slides 2,3: completed 90.91 and 90.92, today discussing removing bands.
slides 5,6: model 90.92, foreground removal is 'better' than 90.91, but we are suspicious because according to Brandon PySM underestimates the variability across the sky. But for now we are proceeding with this model …
slide 7: MR removed both high and low frequencies → there is larger residual. It is factor of ~5 at the lowest \ell.
slide 12,13: Comparing these two slides we see that there is more bias with the restricted frequency range, but even so, it is an upper limit. SH and HK say that “removing the high/low frequency bands would not have led to the wrong conclusion”.
Also, this conclusion (“would not have led to bias”) is true for the PICO noise level. May not be true at other noise levels.
Action Item: Redo analysis removing only high or only low frequency bands.