osmose.calibration.uq.output_statsΒΆ

Reduce one OsmoseResults to a per-species UQ stat dict.

UQ-scoped: emits distinct {sp}_biomass_mean / {sp}_ssb_mean / {sp}_yield_mean keys (see keying.py); does not touch losses.quantity_key so NSGA/DE scoring is unaffected. Values are linear-scale trailing-window means β€” the natural-log and seed-mean transforms happen later in design.py (Phase 1). SSB extraction is net-new relative to scripts/calibrate_baltic.py (which computes only mean/yield/cv/trend).

Functions

compute_uq_stats(results, species_names[, ...])

Per-species linear-scale trailing-window means keyed for the UQ emulator.