Commander slides --22 Nov 2017, Mathieu Remazeilles

Mathieu shows results after removal of two lowest and two highest frequency bands, and also further reduction of frequency bands. Bias in r increases, as well as sigmar. The distribution of parameters suggests that the issue is mostly with mis-estimating the dust temperature, so removal of the higher frequencies not the lower.

One way to check that is to repeat but only removing the lower frequencies. We should see an increase in sigmar, but no increase in bias.

We discuss the CBASS data. With simple analysis (not Commander) Mathieu shows a reduction of 30% in error on CMB polarization by using CBASS data (in which configuration?). Perhaps we can eliminate the 21, 25 GHz bands and rely on CBASS for PICO? Need to check the improvement in sensitivity by increasing the number of detectors at other frequency bands, and need to run the full commander simulation with CBASS data.