Soumen ran NILC with 4 lensing amplitudes and 6 foreground models. He finds the following consistent results:
For 90.99 the residual foreground is high across all \ells.
For all models with the exception of 90.99 NILC reproduces the input CMB when A_l=1.
For all models with the exception of 90.99 NILC reproduces the input CMB for \ell<100 when A_L=0.003.
For all models with NILC fails to reproduce the input CMB for \ell>100 when A_L is different than 1.
Why?
Soumen: perhaps high \ell discrepancy is due to point sources?
SH: I don't think point source have been included in the maps, but not sure. To check with Clem.
SH suggests that the disagreement at high \ell may not be 'foreground' related. Perhaps it is something to do with the way we do 'delensing'. Perhaps run NILC on CMB skies with no foregrounds, but do have successive levels of delensing.
Touch base with Clem.
Touch base with Graca to check on status of her work.