# BK-style power spectra of 06b.00/07/09 with iterative lensing templates

— C. Pryke

One interesting thing to note is the correlation between the lensing template and the high and low frequency bands for models 07 and 09 at $$\ell>200$$ - see 270GHz and LT for extreme case. This appears to be associated with the step in the correlation between at $$\ell\sim200$$ which is visible in this post. In this post Julien says that he uses E-modes only for $$\ell<200$$ and both E and B above. Apparently the reconstructed LT contains a small amount of residual foreground for $$\ell>200$$.
The next step is ML searches to determine $$\sigma(r)$$ similar to those presented in 20200928_ML_dc06. Traditionally we use bandpowers 2 through 10 - shown in red below. Likely doing this will result in parameter biases due to the residual forground in the LT - which the fit model does not know about. Restricting to bandpowers 2-6 will presumably much reduce such problems but might affect $$\sigma(r)$$ - we should find out.