Josquin's dropbox slides

Josquin has been optimizing the bands for the probe. The input is r=0. We noticed (pg. 3) that the lowest r achievable was 0.008 (=substantial bias) and sigma®=8e-4. Questions:

• why is the bias high?
• why is sigma® not lower, like for S4?

Josquin has been working on upgrading his code so that he can compare to S4 results. The goal is to make sure that his code works as expected (up to slide 8).

In the telecon of October 8 we decided that in parallel to reproducing the results of S4, we will also check the following (slide 9):

• Perhaps the bias is due to the high level of Foregrounds. Reduce analysis from 60% of sky to 20% of sky.
• Assume that the code can produce proper relative results. Remove the two highest/lowest frequency bands to check how necessary they are.
• Filter the foregrounds in \ell space. Reproduce r constraints with \ell>~20.

We didn't have a telecon on October 15. But Josquin reports that:

• removal of extreme frequency bands decreases the bias (slide 10)
• reducing to 20% (cleaner?) sky region reduces bias to 0.001 with slight increase in sigma® (slide 11)