Found by the Copilot pass over Wave D PR #345 (2026-07-08); TODO(BUG) marker at src/underworld3/meshing/smoothing/metrics.py (commit 300bba4) with a fix sketch.
mesh_metric_mismatch builds A_target (and the clamp's A_mean) from rank-local sums, so under MPI the returned δ moments (rms/max/median_abs) are partition-dependent and inconsistent with the docstring's global equidistribution target. The live skip signal (alignment/misalignment used by follow_metric/mmpde) IS globally reduced and unaffected — this only mis-reports the diagnostic moments.
Fix: global reductions for the sums/counts feeding A_target/A_mean.
Underworld development team with AI support from Claude Code
Found by the Copilot pass over Wave D PR #345 (2026-07-08);
TODO(BUG)marker atsrc/underworld3/meshing/smoothing/metrics.py(commit 300bba4) with a fix sketch.mesh_metric_mismatchbuildsA_target(and the clamp'sA_mean) from rank-local sums, so under MPI the returned δ moments (rms/max/median_abs) are partition-dependent and inconsistent with the docstring's global equidistribution target. The live skip signal (alignment/misalignment used by follow_metric/mmpde) IS globally reduced and unaffected — this only mis-reports the diagnostic moments.Fix: global reductions for the sums/counts feeding A_target/A_mean.
Underworld development team with AI support from Claude Code