COS–NUM is the Python-based numerical environment and reproducibility layer of the Collapsing-Structure (COS) program. It provides reference numerical representations, simulation tools, data-processing chains, and audit conventions for the COS modules, including COS-S, COS-QF, COS-QSF, COS-QD, COS-SM, COS-SUSY, COS-GUT, and COS-C. The module documents how quantized shell–filament graphs, discrete labels, operator matrices, spectra, collapse/update rules, and topology-related diagnostics can be represented and studied computationally. The emphasis is on executable reference implementations, I/O contracts, configuration files, metadata, logging, benchmarks, convergence diagnostics, runtime/memory profiling, and reproducibility standards, rather than on introducing new physical theorems. COS–NUM includes internal COS-graph simulations using sparse-matrix and spectral methods, together with three first-generation data-processing chains: a COS–Planck CMB-anomaly pipeline, a mutual-information-based CMB scale-asymmetry pipeline, and a COS+astrophysical SGWB fitting pipeline. These CMB and SGWB components are treated as auditable statistical and phenomenological pipelines, not as complete end-to-end derivations from microscopic COS graph dynamics. The paper and appendices describe code structure, public API conventions, example scripts, configuration files, reference simulations, convergence tests, logging formats, and archival requirements. COS–NUM therefore serves as the technical numerical backbone of the COS program, supporting transparent, reproducible, and extensible computational checks and future COS-EXP analyses.
Attila Görhöny (Thu,) studied this question.