Fractal-Informational Seismic Regime Detection (QO3/FIO) v2.2: Declustering, KL-SID, Block Bootstrap, Calibration

QO3/FIO Framework v2.2 is a production-ready research software package for seismic regime detection (background vs elevated activity) using fractal–informational observables (FIO) and operator-spectral invariants (QO3). The system is designed for regime detection rather than deterministic earthquake prediction, with explicit anti-leakage and time-series validation controls.Core methodology: (i) causality-preserving Gardner–Knopoff declustering to eliminate aftershock leakage; (ii) bias-corrected Gutenberg–Richter b-value estimation (Tinti–Mulargia) with uncertainty; (iii) robust variability metrics (MAD-based CV); (iv) Seismic Information Deficit via discrete KL divergence (KL-SID); (v) moving block bootstrap confidence intervals for temporally correlated labels/scores; (vi) rolling/expanding time-series cross-validation; (vii) probability calibration metrics including Brier Skill Score (BSS), Expected Calibration Error (ECE), and reliability diagram data.Reproducibility: the archive includes source code, configuration defaults, and a LaTeX manuscript. All formal claims are explicitly categorized as Theorem / Proposition / Conjecture to maintain mathematical rigor