Xth School of Cosmology
  5 -10 July 2010 at IESC, Cargèse
 The Cosmic Microwave Background at High Angular Resolution

Component separation for CMB observations

Jean-Francois CARDOSO
Laboratoire traitement et communication de l'information (LTCI)

Course Chapters
  1. Multi-channel CMB observations
    1. The microwave sky
    2. Astrophysical foregrounds
    3. Foreground models
    4. The component separation problem
  2. Elements of optimal filtering
    1. Component separation as a filtering problem
    2. Pixel domain, harmonic domain, wavelet domain
    3. Best linear predictor, Wiener filter
    4. Non linear estimates for non Gaussian data
  3. Some component separation strategies
    1. Template fitting
    2. Wiener filters
    3. Independent component analysis
    4. Localization strategies
    5. Maximum likelihood methods
    6. Dealing with uncertainties
Abstract   
 After Component separation is a key step in processing CMB observations. Our instruments map the sky at several frequencies but none of these channels offers a pure view of the CMB: at all frequencies, astrophysical `foregrounds' (dust emission, synchrotron emission, SZ effect...) are superimposed on top of the Cosmic Background. Component separation is the art of taking advantage of the diversity offered by multi-frequency observations in order to recover the cleanest possible CMB map as well as maps of the other astrophysical foregrounds.

These lectures will briefly review the Physics of foregrounds and will focus on the theory and practice of component separation with emphasis on the statistical aspects of the problem.
Bibliography
  • Massiv
    J.

Program