Modelling stellar activity with Gaussian process regression networks

Stellar photospheric activity is known to limit the detection and characterisation of extra-solar planets. In particular, study Earth-like planets around Sun-like stars requires data analysis methods that can accurately model stellar phenomena affecting radial velocity (RV) measurements. Gaussian Process Regression Networks (GPRNs) offer a principled approach simultaneous time-series, combining structural properties Bayesian neural networks with non-parametric flexibility Processes. Using HARPS-N solar spectroscopic observations encompassing three years, we demonstrate this framework capable jointly modelling RV traditional indicators. Although consider only simplest GPRN configuration, are able describe behaviour at least as previously published methods. We confirm correlation between time series reaches maximum separations few days, find evidence non-stationary in series, associated an approaching minimum.