5 Feb
2013
5 Feb
'13
3:43 p.m.
Hi all,
Sorry that we got cut off and you had to listen to me and John debating things.
To reiterate:
- I do not disagree with John - we should approach the process of setting
upper limits on un-deprojected residuals in a systematic manner a la Takahashi
et al. But these are upper limits - in full blown sims there are often
additional cancellations (Cf Chris' inductive cross talk sims).
- However, based on what I am seeing in terms of:
- beam maps constaints on sidelobes
- degree to which false B-mode disappears in jackknifes
- full blown sims of uniform monopole and quadrupole sims from Chris and
Jamie T
- degree of sidelobe which is consistent with WMAP cross comparison
I am skeptical that there is room to produce a false B as big as we are seeing
without violating some of these. But I may be wrong!
I would like to hear a consensus summary from Jamie T and Chris: In which of
the existing sims so far have you managed to produce false B equal in
magnitude to the real data but also passing jackknifes? How absurd (or not)
are these scenarios?
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
**********************************************************************
6 Feb
6 Feb
2:47 p.m.
New subject: [Bispud-pipeline-list] BICEP2 Team meeting tomorrow (05 Feb 2013) noon Eastern
Hi Clem,
In my posting:
http://bicep.caltech.edu/~spuder/analysis_logbook/analysis/20130122_monopol…
a 10% differenetial monopole sidelobe produces enough BB signal to
explain what we see in the real BB and passes jacknives. However, it
appears a 10% differential sidelobe ruled out by Randol's posting. A
2-3% differential sidelobe may be reasonable, but probably doesn't
produce enough BB signal, although I have not done that sim. The 1%
differential sidelobe that I have run does not produce enough power.
I believe both the 1% and 10% pass jacknives. (There is an error in
my posting which makes it look like 10% didn't pass scan dir and FPU
jacknives, but I am confident when I fix my mistake, jacknives will
pass).
-Jamie
On Tue, Feb 5, 2013 at 12:43 PM, Clem Pryke <pryke(a)physics.umn.edu> wrote:
Hi all,
Sorry that we got cut off and you had to listen to me and John debating things.
To reiterate:
- I do not disagree with John - we should approach the process of setting
upper limits on un-deprojected residuals in a systematic manner a la Takahashi
et al. But these are upper limits - in full blown sims there are often
additional cancellations (Cf Chris' inductive cross talk sims).
- However, based on what I am seeing in terms of:
- beam maps constaints on sidelobes
- degree to which false B-mode disappears in jackknifes
- full blown sims of uniform monopole and quadrupole sims from Chris and
Jamie T
- degree of sidelobe which is consistent with WMAP cross comparison
I am skeptical that there is room to produce a false B as big as we are seeing
without violating some of these. But I may be wrong!
I would like to hear a consensus summary from Jamie T and Chris: In which of
the existing sims so far have you managed to produce false B equal in
magnitude to the real data but also passing jackknifes? How absurd (or not)
are these scenarios?
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
**********************************************************************
_______________________________________________
Bispud-pipeline-list mailing list
Bispud-pipeline-list(a)lists.fas.harvard.edu
https://lists.fas.harvard.edu/mailman/listinfo/bispud-pipeline-list
4:06 p.m.
New subject: [Bispud-pipeline-list] BICEP2 Team meeting tomorrow (05 Feb 2013) noon Eastern
In my posting:
http://bicep.caltech.edu/~spuder/analysis_logbook/analysis/20130122_monopol…
a 10% differenetial monopole sidelobe produces enough BB signal to
explain what we see in the real BB and passes jacknives. However, it
appears a 10% differential sidelobe ruled out by Randol's posting. A
2-3% differential sidelobe may be reasonable, but probably doesn't
produce enough BB signal, although I have not done that sim. The 1%
differential sidelobe that I have run does not produce enough power.
I believe both the 1% and 10% pass jacknives. (There is an error in
my posting which makes it look like 10% didn't pass scan dir and FPU
jacknives, but I am confident when I fix my mistake, jacknives will
pass).
Given this information, this seems like a tantalizing possibility to me.
I think we should search/constrain such polarized sidelobe more closely
in the observed differenced beam (Randol?) AND use the actual WMAP T map to
generate expected B and TB from such sidelobe (Jamie?).
There is an ongoing effort to clean TB in fourier space directly and
of course that should continue.
9:19 p.m.
New subject: [Bispud-pipeline-list] BICEP2 Team meeting tomorrow (05 Feb 2013) noon Eastern
Hi All,
I've gone ahead and tried to cover other sims and analyses I've done, in
addition to ones that have actually *caused* spurious BB. I've bolded items
that actually cause excess BB:
1)
http://bicep0.caltech.edu/~spuder/analysis_logbook/analysis/20130125_ellipt…
- comparison of B2 to WMAP temperature maps (thus far only in map space,
not in ell space) rules out few degree FWHM sidelobes with more than ~20%
the power of the main beam. *In fact, the WMAP/B2 comparison looks to even
*suggest* an unmodeled sidelobe.*
- *Given Randol's figure 7* (
http://bicep0.caltech.edu/~spuder/analysis_logbook/analysis/20130124_zemax_…)
*it looks like we *do* have a stacked beam that can be described well as a
Gaussian main beam plus a few degree FWHM, **few percent power sidelobe.*
2)
http://bicep0.caltech.edu/~spuder/analysis_logbook/analysis/20130125_ellipt…
- *A 3 degree FWHM elliptical sidelobe, with 10% ellipticity, 90 degrees
offset between A and B, and in which the ellipses are oriented at 45
degrees to the detector pol angles, induce false B at about the level we're
seeing.* (When the ellipse angle is coaligned with the detector pol angle,
it causes no noticeable spurious BB.)
- It does almost nothing to any of the BB jackknifes, as we'd expect for a
pure quadrupole.
- *However, even though it causes spurious BB, the TB pattern that results
looks inconsistent with what we observe.*
3)
http://bicep0.caltech.edu/~spuder/analysis_logbook/analysis/20130121_ellipt…
- Turning ground subtraction off impacts the real maps more than sims (i.e.
we're seeing some scan fixed pick up) but appears to be operating at an ell
range below where we're taking to be excess BB.
- Excess BB is probably not due to a failure of ground subtraction.
4)
http://bicep0.caltech.edu/~spuder/analysis_logbook/analysis/20130121_ellipt…
- Spurious BB is not a random fluctuation.
- Spurious BB is definitely not there in Keck.
- Spurious BB does not go away by deprojecting real data with a WMAP W-band
template (but does get a tad noisier.)
4)
http://bicep0.caltech.edu/~spuder/analysis_logbook/analysis/20121210_relgai…
- *Simulating a 5.0% random relgain mismatch causes BB at about the level
we're seeing.*
- *Simulating a 0.5% systematic relgain mismatch causes BB at about the
level we're seeing.*
*- The 0.5% systematic mismatch causes a TB pattern like what we observe.*
*- I might be crazy, but I see an X pattern in the real BB spectrum. I also
see an X pattern in the 0.5% systematic relgain sim.*
- *However, both these effects deproject cleanly. Deprojecting with a
noiseless template yields basically the same deprojected spectrum.*
- Simulating relgain mismatch that reproduces Walt's empirically derived
abscals yields no sprious BB, even without deprojecting.
- A 1 degree systematic pol angle error causes no BB.
5)
http://bicep0.caltech.edu/~spuder/analysis_logbook/analysis/20121217_xtalk_…
- Simulating 1.0% inductive crosstalk (even effectively doing so *before*
elnod relgain calibration, c.f.
http://bicep0.caltech.edu/~spuder/analysis_logbook/analysis/20130121_ellipt…)
yields no spurious BB, owing to strong natural cancellation.
- The correct explanation for that cancellation is here:
http://bicep0.caltech.edu/~spuder/analysis_logbook/analysis/20121228_xtalk_…
6)
http://bicep0.caltech.edu/~spuder/analysis_logbook/analysis/20121217_new_re…
- Deprojecting by scanset leaves discrete B modes that we're identifying as
spurious in place (while adding noise to the BB spectrum, of course.)
7)
http://bicep0.caltech.edu/~spuder/analysis_logbook/analysis/20121210_sim123…
- Adding beamwidth or ellipticity deprojection to A/B offset deprojection
does nothing.
8)
http://bicep0.caltech.edu/~spuder/analysis_logbook/analysis/20121204_sim123…
- Accounting for point sources in V-band that are not the same at 100 GHz
does not alter the BB spectrum, even if it does alter the dk jack.
9)
http://bicep0.caltech.edu/~spuder/analysis_logbook/analysis/20121125_new_te…
- Messing around with simulated template noise and assumed WMAP beam
profile does not really alter the simulated BB spectra.
-------------------------------------------------------
Jamie, I want to make sure I understand your posting
http://bicep.caltech.edu/~spuder/analysis_logbook/analysis/20130122_monopol…
.
Your maps pager shows:
- A simulated sidelobe, 4 degree FWHM, that has 10% the power of the main
beam. (You said you used two "sidelobe power fractions": 100% and 10%. It
looks like the pager you show has only the 10% case?)
- Am I correct to interpret the sidelobe percentage in your maps pager as
"differential sidelobe power," and that 10% is equivalent to simulating a
case where the A detector has no sidelobe and the B detector has a sidelobe
with 10% the power of the main beam? (Or maybe 5% and 15%?) That would be
maximally pessimistic, I guess, as there'd be no way to decrease the amount
of power in the stacked sidelobe while maintaining the spurious BB it
causes by increasing the differential.
You say your procedure was this:
Pairsum=pairsum*(sidelobe power fraction)
Pairdiff=pairdiff+pairsum*(differential sidelobe power fraction)
but I'm getting a little confused trying to connect this to actual beams.
Also, I'd say that your sims *do* cause a butterfly pattern in TB, just
sign flipped from what we see, and probably a little too strong as compared
to the BB that is produced.
-Chris
On Wed, Feb 6, 2013 at 3:06 PM, Chao-Lin Kuo <clkuo(a)stanford.edu> wrote:
--
**********************************************************************
Christopher Sheehy - Graduate Student - University of Chicago
University of Minnesota,
Room 220 Tate, 116 Church Street S.E. Minneapolis, MN, 55455
Tel: 612-625-1802 Fax: 612-624-4578 email: csheehy(a)uchicago.edu
**********************************************************************
In my posting:
http://bicep.caltech.edu/~spuder/analysis_logbook/analysis/20130122_monopol…
a 10% differenetial monopole sidelobe produces enough BB signal to
explain what we see in the real BB and passes jacknives. However, it
appears a 10% differential sidelobe ruled out by Randol's posting. A
2-3% differential sidelobe may be reasonable, but probably doesn't
produce enough BB signal, although I have not done that sim. The 1%
differential sidelobe that I have run does not produce enough power.
I believe both the 1% and 10% pass jacknives. (There is an error in
my posting which makes it look like 10% didn't pass scan dir and FPU
jacknives, but I am confident when I fix my mistake, jacknives will
pass).
Given this information, this seems like a tantalizing possibility to me.
I think we should search/constrain such polarized sidelobe more closely
in the observed differenced beam (Randol?) AND use the actual WMAP T map to
generate expected B and TB from such sidelobe (Jamie?).
There is an ongoing effort to clean TB in fourier space directly and
of course that should continue.
_______________________________________________
Bispud-pipeline-list mailing list
Bispud-pipeline-list(a)lists.fas.harvard.edu
https://lists.fas.harvard.edu/mailman/listinfo/bispud-pipeline-list
7 Feb
7 Feb
3:16 p.m.
New subject: [Bispud-pipeline-list] BICEP2 Team meeting tomorrow (05 Feb 2013) noon Eastern
Hi Chris,
- A simulated sidelobe, 4 degree FWHM, that has 10% the power of the main
--
**********************************************************************
Christopher Sheehy - Graduate Student - University of Chicago
University of Minnesota,
Room 220 Tate, 116 Church Street S.E. Minneapolis, MN, 55455
Tel: 612-625-1802 Fax: 612-624-4578 email: csheehy(a)uchicago.edu
**********************************************************************
beam. (You said you used two "sidelobe power
fractions": 100% and 10%. It
looks like the pager you show has only the 10% case?)
In the text I refer to both the pairsum side lobe power (100% and 10%) as
well as the differential sidelobe power (10% and 1%). In the pager, I
specify the amount of sidelobe power by the differential sidelobe power.
- Am I correct to interpret the sidelobe percentage in your maps pager as
"differential sidelobe power," and that 10%
is equivalent to simulating a
case where the A detector has no sidelobe and the B detector has a sidelobe
with 10% the power of the main beam? (Or maybe 5% and 15%?) That would be
maximally pessimistic, I guess, as there'd be no way to decrease the amount
of power in the stacked sidelobe while maintaining the spurious BB it
causes by increasing the differential.
Yes to all the above.
You say your procedure was this:
Pairsum=pairsum*(sidelobe power fraction)
Pairdiff=pairdiff+pairsum*(differential sidelobe power fraction)
but I'm getting a little confused trying to connect this to actual beams.
What I mean here is just that I add a sidelobe component to both pairsum
and pairdiff, since we are concerned with TB as well as BB. Is this
different from what you are doing?
Also, I'd say that your sims *do* cause a butterfly pattern in TB, just
sign flipped from what we see, and probably a little
too strong as compared
to the BB that is produced.
Yes, this may be true.
-Jamie
On Wed, Feb 6, 2013 at 6:19 PM, Chris Sheehy <csheehy(a)uchicago.edu> wrote:
Hi All,
I've gone ahead and tried to cover other sims and analyses I've done, in
addition to ones that have actually *caused* spurious BB. I've bolded
items
that actually cause excess BB:
1)
http://bicep0.caltech.edu/~spuder/analysis_logbook/analysis/20130125_ellipt…
- comparison of B2 to WMAP temperature maps (thus far
only in map space,
not
in ell space) rules out few degree FWHM sidelobes with
more than ~20% the
power of the main beam. In fact, the WMAP/B2 comparison looks to even
*suggest* an unmodeled sidelobe.
- Given Randol's figure 7
(
http://bicep0.caltech.edu/~spuder/analysis_logbook/analysis/20130124_zemax_…
)
it looks like we *do* have a stacked beam that can be
described well as a
Gaussian main beam plus a few degree FWHM, few percent power sidelobe.
2)
http://bicep0.caltech.edu/~spuder/analysis_logbook/analysis/20130125_ellipt…
- A 3 degree FWHM elliptical sidelobe, with 10%
ellipticity, 90 degrees
offset between A and B, and in which the ellipses are oriented at 45
degrees
to the detector pol angles, induce false B at about
the level we're
seeing.
(When the ellipse angle is coaligned with the detector
pol angle, it
causes
no noticeable spurious BB.)
- It does almost nothing to any of the BB jackknifes, as we'd expect for a
pure quadrupole.
- However, even though it causes spurious BB, the TB pattern that results
looks inconsistent with what we observe.
3)
http://bicep0.caltech.edu/~spuder/analysis_logbook/analysis/20130121_ellipt…
- Turning ground subtraction off impacts the real maps
more than sims
(i.e.
we're seeing some scan fixed pick up) but appears
to be operating at an
ell
range below where we're taking to be excess BB.
- Excess BB is probably not due to a failure of ground subtraction.
4)
http://bicep0.caltech.edu/~spuder/analysis_logbook/analysis/20130121_ellipt…
- Spurious BB is not a random fluctuation.
- Spurious BB is definitely not there in Keck.
- Spurious BB does not go away by deprojecting real data with a WMAP
W-band
template (but does get a tad noisier.)
4)
http://bicep0.caltech.edu/~spuder/analysis_logbook/analysis/20121210_relgai…
- Simulating a 5.0% random relgain mismatch causes BB
at about the level
we're seeing.
- Simulating a 0.5% systematic relgain mismatch causes BB at about the
level
we're seeing.
- The 0.5% systematic mismatch causes a TB pattern like what we observe.
- I might be crazy, but I see an X pattern in the real BB spectrum. I also
see an X pattern in the 0.5% systematic relgain sim.
- However, both these effects deproject cleanly. Deprojecting with a
noiseless template yields basically the same deprojected spectrum.
- Simulating relgain mismatch that reproduces Walt's empirically derived
abscals yields no sprious BB, even without deprojecting.
- A 1 degree systematic pol angle error causes no BB.
5)
http://bicep0.caltech.edu/~spuder/analysis_logbook/analysis/20121217_xtalk_…
- Simulating 1.0% inductive crosstalk (even
effectively doing so *before*
elnod relgain calibration, c.f.
http://bicep0.caltech.edu/~spuder/analysis_logbook/analysis/20130121_ellipt…
)
yields no spurious BB, owing to strong natural
cancellation.
- The correct explanation for that cancellation is here:
http://bicep0.caltech.edu/~spuder/analysis_logbook/analysis/20121228_xtalk_…
6)
http://bicep0.caltech.edu/~spuder/analysis_logbook/analysis/20121217_new_re…
- Deprojecting by scanset leaves discrete B modes that
we're identifying
as
spurious in place (while adding noise to the BB
spectrum, of course.)
7)
http://bicep0.caltech.edu/~spuder/analysis_logbook/analysis/20121210_sim123…
- Adding beamwidth or ellipticity deprojection to A/B
offset deprojection
does nothing.
8)
http://bicep0.caltech.edu/~spuder/analysis_logbook/analysis/20121204_sim123…
- Accounting for point sources in V-band that are not
the same at 100 GHz
does not alter the BB spectrum, even if it does alter the dk jack.
9)
http://bicep0.caltech.edu/~spuder/analysis_logbook/analysis/20121125_new_te…
- Messing around with simulated template noise and
assumed WMAP beam
profile
does not really alter the simulated BB spectra.
-------------------------------------------------------
Jamie, I want to make sure I understand your posting
http://bicep.caltech.edu/~spuder/analysis_logbook/analysis/20130122_monopol…
.
>
> Your maps pager shows:
> - A simulated sidelobe, 4 degree FWHM, that has 10% the power of the main
beam. (You said you used two "sidelobe power
fractions": 100% and 10%. It
looks like the pager you show has only the 10% case?)
> - Am I correct to
interpret the sidelobe percentage in your maps pager as
> "differential sidelobe power," and that 10% is equivalent to simulating a
> case where the A detector has no sidelobe and the B detector has a
sidelobe
with 10% the power of the main beam? (Or maybe 5% and
15%?) That would be
maximally pessimistic, I guess, as there'd be no way to decrease the
amount
of power in the stacked sidelobe while maintaining the
spurious BB it
causes
by increasing the differential.
You say your procedure was this:
Pairsum=pairsum*(sidelobe power fraction)
Pairdiff=pairdiff+pairsum*(differential sidelobe power fraction)
but I'm getting a little confused trying to connect this to actual beams.
Also, I'd say that your sims *do* cause a butterfly pattern in TB, just
sign
flipped from what we see, and probably a little too
strong as compared to
the BB that is produced.
-Chris
On Wed, Feb 6, 2013 at 3:06 PM, Chao-Lin Kuo <clkuo(a)stanford.edu> wrote:
>
>
> > In my posting:
> >
> >
http://bicep.caltech.edu/~spuder/analysis_logbook/analysis/20130122_monopol…
a 10% differenetial monopole sidelobe produces enough BB signal to
explain what we see in the real BB and passes jacknives. However, it
appears a 10% differential sidelobe ruled out by Randol's posting. A
2-3% differential sidelobe may be reasonable, but probably doesn't
produce enough BB signal, although I have not done that sim. The 1%
differential sidelobe that I have run does not produce enough power.
I believe both the 1% and 10% pass jacknives. (There is an error in
my posting which makes it look like 10% didn't pass scan dir and FPU
jacknives, but I am confident when I fix my mistake, jacknives will
pass).
Given this information, this seems like a tantalizing possibility to me.
I think we should search/constrain such polarized sidelobe more closely
in the observed differenced beam (Randol?) AND use the actual WMAP T map
to
generate expected B and TB from such sidelobe (Jamie?).
There is an ongoing effort to clean TB in fourier space directly and
of course that should continue.
_______________________________________________
Bispud-pipeline-list mailing list
Bispud-pipeline-list(a)lists.fas.harvard.edu
https://lists.fas.harvard.edu/mailman/listinfo/bispud-pipeline-list
3:25 p.m.
New subject: [Bispud-pipeline-list] BICEP2 Team meeting tomorrow (05 Feb 2013) noon Eastern
Hi Jamie,
Okay, thanks. That makes sense.
You say your procedure was this:
I think you said you're working in pairmap space, so you're not really
operating on pair sum and pair diffs, but the accumulated pairmaps, some of
which are formed form pair sum data and some of which are formed form pair
diff data. Right?
I'm doing this slightly differently. I'm actually using Stefan's machinery
for simulating differential beamwidth and differential ellipticity directly
in the TODs, using highly smoothed synfast maps as the inputs. I'm doing
these signal only sims with no polarization. Then I add a fraction of the
final coadded map to the regular signal+noise sims, and interpret that
fraction as the fraction of power in the sidelobe compared to the main
beam. This seemed most straightforward to me. However, if I wanted to
simulate beams with no differential width / ellipticity but with different
power fractions between A/B, I'd have to add things together in the pairmap
stage and then make final maps, more similar to what you're doing.
-Chris
On Thu, Feb 7, 2013 at 2:16 PM, jamie tolan <jetolan(a)stanford.edu> wrote:
> Hi Chris,
>
>
> - A simulated sidelobe, 4 degree FWHM, that has 10% the power of the main
>> beam. (You said you used two "sidelobe power fractions": 100% and 10%.
It
>> looks like the pager you show has only the 10% case?)
>
>
> In the text I refer to both the pairsum side lobe power (100% and 10%) as
> well as the differential sidelobe power (10% and 1%). In the pager, I
> specify the amount of sidelobe power by the differential sidelobe power.
>
> - Am I correct to interpret the sidelobe percentage in your maps pager as
>> "differential sidelobe power," and that 10% is equivalent to simulating
a
>> case where the A detector has no sidelobe and the B detector has a sidelobe
>> with 10% the power of the main beam? (Or maybe 5% and 15%?) That would be
>> maximally pessimistic, I guess, as there'd be no way to decrease the amount
>> of power in the stacked sidelobe while maintaining the spurious BB it
>> causes by increasing the differential.
>
>
> Yes to all the above.
>
> You say your procedure was this:
> Also, I'd say that your sims *do* cause a butterfly pattern in TB, just
>> sign flipped from what we see, and probably a little too strong as compared
>> to the BB that is produced.
>
>
> Yes, this may be true.
>
> -Jamie
>
> On Wed, Feb 6, 2013 at 6:19 PM, Chris Sheehy <csheehy(a)uchicago.edu> wrote:
> > Hi All,
> >
> > I've gone ahead and tried to cover other sims and analyses I've done,
in
> > addition to ones that have actually *caused* spurious BB. I've bolded
> items
> > that actually cause excess BB:
> >
> > 1)
> >
>
http://bicep0.caltech.edu/~spuder/analysis_logbook/analysis/20130125_ellipt…
> > - comparison of B2 to WMAP temperature maps (thus far only in map space,
> not
> > in ell space) rules out few degree FWHM sidelobes with more than ~20% the
> > power of the main beam. In fact, the WMAP/B2 comparison looks to even
> > *suggest* an unmodeled sidelobe.
> > - Given Randol's figure 7
> > (
> http://bicep0.caltech.edu/~spuder/analysis_logbook/analysis/20130124_zemax_…
> )
> > it looks like we *do* have a stacked beam that can be described well as a
> > Gaussian main beam plus a few degree FWHM, few percent power sidelobe.
> >
> > 2)
> >
>
http://bicep0.caltech.edu/~spuder/analysis_logbook/analysis/20130125_ellipt…
> > - A 3 degree FWHM elliptical sidelobe, with 10% ellipticity, 90 degrees
> > offset between A and B, and in which the ellipses are oriented at 45
> degrees
> > to the detector pol angles, induce false B at about the level we're
> seeing.
> > (When the ellipse angle is coaligned with the detector pol angle, it
> causes
> > no noticeable spurious BB.)
> > - It does almost nothing to any of the BB jackknifes, as we'd expect for
> a
> > pure quadrupole.
> > - However, even though it causes spurious BB, the TB pattern that results
> > looks inconsistent with what we observe.
> >
> > 3)
> >
>
http://bicep0.caltech.edu/~spuder/analysis_logbook/analysis/20130121_ellipt…
> > - Turning ground subtraction off impacts the real maps more than sims
> (i.e.
> > we're seeing some scan fixed pick up) but appears to be operating at an
> ell
> > range below where we're taking to be excess BB.
> > - Excess BB is probably not due to a failure of ground subtraction.
> >
> > 4)
> >
>
http://bicep0.caltech.edu/~spuder/analysis_logbook/analysis/20130121_ellipt…
> > - Spurious BB is not a random fluctuation.
> > - Spurious BB is definitely not there in Keck.
> > - Spurious BB does not go away by deprojecting real data with a WMAP
> W-band
> > template (but does get a tad noisier.)
> >
> > 4)
> >
> http://bicep0.caltech.edu/~spuder/analysis_logbook/analysis/20121210_relgai…
> > - Simulating a 5.0% random relgain mismatch causes BB at about the level
> > we're seeing.
> > - Simulating a 0.5% systematic relgain mismatch causes BB at about the
> level
> > we're seeing.
> > - The 0.5% systematic mismatch causes a TB pattern like what we observe.
> > - I might be crazy, but I see an X pattern in the real BB spectrum. I
> also
> > see an X pattern in the 0.5% systematic relgain sim.
> > - However, both these effects deproject cleanly. Deprojecting with a
> > noiseless template yields basically the same deprojected spectrum.
> > - Simulating relgain mismatch that reproduces Walt's empirically derived
> > abscals yields no sprious BB, even without deprojecting.
> > - A 1 degree systematic pol angle error causes no BB.
> >
> > 5)
> >
> http://bicep0.caltech.edu/~spuder/analysis_logbook/analysis/20121217_xtalk_…
> > - Simulating 1.0% inductive crosstalk (even effectively doing so *before*
> > elnod relgain calibration, c.f.
> >
>
http://bicep0.caltech.edu/~spuder/analysis_logbook/analysis/20130121_ellipt…
> )
> > yields no spurious BB, owing to strong natural cancellation.
> > - The correct explanation for that cancellation is here:
> >
> http://bicep0.caltech.edu/~spuder/analysis_logbook/analysis/20121228_xtalk_…
> >
> > 6)
> >
>
http://bicep0.caltech.edu/~spuder/analysis_logbook/analysis/20121217_new_re…
> > - Deprojecting by scanset leaves discrete B modes that we're identifying
> as
> > spurious in place (while adding noise to the BB spectrum, of course.)
> >
> > 7)
> >
>
http://bicep0.caltech.edu/~spuder/analysis_logbook/analysis/20121210_sim123…
> > - Adding beamwidth or ellipticity deprojection to A/B offset deprojection
> > does nothing.
> >
> > 8)
> >
>
http://bicep0.caltech.edu/~spuder/analysis_logbook/analysis/20121204_sim123…
> > - Accounting for point sources in V-band that are not the same at 100 GHz
> > does not alter the BB spectrum, even if it does alter the dk jack.
> >
> > 9)
> >
>
http://bicep0.caltech.edu/~spuder/analysis_logbook/analysis/20121125_new_te…
> > - Messing around with simulated template noise and assumed WMAP beam
> profile
> > does not really alter the simulated BB spectra.
> >
> > -------------------------------------------------------
> >
> >
> > Jamie, I want to make sure I understand your posting
> >
>
http://bicep.caltech.edu/~spuder/analysis_logbook/analysis/20130122_monopol…
> .
> >
> > Your maps pager shows:
> > - A simulated sidelobe, 4 degree FWHM, that has 10% the power of the main
> > beam. (You said you used two "sidelobe power fractions": 100% and 10%.
It
> > looks like the pager you show has only the 10% case?)
> > - Am I correct to interpret the sidelobe percentage in your maps pager as
> > "differential sidelobe power," and that 10% is equivalent to
simulating a
> > case where the A detector has no sidelobe and the B detector has a
> sidelobe
> > with 10% the power of the main beam? (Or maybe 5% and 15%?) That would be
> > maximally pessimistic, I guess, as there'd be no way to decrease the
> amount
> > of power in the stacked sidelobe while maintaining the spurious BB it
> causes
> > by increasing the differential.
> >
> > You say your procedure was this:
> >
> > Pairsum=pairsum*(sidelobe power fraction)
> > Pairdiff=pairdiff+pairsum*(differential sidelobe power fraction)
> >
> > but I'm getting a little confused trying to connect this to actual beams.
> > Also, I'd say that your sims *do* cause a butterfly pattern in TB, just
> sign
> > flipped from what we see, and probably a little too strong as compared to
> > the BB that is produced.
> >
> > -Chris
> >
> >
> >
> > On Wed, Feb 6, 2013 at 3:06 PM, Chao-Lin Kuo <clkuo(a)stanford.edu> wrote:
> >>
> >>
> >> > In my posting:
> >> >
> >> >
>
http://bicep.caltech.edu/~spuder/analysis_logbook/analysis/20130122_monopol…
> >> >
> >> > a 10% differenetial monopole sidelobe produces enough BB signal to
> >> > explain what we see in the real BB and passes jacknives. However, it
> >> > appears a 10% differential sidelobe ruled out by Randol's posting.
A
> >> > 2-3% differential sidelobe may be reasonable, but probably
doesn't
> >> > produce enough BB signal, although I have not done that sim. The 1%
> >> > differential sidelobe that I have run does not produce enough power.
> >> >
> >> > I believe both the 1% and 10% pass jacknives. (There is an error in
> >> > my posting which makes it look like 10% didn't pass scan dir and
FPU
> >> > jacknives, but I am confident when I fix my mistake, jacknives will
> >> > pass).
> >>
> >> Given this information, this seems like a tantalizing possibility to me.
> >> I think we should search/constrain such polarized sidelobe more closely
> >> in the observed differenced beam (Randol?) AND use the actual WMAP T map
> >> to
> >> generate expected B and TB from such sidelobe (Jamie?).
> >>
> >> There is an ongoing effort to clean TB in fourier space directly and
> >> of course that should continue.
> >>
> >> _______________________________________________
> >> Bispud-pipeline-list mailing list
> >> Bispud-pipeline-list(a)lists.fas.harvard.edu
> >> https://lists.fas.harvard.edu/mailman/listinfo/bispud-pipeline-list
> >
> >
> >
> >
> > --
> > **********************************************************************
> > Christopher Sheehy - Graduate Student - University of Chicago
> > University of Minnesota,
> > Room 220 Tate, 116 Church Street S.E. Minneapolis, MN, 55455
> > Tel: 612-625-1802 Fax: 612-624-4578 email: csheehy(a)uchicago.edu
> > **********************************************************************
>
--
**********************************************************************
Christopher Sheehy - Graduate Student - University of Chicago
University of Minnesota,
Room 220 Tate, 116 Church Street S.E. Minneapolis, MN, 55455
Tel: 612-625-1802 Fax: 612-624-4578 email: csheehy(a)uchicago.edu
**********************************************************************
Pairsum=pairsum*(sidelobe power fraction)
Pairdiff=pairdiff+pairsum*(differential sidelobe power fraction)
but I'm getting a little confused trying to connect this to actual beams.
What I mean here is just that I add a sidelobe component to both pairsum
and pairdiff, since we are concerned with TB as well as BB. Is this
different from what you are doing?
Pairsum=pairsum*(sidelobe power fraction)
Pairdiff=pairdiff+pairsum*(differential sidelobe power fraction)
but I'm getting a little confused trying to connect this to actual beams.
What I mean here is just that I add a sidelobe component to both pairsum
and pairdiff, since we are concerned with TB as well as BB. Is this
different from what you are doing?
4118
days inactive
4120
days old
bicep2-list@lists.fas.harvard.edu
5 comments
4 participants
participants (4)
-
Chao-Lin Kuo -
Chris Sheehy -
Clem Pryke -
jamie tolan