Detection of the closest Jovian exoplanet in the Epsilon Indi triple system

Abstract: We confirm the trend in the radial velocity data for Epsilon Indi A suggesting a long-period planetary companion and find significant curvature is present, sufficient to quantify Epsilon Indi Ab as a cold Jupiter with a minimum mass of $2.71_{-0.44}^{+2.19}~M_{\rm Jup}$ on a nearly circular orbit with a semi-major axis of $12.82_{-0.71}^{+4.18}$ au and an orbital period of $52.62_{-4.12}^{+27.70}$ yr. We also identify other significant signals in the radial velocity data. We investigate a variety of spectral diagnostics and interpret these signals as arising from activity-induced radial velocity variations. In particular, the 2500 and 278 d signals are caused by magnetic cycles. While a planetary signal might be present in the 17.8 d signal, the origin of 17.8 and 11 d signals are most easily interpreted as arising in the rotation of the star with a period of about 35 d. We find that traditional activity indicators have a variety of sensitivities. In particular, the sodium lines and CaHK index are sensitive to all activity-induced signals. The line bisector measurement is sensitive to stellar rotation signal while H$\alpha$ is sensitive to the secondary magnetic cycle. In general, because of their different sensitivities these activity indicators introduce extra noise if included in the noise model whereas differential RVs provide a robust proxy to remove wavelength-dependent noise efficiently. Based on these analyses, we propose an activity diagnostics procedure for the detection of low amplitude signals in high precision radial velocity data. Thus the Epsilon Indi system comprises of at least Epsilon Indi A, Ab as well as a long period brown dwarf binary Ba and Bb; so it provides a benchmark case for our understanding of the formation of gas giants and brown dwarfs.