Re: [Bicep2-list] [Bispud-pipeline-list] [keckarray] coaddtype for keck sims -- Q for Chin Lin or Sarah
22 Jul
2013
22 Jul
'13
10:28 a.m.
Sounds like you are talking specifically about aboffset.
I think this is likely to pass, but I strongly agree it would be
powerful to be able to make conclusive statements, e.g.
1. "the amplitude of the observed signal does not correlate with the
size of the ab offset," or
2. "the observed signal is common to the best and worst half of the
data split by ab offset."
It is true we have more diversity now in K2013, and I have asked Sarah
to start preparing for 2013 Keck sims, motivated by the considerable
statistical power we can expect from forming cross spectra with
offset-fixed 2013 Rx4 x B2. But it seemed clear from our discussion
last week that this is at least a few weeks out. And unlike the 2012
per receiver coadds which I think are already clearly within the scope
of the current analysis, I think there may be a valid argument that we
won't want to invoke 2013 data at least for what we aim to publish
immediately.
So,
--> Chin Lin or Sarah: can either of you easily make plots by tomorrow
showing (a) the 6 distributions and means of aboffset for each of the rx
of K2012 and B2. And then (b) show the same thing but with each
distribution split into a best half and worst half?
(If you have estimates of per-pair weights in the final coadds, it would
be preferable to form the means as weighted means, and to form the
best/worst split assigning equal weight. I worry that simply splitting
in half by number may yield significantly unequal weights.)
The idea is that (a) will tell us how strong a statement of type 1 we
could make just given the per-receiver coadds we are already planning,
and (b) would inform the possibility of forming a new channel jack
within the current dataset, allowing statements of both type 1 and 2.
--> Can people suggest alternative specific lines of attack on this point?
John
p.s. Once again, I will remind people once again that cc'ing keckarray
and bispud is
On 7/22/13 12:47 AM, Jamie Bock wrote:
Hi John,
I was thinking specifically this would be better to do across Keck and
BICEP2 to get more diversity in parameters (which is why I mentioned 2013).
The question is how we do this split exactly, by pixels, by tile, by
receiver. Might be good to look at e.g. dipole amplitude for choices of
grouping. I expect we will pass this test, and it would be nice to show
that.
Jamie
-----Original Message-----
From: John Kovac [mailto:jmkovac@cfa.harvard.edu]
Sent: Sunday, July 21, 2013 4:04 PM
To: Jamie Bock
Cc: 'Sarah Stokes Kernasovskiy'; 'bispud-pipeline-list';
'keckarray'
Subject: Re: [Bispud-pipeline-list] [keckarray] coaddtype for keck sims
This "best/worst beam match" jack sounds like a good idea, Jamie, but a
challenge in forming it for BICEP2 is how to define these groups.
Of the beam parameters we usually report, the pointing mismatch is the
dominant effect. We believe this specific effect deprojects nearly
perfectly, but if we doubted that, or wondered about other correlated
systematics, or just want to demonstrate control, the problem in forming a
jack is that its amplitude is actually fairly uniform across B2's focal
plane--unlike for Keck, there is no "low pointing offset" subset of B2
pairs. I guess we could still try to define a split this way, but I bet the
median pointing offset in the two groups would not differ by much.
More probing might be a split based on the amplitude of the undeprojected
residual for each pair, as characterized by the high S/N composite beam maps
Chin Lin is making right now. We will certainly not have this statistic in
time to form a jack for the current simset.
Other ideas are welcome.
John
On 7/21/13 6:16 PM, Jamie Bock wrote:
--
___________________________________________________________________
John Kovac jmkovac(a)cfa.harvard.edu
Assistant Professor, Astronomy and Physics, Harvard University
160 Concord Ave rm 310, Cambridge MA 02138, 617-496-0611
I think it would be good to do a jack test based
on a sorting of
beam-matching into best ~half vs. worst ~half of the detector pairs.
This is the obvious systematic we're worried about and another way to
demonstrate control. It could be done by receiver I suppose, though
the differences in the 2013 season are where this is most pronounced
by receiver, which we are not using right now.
Jamie
*From:*keckarray-bounces@mailman.stanford.edu
[mailto:keckarray-bounces@mailman.stanford.edu] *On Behalf Of *Sarah
Stokes Kernasovskiy
*Sent:* Sunday, July 21, 2013 9:46 AM
*To:* bispud-pipeline-list; keckarray
*Subject:* [keckarray] coaddtype for keck sims
Hi,
Stefan and I were discussing which coaddtypes to use for the Keck sims.
It seems like we probably want to coadd by receiver and well as
overall. This does complicate the process a little though - that
makes a lot of more maps/coadds.
One option we discussed is to discussed is to just coadd over receiver
- and then since we save the ac structure, we can further coadd that
over the array. This should be alright with the signflip noise sims
because the signs are flipped by scanset.
Anyone have any opinions as to weather we need the by-rx coadds?
Thanks!
-Sarah
_______________________________________________
Bispud-pipeline-list mailing list
Bispud-pipeline-list(a)lists.fas.harvard.edu
https://lists.fas.harvard.edu/mailman/listinfo/bispud-pipeline-list
--
___________________________________________________________________
John Kovac jmkovac(a)cfa.harvard.edu
Assistant Professor, Astronomy and Physics, Harvard University
160 Concord Ave rm 310, Cambridge MA 02138, 617-496-0611
22 Jul
22 Jul
11:43 a.m.
New subject: [Bispud-pipeline-list] [keckarray] coaddtype for keck sims -- Q for Chin Lin or Sarah
That sounds good. Do you think there may be coverage issues with the
(weighted) best/worst channel split? It would be better to do that of
course if we can. Not only will passing this test reinforce conclusions in
the upcoming paper it will help guide fabrication priorities this summer.
Jamie
-----Original Message-----
From: John Kovac [mailto:jmkovac@cfa.harvard.edu]
Sent: Monday, July 22, 2013 7:28 AM
To: Jamie Bock; 'Sarah Stokes Kernasovskiy'; Chin Lin Wong
Cc: bicep2-list; 'keckarray'
Subject: Re: [Bispud-pipeline-list] [keckarray] coaddtype for keck sims -- Q
for Chin Lin or Sarah
Sounds like you are talking specifically about aboffset.
I think this is likely to pass, but I strongly agree it would be powerful to
be able to make conclusive statements, e.g.
1. "the amplitude of the observed signal does not correlate with the size
of the ab offset," or
2. "the observed signal is common to the best and worst half of the data
split by ab offset."
It is true we have more diversity now in K2013, and I have asked Sarah to
start preparing for 2013 Keck sims, motivated by the considerable
statistical power we can expect from forming cross spectra with offset-fixed
2013 Rx4 x B2. But it seemed clear from our discussion last week that this
is at least a few weeks out. And unlike the 2012 per receiver coadds which
I think are already clearly within the scope of the current analysis, I
think there may be a valid argument that we won't want to invoke 2013 data
at least for what we aim to publish immediately.
So,
--> Chin Lin or Sarah: can either of you easily make plots by tomorrow
showing (a) the 6 distributions and means of aboffset for each of the rx of
K2012 and B2. And then (b) show the same thing but with each distribution
split into a best half and worst half?
(If you have estimates of per-pair weights in the final coadds, it would be
preferable to form the means as weighted means, and to form the best/worst
split assigning equal weight. I worry that simply splitting in half by
number may yield significantly unequal weights.)
The idea is that (a) will tell us how strong a statement of type 1 we could
make just given the per-receiver coadds we are already planning, and (b)
would inform the possibility of forming a new channel jack within the
current dataset, allowing statements of both type 1 and 2.
--> Can people suggest alternative specific lines of attack on this point?
John
p.s. Once again, I will remind people once again that cc'ing keckarray and
bispud is
On 7/22/13 12:47 AM, Jamie Bock wrote:
Hi John,
I was thinking specifically this would be better to do across Keck and
BICEP2 to get more diversity in parameters (which is why I mentioned
2013).
The question is how we do this split exactly, by
pixels, by tile, by
receiver. Might be good to look at e.g. dipole amplitude for choices of
grouping. I expect we will pass this test, and it would be nice to show
that.
Jamie
-----Original Message-----
From: John Kovac [mailto:jmkovac@cfa.harvard.edu]
Sent: Sunday, July 21, 2013 4:04 PM
To: Jamie Bock
Cc: 'Sarah Stokes Kernasovskiy'; 'bispud-pipeline-list';
'keckarray'
Subject: Re: [Bispud-pipeline-list] [keckarray] coaddtype for keck sims
This "best/worst beam match" jack sounds like a good idea, Jamie, but a
challenge in forming it for BICEP2 is how to define these groups.
Of the beam parameters we usually report, the pointing mismatch is the
dominant effect. We believe this specific effect deprojects nearly
perfectly, but if we doubted that, or wondered about other correlated
systematics, or just want to demonstrate control, the problem in forming a
jack is that its amplitude is actually fairly uniform across B2's focal
plane--unlike for Keck, there is no "low pointing offset" subset of B2
pairs. I guess we could still try to define a split this way, but I bet
the
median pointing offset in the two groups would not
differ by much.
More probing might be a split based on the amplitude of the undeprojected
residual for each pair, as characterized by the high S/N composite beam
maps
Chin Lin is making right now. We will certainly not
have this statistic
in
time to form a jack for the current simset.
Other ideas are welcome.
John
On 7/21/13 6:16 PM, Jamie Bock wrote:
--
___________________________________________________________________
John Kovac jmkovac(a)cfa.harvard.edu
Assistant Professor, Astronomy and Physics, Harvard University
160 Concord Ave rm 310, Cambridge MA 02138, 617-496-0611
I think it would be good to do a jack test based
on a sorting of
beam-matching into best ~half vs. worst ~half of the detector pairs.
This is the obvious systematic we're worried about and another way to
demonstrate control. It could be done by receiver I suppose, though
the differences in the 2013 season are where this is most pronounced
by receiver, which we are not using right now.
Jamie
*From:*keckarray-bounces@mailman.stanford.edu
[mailto:keckarray-bounces@mailman.stanford.edu] *On Behalf Of *Sarah
Stokes Kernasovskiy
*Sent:* Sunday, July 21, 2013 9:46 AM
*To:* bispud-pipeline-list; keckarray
*Subject:* [keckarray] coaddtype for keck sims
Hi,
Stefan and I were discussing which coaddtypes to use for the Keck sims.
It seems like we probably want to coadd by receiver and well as
overall. This does complicate the process a little though - that
makes a lot of more maps/coadds.
One option we discussed is to discussed is to just coadd over receiver
- and then since we save the ac structure, we can further coadd that
over the array. This should be alright with the signflip noise sims
because the signs are flipped by scanset.
Anyone have any opinions as to weather we need the by-rx coadds?
Thanks!
-Sarah
_______________________________________________
Bispud-pipeline-list mailing list
Bispud-pipeline-list(a)lists.fas.harvard.edu
https://lists.fas.harvard.edu/mailman/listinfo/bispud-pipeline-list
--
___________________________________________________________________
John Kovac jmkovac(a)cfa.harvard.edu
Assistant Professor, Astronomy and Physics, Harvard University
160 Concord Ave rm 310, Cambridge MA 02138, 617-496-0611
5:56 p.m.
New subject: [Bispud-pipeline-list] [keckarray] coaddtype for keck sims -- Q for Chin Lin or Sarah
Hi
A first pass at the plots requested can be found here:
http://bmode.caltech.edu/~spuder/analysis_logbook/analysis/20130722_AB/
Thanks!
Chin Lin
On Mon, Jul 22, 2013 at 11:43 AM, Jamie Bock <jjb(a)astro.caltech.edu> wrote:
That sounds good. Do you think there may be coverage issues with the
(weighted) best/worst channel split? It would be better to do that of
course if we can. Not only will passing this test reinforce conclusions in
the upcoming paper it will help guide fabrication priorities this summer.
Jamie
-----Original Message-----
From: John Kovac [mailto:jmkovac@cfa.harvard.edu]
Sent: Monday, July 22, 2013 7:28 AM
To: Jamie Bock; 'Sarah Stokes Kernasovskiy'; Chin Lin Wong
Cc: bicep2-list; 'keckarray'
Subject: Re: [Bispud-pipeline-list] [keckarray] coaddtype for keck sims --
Q
for Chin Lin or Sarah
Sounds like you are talking specifically about aboffset.
I think this is likely to pass, but I strongly agree it would be powerful
to
be able to make conclusive statements, e.g.
1. "the amplitude of the observed signal does not correlate with the
size
of the ab offset," or
2. "the observed signal is common to the best and worst half of the data
split by ab offset."
It is true we have more diversity now in K2013, and I have asked Sarah to
start preparing for 2013 Keck sims, motivated by the considerable
statistical power we can expect from forming cross spectra with
offset-fixed
2013 Rx4 x B2. But it seemed clear from our discussion last week that this
is at least a few weeks out. And unlike the 2012 per receiver coadds which
I think are already clearly within the scope of the current analysis, I
think there may be a valid argument that we won't want to invoke 2013 data
at least for what we aim to publish immediately.
So,
--> Chin Lin or Sarah: can either of you easily make plots by tomorrow
showing (a) the 6 distributions and means of aboffset for each of the rx of
K2012 and B2. And then (b) show the same thing but with each distribution
split into a best half and worst half?
(If you have estimates of per-pair weights in the final coadds, it would be
preferable to form the means as weighted means, and to form the best/worst
split assigning equal weight. I worry that simply splitting in half by
number may yield significantly unequal weights.)
The idea is that (a) will tell us how strong a statement of type 1 we could
make just given the per-receiver coadds we are already planning, and (b)
would inform the possibility of forming a new channel jack within the
current dataset, allowing statements of both type 1 and 2.
--> Can people suggest alternative specific lines of attack on this point?
John
p.s. Once again, I will remind people once again that cc'ing keckarray and
bispud is
On 7/22/13 12:47 AM, Jamie Bock wrote:
Hi John,
I was thinking specifically this would be better to do across Keck and
BICEP2 to get more diversity in parameters (which is why I mentioned
2013).
The question is how we do this split exactly, by
pixels, by tile, by
receiver. Might be good to look at e.g. dipole amplitude for choices of
grouping. I expect we will pass this test, and it would be nice to show
that.
Jamie
-----Original Message-----
From: John Kovac [mailto:jmkovac@cfa.harvard.edu]
Sent: Sunday, July 21, 2013 4:04 PM
To: Jamie Bock
Cc: 'Sarah Stokes Kernasovskiy'; 'bispud-pipeline-list';
'keckarray'
Subject: Re: [Bispud-pipeline-list] [keckarray] coaddtype for keck sims
This "best/worst beam match" jack sounds like a good idea, Jamie, but a
challenge in forming it for BICEP2 is how to define these groups.
Of the beam parameters we usually report, the pointing mismatch is the
dominant effect. We believe this specific effect deprojects nearly
perfectly, but if we doubted that, or wondered about other correlated
systematics, or just want to demonstrate control, the problem in forming
a
jack is that its amplitude is actually fairly
uniform across B2's focal
plane--unlike for Keck, there is no "low pointing offset" subset of B2
pairs. I guess we could still try to define a split this way, but I bet
the
median pointing offset in the two groups would
not differ by much.
More probing might be a split based on the amplitude of the undeprojected
residual for each pair, as characterized by the high S/N composite beam
maps
Chin Lin is making right now. We will certainly
not have this statistic
in
time to form a jack for the current simset.
Other ideas are welcome.
John
On 7/21/13 6:16 PM, Jamie Bock wrote:
--
___________________________________________________________________
John Kovac jmkovac(a)cfa.harvard.edu
Assistant Professor, Astronomy and Physics, Harvard University
160 Concord Ave rm 310, Cambridge MA 02138, 617-496-0611
I think it would be good to do a jack test based
on a sorting of
beam-matching into best ~half vs. worst ~half of the detector pairs.
This is the obvious systematic we're worried about and another way to
demonstrate control. It could be done by receiver I suppose, though
the differences in the 2013 season are where this is most pronounced
by receiver, which we are not using right now.
Jamie
*From:*keckarray-bounces@mailman.stanford.edu
[mailto:keckarray-bounces@mailman.stanford.edu] *On Behalf Of *Sarah
Stokes Kernasovskiy
*Sent:* Sunday, July 21, 2013 9:46 AM
*To:* bispud-pipeline-list; keckarray
*Subject:* [keckarray] coaddtype for keck sims
Hi,
Stefan and I were discussing which coaddtypes to use for the Keck sims.
It seems like we probably want to coadd by receiver and well as
overall. This does complicate the process a little though - that
makes a lot of more maps/coadds.
One option we discussed is to discussed is to just coadd over receiver
- and then since we save the ac structure, we can further coadd that
over the array. This should be alright with the signflip noise sims
because the signs are flipped by scanset.
Anyone have any opinions as to weather we need the by-rx coadds?
Thanks!
-Sarah
_______________________________________________
Bispud-pipeline-list mailing list
Bispud-pipeline-list(a)lists.fas.harvard.edu
https://lists.fas.harvard.edu/mailman/listinfo/bispud-pipeline-list
--
___________________________________________________________________
John Kovac jmkovac(a)cfa.harvard.edu
Assistant Professor, Astronomy and Physics, Harvard University
160 Concord Ave rm 310, Cambridge MA 02138, 617-496-0611
11:45 a.m.
New subject: [Bispud-pipeline-list] [keckarray] coaddtype for keck sims -- Q for Chin Lin or Sarah
Hi
I can try to make some plots today, probably with some help. I am not sure
if we have estimates of per-pair weights. Can someone point me?
Chin Lin
On Mon, Jul 22, 2013 at 10:28 AM, John Kovac <jmkovac(a)cfa.harvard.edu>wrote;wrote:
Sounds like you are talking specifically about aboffset.
I think this is likely to pass, but I strongly agree it would be powerful
to be able to make conclusive statements, e.g.
1. "the amplitude of the observed signal does not correlate with the
size of the ab offset," or
2. "the observed signal is common to the best and worst half of the data
split by ab offset."
It is true we have more diversity now in K2013, and I have asked Sarah to
start preparing for 2013 Keck sims, motivated by the considerable
statistical power we can expect from forming cross spectra with
offset-fixed 2013 Rx4 x B2. But it seemed clear from our discussion last
week that this is at least a few weeks out. And unlike the 2012 per
receiver coadds which I think are already clearly within the scope of the
current analysis, I think there may be a valid argument that we won't want
to invoke 2013 data at least for what we aim to publish immediately.
So,
--> Chin Lin or Sarah: can either of you easily make plots by tomorrow
showing (a) the 6 distributions and means of aboffset for each of the rx of
K2012 and B2. And then (b) show the same thing but with each distribution
split into a best half and worst half?
(If you have estimates of per-pair weights in the final coadds, it would
be preferable to form the means as weighted means, and to form the
best/worst split assigning equal weight. I worry that simply splitting in
half by number may yield significantly unequal weights.)
The idea is that (a) will tell us how strong a statement of type 1 we
could make just given the per-receiver coadds we are already planning, and
(b) would inform the possibility of forming a new channel jack within the
current dataset, allowing statements of both type 1 and 2.
--> Can people suggest alternative specific lines of attack on this point?
John
p.s. Once again, I will remind people once again that cc'ing keckarray and
bispud is
On 7/22/13 12:47 AM, Jamie Bock wrote:
Hi John,
I was thinking specifically this would be better to do across Keck and
BICEP2 to get more diversity in parameters (which is why I mentioned
2013).
The question is how we do this split exactly, by pixels, by tile, by
receiver. Might be good to look at e.g. dipole amplitude for choices of
grouping. I expect we will pass this test, and it would be nice to show
that.
Jamie
-----Original Message-----
From: John Kovac [mailto:jmkovac@cfa.harvard.**edu<jmkovac@cfa.harvard.edu>
]
Sent: Sunday, July 21, 2013 4:04 PM
To: Jamie Bock
Cc: 'Sarah Stokes Kernasovskiy'; 'bispud-pipeline-list';
'keckarray'
Subject: Re: [Bispud-pipeline-list] [keckarray] coaddtype for keck sims
This "best/worst beam match" jack sounds like a good idea, Jamie, but a
challenge in forming it for BICEP2 is how to define these groups.
Of the beam parameters we usually report, the pointing mismatch is the
dominant effect. We believe this specific effect deprojects nearly
perfectly, but if we doubted that, or wondered about other correlated
systematics, or just want to demonstrate control, the problem in forming a
jack is that its amplitude is actually fairly uniform across B2's focal
plane--unlike for Keck, there is no "low pointing offset" subset of B2
pairs. I guess we could still try to define a split this way, but I bet
the
median pointing offset in the two groups would not differ by much.
More probing might be a split based on the amplitude of the undeprojected
residual for each pair, as characterized by the high S/N composite beam
maps
Chin Lin is making right now. We will certainly not have this statistic
in
time to form a jack for the current simset.
Other ideas are welcome.
John
On 7/21/13 6:16 PM, Jamie Bock wrote:
--
______________________________**______________________________**_______
John Kovac jmkovac(a)cfa.harvard.edu
Assistant Professor, Astronomy and Physics, Harvard University
160 Concord Ave rm 310, Cambridge MA 02138, 617-496-0611
I think it would be good to do a jack test based
on a sorting of
beam-matching into best ~half vs. worst ~half of the detector pairs.
This is the obvious systematic we're worried about and another way to
demonstrate control. It could be done by receiver I suppose, though
the differences in the 2013 season are where this is most pronounced
by receiver, which we are not using right now.
Jamie
*From:*keckarray-bounces@**mailman.stanford.edu<keckarray-bounces@mailman.stanford.edu>
[mailto:keckarray-bounces@**mailman.stanford.edu<keckarray-bounces@mailman.stanford.edu>]
*On Behalf Of *Sarah
Stokes Kernasovskiy
*Sent:* Sunday, July 21, 2013 9:46 AM
*To:* bispud-pipeline-list; keckarray
*Subject:* [keckarray] coaddtype for keck sims
Hi,
Stefan and I were discussing which coaddtypes to use for the Keck sims.
It seems like we probably want to coadd by receiver and well as
overall. This does complicate the process a little though - that
makes a lot of more maps/coadds.
One option we discussed is to discussed is to just coadd over receiver
- and then since we save the ac structure, we can further coadd that
over the array. This should be alright with the signflip noise sims
because the signs are flipped by scanset.
Anyone have any opinions as to weather we need the by-rx coadds?
Thanks!
-Sarah
______________________________**_________________
Bispud-pipeline-list mailing list
Bispud-pipeline-list@lists.**fas.harvard.edu<Bispud-pipeline-list@lists.fas.harvard.edu>
https://lists.fas.harvard.edu/**mailman/listinfo/bispud-**pipeline-list<…
--
______________________________**______________________________**_______
John Kovac jmkovac(a)cfa.harvard.edu
Assistant Professor, Astronomy and Physics, Harvard University
160 Concord Ave rm 310, Cambridge MA 02138, 617-496-0611
11:48 a.m.
New subject: [Bispud-pipeline-list] [keckarray] coaddtype for keck sims -- Q for Chin Lin or Sarah
I can help you with the weights, Chin lin. They can be taken directly from
maps.
On Jul 22, 2013 8:45 AM, "Chin Lin Wong" <chinlinnet(a)gmail.com> wrote:
Hi
I can try to make some plots today, probably with some help. I am not sure
if we have estimates of per-pair weights. Can someone point me?
Chin Lin
On Mon, Jul 22, 2013 at 10:28 AM, John Kovac <jmkovac(a)cfa.harvard.edu>wrote;wrote:
Sounds like you are talking specifically about aboffset.
I think this is likely to pass, but I strongly agree it would be powerful
to be able to make conclusive statements, e.g.
1. "the amplitude of the observed signal does not correlate with the
size of the ab offset," or
2. "the observed signal is common to the best and worst half of the
data split by ab offset."
It is true we have more diversity now in K2013, and I have asked Sarah to
start preparing for 2013 Keck sims, motivated by the considerable
statistical power we can expect from forming cross spectra with
offset-fixed 2013 Rx4 x B2. But it seemed clear from our discussion last
week that this is at least a few weeks out. And unlike the 2012 per
receiver coadds which I think are already clearly within the scope of the
current analysis, I think there may be a valid argument that we won't want
to invoke 2013 data at least for what we aim to publish immediately.
So,
--> Chin Lin or Sarah: can either of you easily make plots by tomorrow
showing (a) the 6 distributions and means of aboffset for each of the rx of
K2012 and B2. And then (b) show the same thing but with each distribution
split into a best half and worst half?
(If you have estimates of per-pair weights in the final coadds, it would
be preferable to form the means as weighted means, and to form the
best/worst split assigning equal weight. I worry that simply splitting in
half by number may yield significantly unequal weights.)
The idea is that (a) will tell us how strong a statement of type 1 we
could make just given the per-receiver coadds we are already planning, and
(b) would inform the possibility of forming a new channel jack within the
current dataset, allowing statements of both type 1 and 2.
--> Can people suggest alternative specific lines of attack on this point?
John
p.s. Once again, I will remind people once again that cc'ing keckarray
and bispud is
On 7/22/13 12:47 AM, Jamie Bock wrote:
Hi John,
I was thinking specifically this would be better to do across Keck and
BICEP2 to get more diversity in parameters (which is why I mentioned
2013).
The question is how we do this split exactly, by pixels, by tile, by
receiver. Might be good to look at e.g. dipole amplitude for choices of
grouping. I expect we will pass this test, and it would be nice to show
that.
Jamie
-----Original Message-----
From: John Kovac [mailto:jmkovac@cfa.harvard.**edu<jmkovac@cfa.harvard.edu>
]
Sent: Sunday, July 21, 2013 4:04 PM
To: Jamie Bock
Cc: 'Sarah Stokes Kernasovskiy'; 'bispud-pipeline-list';
'keckarray'
Subject: Re: [Bispud-pipeline-list] [keckarray] coaddtype for keck sims
This "best/worst beam match" jack sounds like a good idea, Jamie, but a
challenge in forming it for BICEP2 is how to define these groups.
Of the beam parameters we usually report, the pointing mismatch is the
dominant effect. We believe this specific effect deprojects nearly
perfectly, but if we doubted that, or wondered about other correlated
systematics, or just want to demonstrate control, the problem in forming
a
jack is that its amplitude is actually fairly uniform across B2's focal
plane--unlike for Keck, there is no "low pointing offset" subset of B2
pairs. I guess we could still try to define a split this way, but I bet
the
median pointing offset in the two groups would not differ by much.
More probing might be a split based on the amplitude of the undeprojected
residual for each pair, as characterized by the high S/N composite beam
maps
Chin Lin is making right now. We will certainly not have this statistic
in
time to form a jack for the current simset.
Other ideas are welcome.
John
On 7/21/13 6:16 PM, Jamie Bock wrote:
--
______________________________**______________________________**_______
John Kovac jmkovac(a)cfa.harvard.edu
Assistant Professor, Astronomy and Physics, Harvard University
160 Concord Ave rm 310, Cambridge MA 02138, 617-496-0611
I think it would be good to do a jack test based
on a sorting of
beam-matching into best ~half vs. worst ~half of the detector pairs.
This is the obvious systematic we're worried about and another way to
demonstrate control. It could be done by receiver I suppose, though
the differences in the 2013 season are where this is most pronounced
by receiver, which we are not using right now.
Jamie
*From:*keckarray-bounces@**mailman.stanford.edu<keckarray-bounces@mailman.stanford.edu>
[mailto:keckarray-bounces@**mailman.stanford.edu<keckarray-bounces@mailman.stanford.edu>]
*On Behalf Of *Sarah
Stokes Kernasovskiy
*Sent:* Sunday, July 21, 2013 9:46 AM
*To:* bispud-pipeline-list; keckarray
*Subject:* [keckarray] coaddtype for keck sims
Hi,
Stefan and I were discussing which coaddtypes to use for the Keck sims.
It seems like we probably want to coadd by receiver and well as
overall. This does complicate the process a little though - that
makes a lot of more maps/coadds.
One option we discussed is to discussed is to just coadd over receiver
- and then since we save the ac structure, we can further coadd that
over the array. This should be alright with the signflip noise sims
because the signs are flipped by scanset.
Anyone have any opinions as to weather we need the by-rx coadds?
Thanks!
-Sarah
______________________________**_________________
Bispud-pipeline-list mailing list
Bispud-pipeline-list@lists.**fas.harvard.edu<Bispud-pipeline-list@lists.fas.harvard.edu>
https://lists.fas.harvard.edu/**mailman/listinfo/bispud-**pipeline-list<…
--
______________________________**______________________________**_______
John Kovac jmkovac(a)cfa.harvard.edu
Assistant Professor, Astronomy and Physics, Harvard University
160 Concord Ave rm 310, Cambridge MA 02138, 617-496-0611
3:55 p.m.
New subject: [Bispud-pipeline-list] [keckarray] coaddtype for keck sims -- Q for Chin Lin or Sarah
Hi Chris and Stefan
Does anyone have per pair coadds from which I can get the per-pair weights
for Bicep2? Sarah's pointed me to Keck ones, but I'm not sure where to look
for Bicep2 ones.
Thanks!
Chin Lin
On Mon, Jul 22, 2013 at 11:48 AM, Sarah Stokes Kernasovskiy <
sstokes(a)stanford.edu> wrote:
I can help you with the weights, Chin lin. They can
be taken directly
from maps.
On Jul 22, 2013 8:45 AM, "Chin Lin Wong" <chinlinnet(a)gmail.com> wrote:
> Hi
>
> I can try to make some plots today, probably with some help. I am not
> sure if we have estimates of per-pair weights. Can someone point me?
>
> Chin Lin
>
>
> On Mon, Jul 22, 2013 at 10:28 AM, John Kovac <jmkovac(a)cfa.harvard.edu>wrote;wrote:
>
>>
>> Sounds like you are talking specifically about aboffset.
>>
>> I think this is likely to pass, but I strongly agree it would be
>> powerful to be able to make conclusive statements, e.g.
>> 1. "the amplitude of the observed signal does not correlate with the
>> size of the ab offset," or
>> 2. "the observed signal is common to the best and worst half of the
>> data split by ab offset."
>>
>> It is true we have more diversity now in K2013, and I have asked Sarah
>> to start preparing for 2013 Keck sims, motivated by the considerable
>> statistical power we can expect from forming cross spectra with
>> offset-fixed 2013 Rx4 x B2. But it seemed clear from our discussion last
>> week that this is at least a few weeks out. And unlike the 2012 per
>> receiver coadds which I think are already clearly within the scope of the
>> current analysis, I think there may be a valid argument that we won't want
>> to invoke 2013 data at least for what we aim to publish immediately.
>>
>> So,
>>
>> --> Chin Lin or Sarah: can either of you easily make plots by tomorrow
>> showing (a) the 6 distributions and means of aboffset for each of the rx of
>> K2012 and B2. And then (b) show the same thing but with each distribution
>> split into a best half and worst half?
>>
>> (If you have estimates of per-pair weights in the final coadds, it would
>> be preferable to form the means as weighted means, and to form the
>> best/worst split assigning equal weight. I worry that simply splitting in
>> half by number may yield significantly unequal weights.)
>>
>> The idea is that (a) will tell us how strong a statement of type 1 we
>> could make just given the per-receiver coadds we are already planning, and
>> (b) would inform the possibility of forming a new channel jack within the
>> current dataset, allowing statements of both type 1 and 2.
>>
>> --> Can people suggest alternative specific lines of attack on this
>> point?
>>
>>
>> John
>>
>>
>> p.s. Once again, I will remind people once again that cc'ing keckarray
>> and bispud is
>>
>>
>>
>> On 7/22/13 12:47 AM, Jamie Bock wrote:
>>
>>> Hi John,
>>>
>>> I was thinking specifically this would be better to do across Keck and
>>> BICEP2 to get more diversity in parameters (which is why I mentioned
>>> 2013).
>>> The question is how we do this split exactly, by pixels, by tile, by
>>> receiver. Might be good to look at e.g. dipole amplitude for choices of
>>> grouping. I expect we will pass this test, and it would be nice to show
>>> that.
>>>
>>> Jamie
>>>
>>> -----Original Message-----
>>> From: John Kovac
[mailto:jmkovac@cfa.harvard.**edu<jmkovac@cfa.harvard.edu>
>>> ]
>>> Sent: Sunday, July 21, 2013 4:04 PM
>>> To: Jamie Bock
>>> Cc: 'Sarah Stokes Kernasovskiy'; 'bispud-pipeline-list';
'keckarray'
>>> Subject: Re: [Bispud-pipeline-list] [keckarray] coaddtype for keck sims
>>>
>>>
>>> This "best/worst beam match" jack sounds like a good idea, Jamie,
but a
>>> challenge in forming it for BICEP2 is how to define these groups.
>>>
>>> Of the beam parameters we usually report, the pointing mismatch is the
>>> dominant effect. We believe this specific effect deprojects nearly
>>> perfectly, but if we doubted that, or wondered about other correlated
>>> systematics, or just want to demonstrate control, the problem in
>>> forming a
>>> jack is that its amplitude is actually fairly uniform across B2's focal
>>> plane--unlike for Keck, there is no "low pointing offset" subset of
B2
>>> pairs. I guess we could still try to define a split this way, but I
>>> bet the
>>> median pointing offset in the two groups would not differ by much.
>>>
>>> More probing might be a split based on the amplitude of the
>>> undeprojected
>>> residual for each pair, as characterized by the high S/N composite beam
>>> maps
>>> Chin Lin is making right now. We will certainly not have this
>>> statistic in
>>> time to form a jack for the current simset.
>>>
>>> Other ideas are welcome.
>>>
>>> John
>>>
>>>
>>>
>>> On 7/21/13 6:16 PM, Jamie Bock wrote:
>>>
>>>> I think it would be good to do a jack test based on a sorting of
>>>> beam-matching into best ~half vs. worst ~half of the detector pairs.
>>>> This is the obvious systematic we're worried about and another way
to
>>>> demonstrate control. It could be done by receiver I suppose, though
>>>> the differences in the 2013 season are where this is most pronounced
>>>> by receiver, which we are not using right now.
>>>>
>>>> Jamie
>>>>
>>>>
*From:*keckarray-bounces@**mailman.stanford.edu<keckarray-bounces@mailman.stanford.edu>
>>>>
[mailto:keckarray-bounces@**mailman.stanford.edu<keckarray-bounces@mailman.stanford.edu>]
>>>> *On Behalf Of *Sarah
>>>> Stokes Kernasovskiy
>>>> *Sent:* Sunday, July 21, 2013 9:46 AM
>>>> *To:* bispud-pipeline-list; keckarray
>>>> *Subject:* [keckarray] coaddtype for keck sims
>>>>
>>>> Hi,
>>>>
>>>> Stefan and I were discussing which coaddtypes to use for the Keck sims.
>>>> It seems like we probably want to coadd by receiver and well as
>>>> overall. This does complicate the process a little though - that
>>>> makes a lot of more maps/coadds.
>>>>
>>>> One option we discussed is to discussed is to just coadd over receiver
>>>> - and then since we save the ac structure, we can further coadd that
>>>> over the array. This should be alright with the signflip noise sims
>>>> because the signs are flipped by scanset.
>>>>
>>>> Anyone have any opinions as to weather we need the by-rx coadds?
>>>>
>>>> Thanks!
>>>>
>>>> -Sarah
>>>>
>>>>
>>>>
>>>> ______________________________**_________________
>>>> Bispud-pipeline-list mailing list
>>>>
Bispud-pipeline-list@lists.**fas.harvard.edu<Bispud-pipeline-list@lists.fas.harvard.edu>
>>>> https://lists.fas.harvard.edu/**mailman/listinfo/bispud-**
>>>>
pipeline-list<https://lists.fas.harvard.edu/mailman/listinfo/bispud-pipe…
>>>>
>>>>
>>>
>>> --
>>> ______________________________**______________________________**_______
>>> John Kovac jmkovac(a)cfa.harvard.edu
>>>
>>> Assistant Professor, Astronomy and Physics, Harvard University
>>> 160 Concord Ave rm 310, Cambridge MA 02138, 617-496-0611
>>>
>>>
>>
>> --
>> ______________________________**______________________________**_______
>> John Kovac jmkovac(a)cfa.harvard.edu
>>
>> Assistant Professor, Astronomy and Physics, Harvard University
>> 160 Concord Ave rm 310, Cambridge MA 02138, 617-496-0611
>>
>
>
3953
days inactive
3953
days old
bicep2-list@lists.fas.harvard.edu
5 comments
4 participants
participants (4)
-
Chin Lin Wong -
Jamie Bock -
John Kovac -
Sarah Stokes Kernasovskiy