Probabilistic inference of basic stellar parameters: application to flickering stars

TL;DR

This paper introduces a probabilistic approach using Hierarchical Bayesian Modelling to recalibrate the relations between stellar flicker and parameters like surface gravity and density, accounting for intrinsic scatter and improving inference accuracy.

Contribution

It presents a novel probabilistic framework for stellar parameter inference, revealing intrinsic scatter in flicker relations that previous deterministic models overlooked.

Findings

Evidence for additional intrinsic scatter in stellar flicker relations

Flicker is a valid proxy for surface gravity and density across star types

Intrinsic scatter is independent of flicker amplitude

Abstract

The relations between observable stellar parameters are usually assumed to be deterministic. That is, given an infinitely precise measurement of independent variable, `$x$', and some model, the value of dependent variable, `$y$' can be known exactly. In practise this assumption is rarely valid and intrinsic stochasticity means that two stars with exactly the same `$x$', will have slightly different `$y$'s. The relation between short-timescale brightness fluctuations (flicker) of stars and both surface gravity and stellar density are two such stochastic relations that have, until now, been treated as deterministic ones. We recalibrate these relations in a probabilistic framework, using Hierarchical Bayesian Modelling (HBM) to constrain the intrinsic scatter in the relations. We find evidence for additional scatter in the relationships, that cannot be accounted for by the observational uncertainties alone. The scatter in surface gravity and stellar density does not depend on flicker, suggesting that using flicker as a proxy for $\log g$ and $\rho_\star$ is equally valid for dwarf and giant stars, despite the fact that the observational uncertainties tend to be larger for dwarfs.