Source Code and Validation Suite for Cross-Scale Residual Correlation Analysis

Python implementation of a model-agnostic cross-scale residual aggregation framework (F_cross_scale). The code evaluates whether structure persists in residuals across disparate astrophysical and cosmological regimes by applying scale-normalized weighting, beta-optimization, and adversarial null tests. Includes validation notebooks, robustness and permutation tests, weak-signal recovery experiments, and a full adversarial testing suite designed to verify correct failure modes under null, corrupted, and heavy-tailed noise conditions. Intended for methodological research, cross-domain residual auditing, and statistical proofing of model-subtracted datasets. This repository does not claim new physical results, but provides a reproducible framework for testing residual coherence across observational silos.