Supplementary data and code for "An identifiable scatter floor and calibration framework for the radial acceleration relation"

Analysis code and intermediate numerical results for "An identifiable scatterfloor and calibration framework for the radial acceleration relation"(Tanigawa 2026, submitted to MNRAS). The deposit contains the Python analysis pipeline (49 scripts), shareddata-loading module, and intermediate CSV/NPZ results required to reproducethe findings. The work analyses 171 SPARC galaxies with a hierarchicalBayesian and DAG-based identifiability framework, proving algebraically thatthe (sigma_halo, sigma_within) decomposition is structurally non-identifiablefrom the hierarchical likelihood and that naive fits under-estimatesigma_halo by 30-50% through a conflation-bias mechanism quantified on a640-realisation synthetic grid. A SPARC-like mock catalogue from IllustrisTNGTNG100-1 is also included for an apples-to-apples comparison against LCDM. Manuscript text and published figures are NOT included (publisher copyrighton acceptance). Figures are reproducible by running the corresponding scriptson the deposited intermediate results.