
>
> Can you explain why the sim fluctuation in TB (and EB) in my latest post is
> way smaller for B2xKeck-KeckxKeck than it is for B2xB2-B2xKeck?
You must be defining the cross TB as B2(T) x Keck(B).
In that case,
cross-Keck = (B2(T) x Keck(B)) - (Keck(T) x Keck(B))
= (B2(T)-Keck(T)) x Keck(B)
<<
B2-cross = (B2(T) x B2(B)) - (B2(T) x Keck(B))
= B2(T) x (B2(B)-Keck(B))
>
> I can't see how the "signal beating against noise" effect distinguishes between the two?
>
> Clem
>
--
___________________________________________________________________
John Kovac jmkovac(a)cfa.harvard.edu
Associate Professor, Harvard University Astronomy Department
160 Concord Ave rm 310, Cambridge MA 02138, 617-496-0611
On the telecon today I mentioned the QUIET-style jackknives that don't
have signal leakage. There are some more details on page 129 of my
thesis (http://arxiv.org/pdf/1209.1277v1). The definition is Equation
(5.49). I think the main difference from what we are doing now is to use
different suppression factors for each half of the jack and the cross.
The expectation value of this null estimator is independent of the
signal spectrum, but the variance can still get contributions from
sample variance. I think the contribution will be smaller than
estimators that have signal contribution and have to debias in the mean.
Also, I think I said something on the telecon that was incorrect: this
method doesn't use different masks for the two halves; it helps correct
for coverage/scanning differences between the two halves. If people are
interested in more details, I can ask Kendrick if I can circulate the
memo he wrote on this subject.
Immanuel Buder
Postdoctoral Fellow
Harvard-Smithsonian Center for Astrophysics
(office) 160 Concord Ave., M-114C
(office) 617 495 7567
(mail) 60 Garden St. MS 42
Cambridge, MA 02138
ibuder(a)cfa.harvard.edu
One more posting:
http://bicep0.caltech.edu/~spuder/analysis_logbook/analysis/20130917_satcom…
This one just shows that BICEP2 and Keck Array both see the signal, and
they see the turn-on / turn-off events at the same times. This proves
it's external to the experiment, and suggests the possibility that this
will cause correlations between the two experiments.
- Walt
Hi everyone,
I've put up a new posting in which I try to confirm that the az-fixed
signal we're seeing is the carrier of the ground station transmitting
to GOES. I do this by comparing the "oddball" tags in our D phases to
the USAP reports, and by seeing whether the signal goes away during
reported multi-day GOES and TDRS Ku-band outages. In both cases, I
don't find any connection. Therefore, I cannot confirm that we're
simply seeing the GOES transmission.
http://bicep0.caltech.edu/~spuder/analysis_logbook/analysis/20130916_satcom…
- Walt
I have a new posting on per-detector abscal. The only differences from
the last version (in late 2012) are:
- use 3yr data set
- use Planck input maps
- no longer including pixel window function in input maps
There's nothing surprising here and we should not need to discuss it
during the telecon.
http://bicep0.caltech.edu/~spuder/analysis_logbook/analysis/20130916_abscal…
Hi Walt,
This is super interesting stuff. Let's all remember that so long as the signal
is az fixed and not evolving in time then it goes out completely in the ground
subtraction. So it is the on/offs that matter. As you mention a large enough
contribution from these will cause failure of the az jack since GOES is not
operating during BEH phases. However given the greater noise in jacks there is
a grey area where the jack can pass but there still be contamination in jack0
- and since we seem to be perennially unlucky perhaps we are in this regime.
This is exactly the sort of thing that could cause partial correlation between
B2 and Keck of the type we appear to see. I am not sure that we would expect
B2xB2 > B2xKeck > KeckxKeck but perhaps.
It seems like something we immediately want to do is prepare a list of the
worst pairs. Since these are a small fraction of the total pairs we should
probably spin a new daughter coadd excluding these and then ask if the shift
versus current is consistent with expectation from sims - or clearly greater.
Are you in a position to provide such a list yet? I guess maybe for B2, but
not even started for Keck2012?
Best,
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 Walt,
I am not at all surprised there are oddballs. I expect what we are seeing is
carrier wave power from the satcom uplink - i.e. simply when the transmitter
is switched on - but not necessarily doing anything. The comms techs may well
mess around sometimes.
For instance in schedules 180 and 182 perhaps the pass was missed due to
technical problems but then somebody switched on the transmitter after the
pass ended to debug? Robert might have routine email messages sent around on
base saying "sorry no GOES this evening" etc - might be worth asking although
he may delete such stuff.
I seriously doubt the comms techs keep proper logbooks detailing such activity
- that would be military grade record keeping. But again might be worth
asking. Would presumably be paper on site if it existed at all.
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
**********************************************************************
