Step 0 is sufficiently instrument/experiment
specific that general analysis tools are often of limited value. Steps
1-3
derive intrinsic statistical properties of the observation data itself.
Step 4 then relates these to the
parameters of a particular class of cosmological models, such as
inflation. MADCAP3 addresses steps 1 - 3 with 3 tools: - MADnes is an iterative timestream noise estimation code developed and maintained by Radek Stompor . - MADmap
is a PCG
map-making code developed and maintained by Chris Cantalupo . - MADspec
is a maximum-likelihood power spectrum estimation code developed and
maintained by Julian Borrill . The overall structure of the MADCAP codes (red boxes) and
data inputs and
outputs (blue boxes) is illustrated below:  MADnes estimates the inverse
time-time noise correlations from the
time-ordered data, pointing, and initial guess at a map; MADmap then
generates an improved map from the time-ordered data, pointing, and
these noise
correlations, and feeds this back to the next iteration of MADnes. When
this process has converged, the final map is inspected in various ways
to determine its acceptability.
Once an acceptable map has been generated, the MADspec noise correlator uses the time-ordered data, pointing, and time-time noise correlations to generate the pixel-pixel noise correlation matrix and map. Several such map/matrix pairs - either complete or after cutting - are then fed to the MADspec power spectra estimator to determine a user-defined combination of the maximum-likelihood T=<TT>, E=<EE>, B=<BB> spectra and the X=<TE>, Y=<TB> and Z=<EB> cross-spectra. Feedback & MADCAP Mailing List Please send comments, questions and requests
to be added
to the MADCAP mailing list to Julian
Borrill
. Acknowledgements MADCAP3 is a product of the COMBAT collaboration, which is supported by NASA's Advanced Information Systems Research Projects under Grant NAG5-3941, and was developed at NERSC, which is supported by the D.o.E's Office of Science under Contract No. DE-AC03-76SF00098. |