Attendance: Clem Pryke, Amy, Shaul, Dan Green, Dave Chuss, Colin Hill, Jim B., Marcel S., Nick B., Tim P., Charles L., Alex E.
Notes by: Karl Young
A/I's given in bold.
Update on Signal Separation (aka - foregrounds) - Clem
currently sets of maps based sims on NERSC.
various (seven) foreground models
00 is just a toy model for sanity checks
3 Pysm models. has AME, dust, free-free, and synchtron (heavily based on Planck component maps)
96 MHD model from galactic magnetic field (simulated realization, not observed galaxy)
98 multilayer dust model. add synchtron from PySM.
99 mulitfrequency model from paper (see Clem's link)
Maps have either small r or no r added.
Multiple groups analyzing the final maps to reconstruct r and clean.
Progress has been slow.
Soumen using NILC
Belen using SMICA
Mateo using Commander
Stalled out. Spectral behavior varies across large pixel size needed for commander. Get residuals at PICO noise levels.
working on getting Oslo group to run Commander on this also. They have v2 of Commander.
Soumen made most progress:
-
slide 5: output power spectra. Titles tells how de-lensing is being faked. upper left is full lensing. upper right is .35, lower left is .15, lower right is .03.
SH: PICO should get to ~ 15% (lower left)
green is residual foregrounds.
00 model works good (trivial)
slide 6 (PySM 1, model 91). foreground still cleaned.
92, 93 also ok.
96, cut off at ell ~300 is becuase model has no foreground. cleaning does ok.
99 (slide 10). still foreground limited.
So less pessimistic foregrounds work with NILC method.
all this has been done on a single realization. Will be done with various CMB skies to test robustness.
for report. qualitative 'this works in most cases' Don't yet have full many realization statistics.
SH: how do we show all the work for interested parties? Beyond inclusion in the PICO report. Can we post an 'Appendix' someplace and link to it? Can't have an actual appendix in report.
CP: Would be good. S4 CMB has ~ 10 page appendix.
SH: We'll consider this. Needs to be prepped in ~ 1-2 month timescale. Could be a white paper or just appendix posted someplace.
Recommendations for external reviewers (deliver report end of Oct.)