Develop the standalone companion methods paper + codameter implementation for processing-choice uncertainty in dv/v: hierarchical estimator with a correlated-window data covariance (block bootstrap), quality-metrics-as-noise (not weights), lapse-time regime segmentation, resolution-as-epistemic, and coverage-validated credible intervals. The naive softmax(λJ) average is rejected.
Prospectus: docs/theory/companion_paper_bayesian_dvv_uq.md; theory + critique: docs/theory/coda_window_selection_metrics.md. NOTE: the real codameter package is the separate repo at ../codameter.
Develop the standalone companion methods paper + codameter implementation for processing-choice uncertainty in dv/v: hierarchical estimator with a correlated-window data covariance (block bootstrap), quality-metrics-as-noise (not weights), lapse-time regime segmentation, resolution-as-epistemic, and coverage-validated credible intervals. The naive softmax(λJ) average is rejected.
Prospectus:
docs/theory/companion_paper_bayesian_dvv_uq.md; theory + critique:docs/theory/coda_window_selection_metrics.md. NOTE: the realcodameterpackage is the separate repo at ../codameter.