Data-driven Stellar Models

We developed a data-driven model to map stellar parameters (Teff, , and ) accurately precisely broadband photometry. This must, does, simultaneously constrain the passband-specific dust reddening vector in Milky Way, R. The uses neural network learn (de-reddened) absolute magnitude one band colors across many bands, given from spectroscopic surveys parallax constraints Gaia. To demonstrate effectiveness of this approach, we train our on data set with LAMOST, APOGEE, GALAH, Gaia parallaxes, optical near-infrared photometry Gaia, Pan-STARRS 1, Two Micron All Sky Survey Wide-field Infrared Explorer. Testing these sets leads an excellent fit precise—and by construction—accurate prediction color–magnitude diagrams bands. flexible approach rigorously links photometric surveys, also results improved, Teff-dependent As such, it provides simple accurate method for predicting evolutionary models. Our will form basis infer properties, distances, extinction data, which should be great use 3D mapping Way. trained can obtained at doi:10.5281/zenodo.3902382.