Next
Meeting in Toulouse 26 November 19999
Contributions:
Participants:
Dear Archeops Collaborators,
Here is some information about the upcoming Archeops Data Analysis
Meeting. In order to prepare, we have some questions. PLEASE respond to
the questions marked with '***' as soon as possible:
*** Who will be coming? We need to know the number of people who will
*** be here in order to make the preparations.
*** Any other suggestions?
*** If you have questions, please contact Xavier Desert (about the
*** content of the meeting) or Jacques Delabrouille (about College
de
*** France logistics).
=== Logistical Information ===
Date : Friday, October 22, 1999
Time : 10 AM
Location: College de France
First Floor
Conference Room, Building F
(See below for
directions)
Directions: We believe that most people know how to get to the College
de France. It is at the intersection of Rue Saint Jacques and Rue des Ecoles,
in the 5th arrondissement in Paris. Once there, go in through
the main entrance and tell the guards that you're here for an "Archeops"
meeting at "PCC" and they will contact us to bring you to the meeting.
For a map of the area and some other information (mostly for the non-French), see the web page
=== Outline of Agenda ===
Morning :
The goal of the morning session will be to present what people have done thus far.
*** Who would like to present something? In order to make the
*** schedule, we need to know how many speakers there will be. The
*** format will be very simple. We will try to organize things in the
*** following groups, but additions and changes are always
*** possible. If you are going to present something, please give us
a
*** sentence or two describing it, and tell us where you think it
*** belongs in the list below.
***
*** Each participant should prepare 2-3 transparents for a short
*** presentation (5-10 minutes) of routine(s) that (s)he makes available
*** to the Archeops collaboration. This should contain
*** - goal of the routine (or set of)
*** - language, and dependencies : e.g. C code need such_and_such library
*** - algorithm
*** - input-output
*** - one example
*** A simple postscript file could be sent to me just before the meeting
*** to be posted in the software repository and on the web
*** http://www-laog.obs.ujf-grenoble.fr/~desert/archeops/#PRESENT
Data Extraction and Documentation
ArchTOI
ArcheDiab
LogBook
Archiving (data, software, papers, etc.)
...
Instrument Characterization
Temperatures
Focal Plane
Telemetry
Filters
Pendulations
...
Data Characterization
Noise
Systematics
...
Data Reduction
Glitches
Filtering
Removing Drifts
...
Pointing
Bolometers
Gyros
Stellar Sensor
...
Afternoon:
The afternoon session will be an open discussion of what was presented in the Morning session, as well as a discussion of what needs to be done, and how it should be organized and scheduled and assigned to people. Possible topics will include:
Goals of First Iteration of Data Processing
Accurate Boresight Pointing
Focal Plane
Beam Shapes
Relative and Absolute Calibration
See suggestions for many tasks below
...
Format of Deliverables
Software Repository and Packaging
Documentation, Paper Preparations
Data, Filters, etc.
...
Data Analysis
Foregrounds
Likelihood Analyses
...
Schedule
What is the most critical task?
Will there be conflicts when others have to return to the lab
for
the next flight?
At the following meeting, we should have a first cut at maps
and
noise.
...
Possible Inclusion of new Collaborators
- Stompor, Borrill & Balbi, suggested by Shaul Hanany from
U.
Minnesota
- Rome people (Vittorio & DeGasperis suggested by Paolo
de Bernardis)
on map projection softwares and C_l estimates.
*****
Ken Ganga, Jacques Delabrouille, Francois Bouchet,
&
F.-Xavier D\'esert
Laboratoire d'Astrophysique, Observatoire de Grenoble, BP 53,
414 rue de la piscine, 38041 Grenoble Cedex 9 FRANCE
Tel: (+33) 4 7663 5512, Fax : (+33) 4
7644 8821
email: Francois-Xavier.Desert@obs.ujf-grenoble.fr
Internet: http://www-laog.obs.ujf-grenoble.fr/~desert
Some Thoughts On The Many Tasks We Need To Do
=============================================
Document and Archive Raw Data
To Tape
To Disk? On the Web?
Obtain Flight Log
PRODUCTS: ArchTOI
Data Tapes
Web-Based Log Book
Extract Raw Data to something easily used (called it "Extracted Data")
Extract Data
Store extracted data on easily accessed disk
Document and Archive Extracted Data
Document and Archive Data Extraction Code (RCS? CVS?)
Characterize Time Stream (Extracted) Data
Visually Inspect Data
Make plots
Flag obviously bad data
Log and Flag Obviously Bad Data
How much is flagged?
Find Explanations for Obviously Bad
Data
(is it in the Log?)
Statistically Inspect Data
Power Spectra
PRODUCT: Estimate of noise level, 1/f knees,
etc. as a function of
time.
Log and Flag statistically bad data
How much is flagged?
Find Explanations for Statistically Bad Data
(is it in the Log?)
Inspect Data in various Coordinates
In Time
Find Missing Data and Flag
it (add new, flagged points?)
Experimental variations?
Changes in Responsivity?
Changes in Noise?
In Azimuth
Earth Shine?
Sun?
In Elevation
Pendulations?
Characterize Time-Constants/Filters
From Planets
From Point Sources?
From First Principles?
Document and Archive Inspection Code and Flags
Reconstruct Pointing
With Galaxy
PRODUCTS: Rough azimuth estimates (good to 1/2 deg.) with
associated sample numbers
Rough estimate of the elevation, assumed to be constant
With Planets
PRODUCTS: Focal Plane Map, with offsets in both elevation
and azimuth
Beam Maps
With Star Sensor
Make Beam Maps/Focal Plane Map
From Time Stream with Jupiter
Compare with Saturn in Time Stream
From Maps with Jupiter
Compare with Saturn in Maps
Other Sources/Galaxy?
Reduce Time Stream Data
Flag Cosmic Rays
By Hand
By Computer
Flag Planet Sightings
Calibrate
Remove Atmosphere and Drifts
Correlate with high-frequency channels
Use FFT or some other filter?
Using cross-linking (best done iteratively with
Map-making)
Deconvolve Time Constants
Document and Archive Time Stream Code
Map Making
Using simple binning
Make error estimates
Using complicated, "optimal" matrix methods
Make noise covariance matrix
Supercomputers?
MADCAP
Best done iteratively with time-stream reduction
"Jack-knife" data
In time
Analyze Maps
Investigate Noise properties
Histogram of noise
Check covariances
Check Gaussianity
Correlate with foregrounds
Dipole
Calibrate
Cross-check with planets
Dust
Syncrotron
H2?
Point sources?
Vary galactic cuts and make custom cuts
Characterize CMB
Make correlation function.
Make power spectrum with FFT methods
Make power spectrum with likelihood methods
Account for dipole/foreground removal
Supercomputers?
MADCAP
Fit power spectrum to CMB power spectra
Fit data directly to CMB power spectra
"Jack-knife" data
In time
On Sky
In Azimuth