Differences

This shows you the differences between two versions of the page.

--- foregroundstelecon20220120 [2022/01/20 09:58] – created hanany
+++ foregroundstelecon20220120 [2022/01/20 11:52] (current) – hanany
@@ Line 4: / Line 4: @@
 [[https://spa-zzz-01.spa.umn.edu/!ipsig/logbook|Link to logbook]]
-Attendance: SH,  \\
+Attendance: SH, Kris, Charles, Julien, Sebastian \\
 Regrets:
@@ Line 15: / Line 15: @@
 === Notes: ===
+  * Julien/Sebastian have been in contact with MR. They now have access to MR's E-maps and they can re-run the delensing analysis. The expectation is that it will be essentially the same as with their own E-maps.
+  * Assuming the results are the same, what are the next steps?
+    -  MR needs to take the delensed maps and produce r estimates.
+    -  do delensing with other foregrounds skies, perhaps 90.92
+  * Hans Kristian sends an e-mail about commander progress and this figure {{::commander_20220120_comm_r003_8real.png?50|}}
+Unfortunately, neither Ragnhild, Ingunn or I will be able to attend today's telecon.
+However, I have been able to make some progress on the binned Blackwell-Rao estimator, and am currently analysing 8 simulations in parallel. The current situation is summarized in the attached plot, although this is preliminary, as some of the realizations only have about 300 samples, while we want 1000 before we're happy in terms of convergence. But the big picture is probably not going to change much.
+Overall, things look much more consistent now than two weeks ago. Each individual realization is consistent with the true input at no worse than 2 sigma. At the same time the combined distribution appears slightly biased positively with a PTE of about 1%. So we're not quite there -- but definitely closer than before. As already mentioned, we need more samples to claim convergence, and we are also missing two realizations, one of which was among the lowest power ones in Ragnhild's previous results. If we're lucky, and things shape up once those things are fixed, then this may finally be ticked off the TODO list. Otherwise, the next thing is to try expanding the mask a little more, in case that helps.

User Tools

Differences

Page Tools