BUG: Backport zero-check guard in CumulativeGaussianCostFunction

[hjmjohnson]

Adapted from d7845fe on main. Adds numberOfElements == 0 check before the size-mismatch check in CalculateFitError to guard against division-by-zero.

On main this was a reorder of an existing compound condition; on release-5.4 the zero check was entirely absent.

Backport for #6051 (Tier 1).

[greptile-apps]

Greptile Summary

This backport adds a numberOfElements == 0 early-return guard in CalculateFitError (line 46) via short-circuit || evaluation, preventing a division-by-zero crash that was entirely absent in the release-5.4 branch. The fix is correctly ordered and the guard logic is sound.

Confidence Score: 5/5

Safe to merge; the zero-check guard is correct and no new defects are introduced.

The primary fix is correct — numberOfElements == 0 is checked before the mismatch check using short-circuit evaluation, fully preventing division-by-zero. The only finding is a pre-existing integer division issue (1 / numberOfElements) that is unrelated to this PR and does not block merge.

No files require special attention.

Important Files Changed

Filename	Overview
Modules/Numerics/Optimizers/src/itkCumulativeGaussianCostFunction.cxx	Adds numberOfElements == 0 guard via short-circuit `

Flowchart

%%{init: {'theme': 'neutral'}}%%
flowchart TD
    A["CalculateFitError(setTestArray)"] --> B["numberOfElements = m_OriginalDataArray.GetNumberOfElements()"]
    B --> C{"numberOfElements == 0?"}
    C -- "Yes (NEW guard)" --> D["return 1"]
    C -- "No" --> E{"numberOfElements != setTestArray->size?"}
    E -- "Yes" --> D
    E -- "No" --> F["Sum squared differences in loop"]
    F --> G["return sqrt((1/numberOfElements) * fitError)"]
    style C fill:#90ee90,stroke:#228B22
    style D fill:#ffcccc

Loading

Comments Outside Diff (1)

1. Modules/Numerics/Optimizers/src/itkCumulativeGaussianCostFunction.cxx, line 55 (link)

   Pre-existing integer division yields incorrect RMSE

   1 / numberOfElements performs integer division because both operands are integer types (int literal and unsigned int). For any numberOfElements > 1, this evaluates to 0, making sqrt(0 * fitError) == 0.0 — the RMSE is silently wrong for all non-trivial arrays. The PR correctly guards against numberOfElements == 0, but since this function is being modified, this is a good opportunity to fix the formula.

_{Reviews (1): Last reviewed commit: "BUG: Add zero-check guard in CumulativeG..." | Re-trigger Greptile}

[hjmjohnson]

greptile comments need to ensure that upstream/main has already fixed this, if not fix upstream/main and PR.

This is a backport from main -> release-5.4.

[thewtex]

@hjmjohnson thanks!