The detectability of single- and multiple-planet systems in Gaia data

Sammanfattning: The all-sky survey of Gaia will generate vast amounts of astrometric data, in which there are expected to be thousands of planets found. Finding a system of $n_\textup{p}$ planets requires fitting of a total $5+7n_\textup{p}$ parameters: five astrometric and seven Keplerian parameters for every planet. The problem is thus highly non-linear and computationally prohibitive. After exploiting the linear properties of the Thiele-Innes constants, the remaining three non-linear parameters still limit the number of stars in the Gaia data that are practicable to probe for planets. The aim of this thesis is to investigate the feasibility of further eliminating two non-linear parameters, which is accomplished by assuming a circular orbit in the fit. If this approach is successful at reliably finding planet candidates even for eccentric orbits, it can possibly be used to expand the number of stars that can be searched for planets in the Gaia data. The approach is tested in simulated Gaia observations of known single- and multiple-planet systems from radial velocity (RV) measurements. Two detection metrics are used to enable comparison with past studies: a simple signal-to-noise (S/N) threshold and a more robust metric based on orbit fitting, the $\Delta\chi^2$ metric. The results indicate that, assuming a circular orbit in the fit, the orbital period can be correctly determined for planets with eccentricity $\lesssim0.8$ in single-planet systems, and $\lesssim0.5$ in multiple-planet systems. It is also found that around one third of the total number of planets in the current RV catalog, and at least one sixth of its multiple-planet system members, can reliably or somewhat reliably be detected and characterized by Gaia.