cosmo_val/cat_config.yaml defines SP_v1.3.6 twice (L148, L522) and SP_v1.4.6 twice (L436, L565). YAML keeps the last one, silently — no error, no warning.
The blocks are not copies. For SP_v1.4.6 the surviving block changes:
shear.R: 1.0 → 0.92 (an ~8% multiplicative shear rescaling)cov_th.A: 2894.03 → 2405.39 and cov_th.n_e: 5.093 → 6.128 (analytic covariance amplitude)- n(z):
redshift_distr: .../nz/v1.4.6/nz_SP_v1.4.6_A.txt → redshift_path: .../CFIS/v1.0/nz/dndz_SP_A.txt (the 2023 pre-SOM distribution)
- mask →
mask_map_v1.4.6_nside_8192.fits
For SP_v1.3.6 the surviving block swaps the mask from the 4096 footprint to the v1.4.6 8192 map, plus plot colour/marker.
So anyone running either version gets the second block, with no symptom other than wrong numbers. Worth a deduplication pass, and a duplicate-key check at config load so it can't recur silently. Adjacent to #294.
cosmo_val/cat_config.yamldefinesSP_v1.3.6twice (L148, L522) andSP_v1.4.6twice (L436, L565). YAML keeps the last one, silently — no error, no warning.The blocks are not copies. For
SP_v1.4.6the surviving block changes:shear.R: 1.0 → 0.92 (an ~8% multiplicative shear rescaling)cov_th.A: 2894.03 → 2405.39 andcov_th.n_e: 5.093 → 6.128 (analytic covariance amplitude)redshift_distr: .../nz/v1.4.6/nz_SP_v1.4.6_A.txt→redshift_path: .../CFIS/v1.0/nz/dndz_SP_A.txt(the 2023 pre-SOM distribution)mask_map_v1.4.6_nside_8192.fitsFor
SP_v1.3.6the surviving block swaps the mask from the 4096 footprint to the v1.4.6 8192 map, plus plot colour/marker.So anyone running either version gets the second block, with no symptom other than wrong numbers. Worth a deduplication pass, and a duplicate-key check at config load so it can't recur silently. Adjacent to #294.