Residual-First Benchmarking of NV-Center ODMR Models Using Public Data: Baseline Spin-Hamiltonian Fits versus a Strictly Reduced, Transparent Comparison Family

This repository provides the complete reproducibility kit for the benchmark study “Residual-First Benchmarking of NV-Center ODMR Models Using Public Data.” The benchmark implements a residual-first methodology in which model adequacy is evaluated primarily through residual structure rather than aggregate fit metrics alone. Public NV-center ODMR datasets are confronted with two explicitly declared model families: (i) a conventional spin-Hamiltonian baseline and (ii) a strictly reduced, transparent comparison family with a fixed parameter budget. The repository includes all raw input data (as obtained from public sources), deterministic analysis scripts, automatically generated figures (model overlays and residual diagnostics), and quantitative summary tables (RMSE, AIC, BIC, Durbin–Watson). No filtering, windowing, or run-specific tuning is applied. All figures and tables appearing in the associated manuscript are reproducible by executing the provided scripts. The benchmark is intended as a methodological and reproducibility-oriented contribution, not as a proposal of new NV-center physics. This dataset supports independent audit, reuse, and extension of residual-first benchmarking approaches in quantum sensing and metrology.