Two weeks ago at AIMS, Martin Bucher from (University Paris Diderot 7) gave us a seminar on the future of CMB experiments. Martin began by reviewing CMB physics, explaining how the measured temperature data is reduced to give power spectra, and the physical mechanisms underlying the features we see in CMB spectra.
Martin then outlined how the addition of noise and sky cuts complicates the analysis; he also stressed that the development of good approximations has been key to CMB astronomy, as the exact data analysis would often be too computationally intensive.
The talk then moved on to polarization. Martin explained that the CMB is a (mildly) polarized source, and this polarization data contains complementary information that can be extracted by current and future polarization experiments. In particular, polarization can tell us something about tensor fluctuations produced during inflation.
Different correlations (TT, TE, EE, BB – E and B are the different components of the polarization field) have different amplitudes, and the BB correlator that would be evidence of tensor modes, is particularly small, and so hard to see even with the data from the PLANCK experiment (depending on one’s theoretical assumptions).
Although future experiments may try to reach out to large multipoles than those reached by PLANCK, Martin showed us that as Silk damping reduces the CMB power, other effects (e.g. from reionization) begin to dominate. Although, this will make it hard to extract the primordial spectrum, it will still be possible to do other interesting physics.
Unfortunately, time constraints prevented us from hearing about CMB detection technologies as well as CMB foregrounds. The former is thus in the slides below, but not discussed here.
Martin opened his discussion of foregrounds by showing us WMAP 7 year sky maps at five different frequencies, ranging from 23GHz to 100GHz, showing us that the CMB is much clearer at higher frequencies (though dust means things get worse again above 100GHz).
To produce clean maps experimenters use linear combinations of different frequency sky maps along with masks to remove things such as point sources. Martin explained that there are varius sources of foreground emissions: power galactic synchrotron, free-free emission and dust. The latter is particularly hard to study, especially with regards to polarization.
Martin concluded by summarizing the foreground removal techniques as falling into four types and stressing that none is a silver bullet and that many approaches are needed. These approaches are:
- Understanding and modelling the detailed physics
- Data analysis
- Blind analysis (e.g. independent component analysis)
- Bayesian modelling