Lessons Learned
from CMBX RR2



Cail Daley

CosmoStat, CEA Paris-Saclay


Euclid Weak Lensing Meeting, Marseille

October 3, 2025

SPT3G Funding Seals

RR2 Mask Products Overview

  • Comparing weighted counts in the catalog to coverage and visibility masks.
  • Largely similar, but there are regions with no counts that aren’t masked.
  • Track ISD weight convergence and lensing-source n(z) consistency alongside the mask products.

Source Counts (Weighted)

  • Applying she_lensmc_weight to build an effective number-density map:

Effective Coverage Mask

  • Not a subset of the weighted counts! Best to take the intersection?

Combined W Mask

  • Fold together coverage and visibility masks:

ISD Weight Iterations

  • Iterative spatial down-sampler (ISD) weights converge within six passes across tomographic bins; the residual wiggles flag masks that still need smoothing.
ISD weight iterations for tomographic bin 1
Bin 1
ISD weight iterations for tomographic bin 2
Bin 2
ISD weight iterations for tomographic bin 3
Bin 3
ISD weight iterations for tomographic bin 4
Bin 4
ISD weight iterations for tomographic bin 5
Bin 5
ISD weight iterations for tomographic bin 6
Bin 6

n(z) Comparisons (Unweighted)

  • Comparing RR2 unweighted tomographic redshift histograms against the Flagship mocks in both coarse and fine binning.

n(z) Comparisons (Weighted)

  • LensMC-weighted distributions smooth the mid-z mismatch but highlight a persistent low-z tilt that the ISD runs will need to address.