[us-commits] [ehb54/ultrascan3] 98979b: SAXS extrap-to-c0: error-band Primus merge test (G...

[emre brookes]

  Branch: refs/heads/ehb54-issue-920
  Home:   https://github.com/ehb54/ultrascan3
  Commit: 98979b73c555a8fed8a702b0a16cdef323cb85f6
      https://github.com/ehb54/ultrascan3/commit/98979b73c555a8fed8a702b0a16cdef323cb85f6
  Author: ehb54 <brookes at uthscsa.edu>
  Date:   2026-07-12 (Sun, 12 Jul 2026)

  Changed paths:
    M us_somo/develop/src/us_hydrodyn_saxs_iqq_extrap_c0.cpp
    M us_somo/somo/doc/manual/somo/somo_saxs_extrap_c0.html

  Log Message:
  -----------
  SAXS extrap-to-c0: error-band Primus merge test (GCV-robust)

The CORMAP sign-run merge test compares the GCV-smoothed extrapolation
against the noisy reference; a small consistent offset (the residual bias
of a nearly-linear regularized slope at high q) produces long same-sign
runs even when the curves agree to within noise, pushing the merge point
much too high (q~0.38 on the asyn data, so Primus extrapolated ~80% of the
curve and left a visible kink at the transition).

Replace it with an error-band test: the merge is the first q from which a
trailing window of the extrapolation agrees with the reference within the
reference's own errors -- reduced chi^2 of (Iex-Iref)/sd_ref <= threshold
(default 4 ~= 2 sigma; gparam saxs_extrap_c0_merge_chi2). On the asyn data
this moves the merge to the knee of the interaction (q~0.18) and copies the
reference over the biased high-q plateau, shrinking the kink. CORMAP is
kept as a fallback when the reference has no error column.

Manual updated. Review-only; not compiled here.

Co-Authored-By: Claude Opus 4.8 <noreply at anthropic.com>



To unsubscribe from these emails, change your notification settings at https://github.com/ehb54/ultrascan3/settings/notifications