Sub-Jupiter gas giants orbiting giant stars uncovered using a Bayesian framework

Giant stars have been shown to be rich hunting grounds for those aiming to detect giant planets orbiting beyond ~0.5 AU. Here we present two planetary systems around bright giant stars, found by combining the radial-velocity (RV) measurements from the EXPRESS and PPPS projects, and using a Bayesian framework. HIP 18606 is a naked-eye (V = 5.8 mag) KOIII star, and is found to host a planet with an orbital period of ~675 days, and to have a minimum mass (m sini) of 0.8 MJ and a circular orbit. HIP 111909 is a bright (V = 7.4 mag) K1III star, and hosts two giant planets on circular orbits with minimum masses of m sini=1.2 MJ and m sini=0.8 MJ, and orbital periods of ~490 d and ~890 d, for planets b and c, respectively, strikingly close to the 5:3 orbital period ratio. An analysis of the 11 known giant star planetary systems arrive at broadly similar parameters to those published, whilst adding two new worlds orbiting these stars. With these new discoveries, we have found a total of 13 planetary systems (including three multiple systems) within the 37 giant stars that comprise the EXPRESS and PPPS common sample. Periodogram analyses of stellar activity indicators present possible peaks at frequencies close to the proposed Doppler signals in at least two planetary systems, HIP 24275 and HIP 90988, calling for more long-term activity studies of giant stars. Even disregarding these possible false positives, extrapolation leads to a fraction of 25–30% of low-luminosity giant stars hosting planets. We find that the mass function exponentially rises towards the lowest planetary masses; however, there exists a ~93% probability that a second population of giant planets with minimum masses in the range 4–5 MJ is present, worlds that could have formed by the gravitational collapse of fragmenting protoplanetary disks. Finally, our noise modelling reveals a lack of statistical evidence for the presence of correlated noise at these RV precision levels, meaning white noise models are favoured for such datasets. However, different eccentricity priors can lead to significantly different results, advocating for model grid analyses such as those applied here to be regularly performed. By using our Bayesian analysis technique to better sample the posteriors, we are helping to extend the planetary mass parameter space to below 1 MJ, building the first vanguard of a new population of super-Saturns orbiting giant stars.