R-forecasting: delensing discussion

From CMB-S4 wiki
Revision as of 18:12, 1 May 2017 by Sehgal (talk | contribs)
(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)
Jump to navigationJump to search

Guide for discussion (copied from Clem's email):

1) Enhance the Fisher matrix code which Victor has been using for band selection optimization to deal with lensing more realistically. No actual maps will be involved.

2) Take the high-res maps already being generated for S4 "data challenges" and reconstruct phi from them. These can then be used to generate a lensing template for use when these maps are being re-analyzed "BK style", and presumably in other styles of analysis.

Action items from S4 delensing and r call:

Fisher forecasting:

* Victor, Colin H., Mat, DW, Blake, Neelima

  • will iterate on values to try for numbers of detectors for high-res and low-res surveys;
  • will also iterate on which sky fractions to try;
  • Colin H. will make foreground-cleaned Nl^BB and Nl^EE curves and pass them to Mat/DW;
  • Mat/DW will generate AL curves and delensing efficiency curves and send them to Victor
  • Victor will run through the range of delensing curves for each of his detector count models and produce a grid from which we can identify which combinations reach the S4 sigma(r) target, and of those, which has the smallest detector usage; he may put the fisher part of his code on github

Map-level explorations:

* Alex, Colin B, Clem, Raphael - put high-res foreground sims on NERSC for all to use

* Kimmy, Yuuki, Alessandro, Kyle, Tom, Alex, DW, Julian C., Blake - run high-res foreground sims through delensing code to explore bias on r

* Kimmy, Yuuki, Alessandro, Kyle, Tom - explore effects of missing modes, edge effects, and small patches

* Julian C., Blake - implement map-based maximum likelihood code