Algorithm Validation
Algorithm Validation
Cross-validation with reference software, public data, and analytical solutions
This section records algorithm comparisons performed with fixed inputs to assess agreement between the software and independent references. Each record provides the shared inputs, configuration evidence, result comparison, and acceptance criteria.
| Validation | Independent reference | Comparison | Main result |
|---|---|---|---|
| Broadband AR Cross-Validation | Essential Macleod | 400–700 nm forward reflectance and four-layer thickness optimization | Overlapping spectra; maximum optimized-thickness difference of 0.782% |
| DBR Cross-Validation | Essential Macleod | 400–800 nm reflectance, transmittance, reflection phase, and a 510 nm angle sweep | Overlapping R/T; converted phase agreement; matching angular curves for three polarizations |
Agreement applies only to the stated model scope and inputs; it does not replace measurement, fabrication-tolerance analysis, or material-uncertainty analysis. For the assumptions that scope shares with every transfer-matrix solver, and the situations in which real samples break them, see Limits of the TMM Model.