2025 workshop program
Spectrophotometric Inference of Stellar and Binary Parameters
The work
Abstract
Understanding the dynamics of binary systems is invaluable to our understanding of stellar evolution, as roughly half of all stars are in binary systems. Inferring the physical properties of binary systems proves to be more difficult than for single stars. This is primarily caused by an increased number of free parameters, and by the complications of modelling a potentially blended spectrum that changes with orbital motion. The Python tool uberMS attempts to overcome the obstacles presented by the complexities of double-lined spectroscopic binaries (SB2s). It employs stochastic variational inference (SVI), a posterior approximation technique that scales well with the number of free parameters. SVI allows uberMS to jointly model observed spectra and photometry by simultaneously fitting all stellar parameters and elemental abundances. The original version of uberMS could model only single stars, and we recently updated it to model SB2s while treating the constituent stars as independent of each other. We now attempt to model the stellar and orbital parameters of SB2s simultaneously by including a simple relationship between the mass ratio, system radial velocity, and component radial velocities. We test this new version of uberMS on MMT/Hectochelle spectra for known SB2s in NGC 6819, and with mock spectra. We will present the preliminary results from these tests and our evaluation of the code’s performance.