ANAorb: A robust framework for the optimal modelling of binaries

Reliable orbital solutions for binary stars are essential for deriving accurate stellar parameters and for testing stellar-evolution models. However, current fitting approaches often struggle with local minima, strong parameter correlations, and the heterogeneous nature of historical and modern astrometric data (e. g. micrometric and interferometric). As the volume and precision of astrometry from missions such as grow, these limitations increasingly risk producing biased or sub-optimal orbital solutions. Gaia We introduce a unified and statistically rigorous framework for modelling binary-star orbits that overcomes the common weaknesses of existing methods. The approach is designed to integrate heterogeneous datasets, mitigate parameter degeneracies, avoid convergence to local minima, and provide reliable and internally consistent uncertainty estimates. We developed a flexible orbital-modelling framework that incorporates several methodological advances. These include: (i) a statistically rigorous outlier-detection and rejection scheme based on M-estimators; (ii) the combined use of multiple global and local optimisation algorithms to ensure efficient and robust convergence; and (iii) uncertainty estimation via a jackknife resampling procedure. The module, embedded within the workflow, provides dynamical parallaxes and individual stellar masses that are directly tied to the fitted orbit. converges rapidly, typically within a few hundred iterations, and systematically improves standard goodness-of-fit diagnostics (e. g. ̧hi², rms residuals) relative to existing solutions. Outliers are removed in a statistically justified manner, and jackknife resampling yields realistic uncertainties for all parameters. Application to the benchmark systems 70 Tau (in the Hyades moving group) and 22 Cas A (a B-type emission-line star with a circumstellar envelope) demonstrates markedly improved orbital solutions, distances, and individual masses. Binary systems are fundamental laboratories for stellar astrophysics. provides a fast, reliable, and statistically consistent framework for deriving precise orbital solutions from heterogeneous data, while the module delivers accurate dynamical parallaxes and individual masses. The methodology establishes a general and reproducible framework suited to forthcoming large-scale analyses of binary-star populations.