LACHESIS: Robust stellar parameters through Bayesian model averaging of isochrone grids

Accurate stellar parameters underpin much of modern astrophysics, from the characterisation of exoplanetary systems to Galactic archaeology. For isolated field stars, masses and especially ages are most often inferred by comparing observed properties against stellar evolution models, yet these inferences depend on the chosen model grid, and the systematic differences between grids can rival the statistical uncertainties of any single pipeline. We present , a package that determines stellar masses, radii, and ages from photometry and spectroscopy while explicitly accounting for the systematic uncertainties introduced by the choice of stellar evolution model grid. lachesis python We interpolated five independent isochrone grids in Fe/H , log age, and equivalent evolutionary phase conditioned on the data; sampled each grid using nested sampling; and combined the resulting per-grid posteriors using Bayesian model averaging, with weights proportional to the Bayesian evidence of each grid. We validated against a benchmark of Kepler asteroseismic dwarfs and subgiants. Masses were recovered with a robust scatter of sim4% and radii to sim2%, with the radius negligibly biased by $<1%$ and the masses carrying a small ∼-3% systematic. Stellar age was recovered with a robust scatter of 30% that depends strongly on the evolutionary state. The grid-to-grid systematic contributes a sim9% floor, a subdominant but real term that a single-grid fit omits. Model averaging is better calibrated than when selecting a single grid, its credible intervals approach nominal coverage once the reference uncertainty is included, and the intervals cover near the nominal rate in injection tests even when the generating model lies outside the ensemble. lachesis By marginalising over an ensemble of stellar model grids, folds the choice-of-grid systematic directly into the posterior, so the reported uncertainties reflect the choice-of-grid systematic rather than only within-grid data noise and stay near a nominal value in injection tests even when the generating model lies outside the ensemble. This delivers homogeneous, well-characterised masses, radii, and ages for exoplanet host characterisation and stellar population studies. lachesis