[us-commits] [ehb54/ultrascan3] f6be52: extrap_c0 doc: multi-discard, studentized residual...

[emre brookes]

  Branch: refs/heads/ehb54-issue-920
  Home:   https://github.com/ehb54/ultrascan3
  Commit: f6be52fa2c657a7ea9ddbc48c3e51d5f2fbf7268
      https://github.com/ehb54/ultrascan3/commit/f6be52fa2c657a7ea9ddbc48c3e51d5f2fbf7268
  Author: ehb54 <brookes at uthscsa.edu>
  Date:   2026-07-20 (Mon, 20 Jul 2026)

  Changed paths:
    M us_somo/somo/doc/manual/somo/somo_saxs_extrap_c0.html

  Log Message:
  -----------
  extrap_c0 doc: multi-discard, studentized residual, and its limitation

The manual was not updated alongside 246fd4b74 (multi-discard) and 3b3ffaec7
(leverage-corrected nomination), leaving two stale claims and two undocumented
directives.

- Outlier QC section: "at most one curve is ever discarded" now correctly scoped to
  the interactive dialog, with the scripted outlier_max limit and the reason raising
  it matters (a single pass judges each curve against a trend still containing the
  other offender). Ranking statistic corrected to the studentized residual. Output
  token generalized _qc1 -> _qcN, noting the red "x" marks the first discard while
  the log records all.
- New paragraph on WHY the residual is leverage-corrected: the highest-concentration
  curve is the extreme x, carries the most leverage, pulls the fit onto itself and
  masks its own residual (measured: leverage 0.80 vs 0.29 average moved a curve from
  5th to 2nd of 7). Var(r) = sigma^2 (1-h), so it is the standard statistic.
- New paragraph on the limitation: when one concentration sits far above the rest its
  leverage approaches 1, the fit passes through it and its residual is near zero
  however it is scaled -- no residual-based test can judge such a curve, and doing so
  needs information from outside the regression (e.g. MW from I(0) vs the known value).
- Scripting section: documented outlier_max and outlier_leverage. Date bumped.

Rendered and checked; the pre-existing bare-<p> imbalance is unchanged (delta 5 both
before and after), so no tag nesting was introduced.

Co-Authored-By: Claude Fable 5 <noreply at anthropic.com>



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