Branch: refs/heads/ehb54-issue-920 Home: https://github.com/ehb54/ultrascan3 Commit: 2c58b44034274c0cff1b5c5a19dd8db216b47a6a https://github.com/ehb54/ultrascan3/commit/2c58b44034274c0cff1b5c5a19dd8db21... Author: ehb54 <brookes@uthscsa.edu> Date: 2026-07-11 (Sat, 11 Jul 2026) Changed paths: M us_somo/develop/src/us_hydrodyn_saxs_iqq_extrap_c0.cpp M us_somo/develop/src/us_hydrodyn_saxs_iqq_extrap_c0_conc.cpp M us_somo/somo/doc/manual/somo/somo_saxs_extrap_c0.html Log Message: ----------- SAXS extrap-to-c0: Primus mode = weighted merge, not per-q extrapolation Primus mode previously did a per-q linear fit of the scaled intensity vs concentration and took the c=0 intercept. Once the curves are scaled onto the reference their residual concentration dependence is negligible, so that per-point fit mostly tracks noise -- it drove the high-q intensity negative (25 negative points on the asyn test set), and since the error column is the copied reference sd, those near-zero/negative points showed enormous error bars on a log plot. This is what a colleague reported as "Primus error bars too large vs ATSAS"; the sd column itself was already byte-identical to the ATSAS reference. Replace the per-q extrapolation with the inverse-variance weighted MERGE of the scaled curves (matching ATSAS almerge). On the asyn data this removes all 25 negative points (ATSAS has 0), eliminates all 37 huge-error-bar points, and matches the ATSAS zc_001.dat curve far better (median rel. diff low/mid/high-q 0.008/0.032/0.104 vs 0.026/0.133/0.470). The error column (reference sd), reference scaling, Conc tagging, and the I*(q) handling are unchanged. The regression pop-up now draws a level (slope 0) at the merged value. Zimm mode is untouched. Comments, editor message, dialog tooltip, and manual page updated. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> To unsubscribe from these emails, change your notification settings at https://github.com/ehb54/ultrascan3/settings/notifications