# On The Statistical Properties of the Lower Main Sequence

## TL;DR

This paper assesses how classical and asteroseismic observations, combined with statistical analysis and machine learning, can reveal detailed properties of stars on the lower main sequence, including unobservable characteristics.

## Contribution

It introduces a comprehensive analysis of the diagnostic power of various stellar observables and identifies optimal combinations for inferring stellar parameters using upcoming survey data.

## Key findings

- Certain observable combinations can tightly constrain stellar properties.
- Principal component analysis reveals redundancy in measurements.
- Machine learning enhances parameter estimation accuracy.

## Abstract

Astronomy is in an era where all-sky surveys are mapping the Galaxy. The plethora of photometric, spectroscopic, asteroseismic and astrometric data allows us to characterise the comprising stars in detail. Here we quantify to what extent precise stellar observations reveal information about the properties of a star, including properties that are unobserved, or even unobservable. We analyse the diagnostic potential of classical and asteroseismic observations for inferring stellar parameters such as age, mass and radius from evolutionary tracks of solar-like oscillators on the lower main sequence. We perform rank correlation tests in order to determine the capacity of each observable quantity to probe structural components of stars and infer their evolutionary histories. We also analyse the principal components of classic and asteroseismic observables to highlight the degree of redundancy present in the measured quantities and demonstrate the extent to which information of the model parameters can be extracted. We perform multiple regression using combinations of observable quantities in a grid of evolutionary simulations and appraise the predictive utility of each combination in determining the properties of stars. We identify the combinations that are useful and provide limits to where each type of observable quantity can reveal information about a star. We investigate the accuracy with which targets in the upcoming TESS and PLATO missions can be characterized. We demonstrate that the combination of observations from GAIA and PLATO will allow us to tightly constrain stellar masses, ages and radii with machine learning for the purposes of Galactic and planetary studies.