Measuring Extreme Precision Radial Velocities of Exoplanet-Hosting Stars in the Presence of Stellar Noise Using Deep Learning

Stellar variability is one of the largest contributors to noise in Extreme Precision Radial Velocity (EPRV) measurements. We are developing deep learning-based approaches to measure small injected planet-like RVs in the presence of larger amplitude RV noise caused by stellar activity. Our networks are trained using the HARPS-N sun-as-a-star extracted (order-by-order) spectra from 2015-2018, with the goal of using NEID sun-as-a-star spectra in the near future. The unprecedented signal-to-noise and cadence of sun-as-a-star spectra allow us to evaluate the effectiveness and limitations of neural networks at separating stellar and planet-induced RVs in the wavelength domain at sub-m/s precision, and determine their applicability to the EPRV community’s goal of mitigating stellar RV variability.