Diana Kossakowski (USA)
kossakowski @ mpia.de
Signal search in RV data
I am working with the CARMENES instrument, where around 300 M-dwarfs are observed for radial velocity (~1m/s) in the optical and near-infrared spectrum. In general, the purpose of applying radial velocity techniques to planetary systems is to obtain the most probable planetary mass, which can then be combined with the calculated radius produced from the transiting method such that the planet’s density can be computed in order to give us more insight on what type of planet we are observing.*
Radial velocity measurements are becoming more and more precise (<1m/s), allowing for the opportunity of more Earth-like planet detections but also introducing the additional cost of picking up unwanted stellar activity. Stellar activity includes magnetic cycles, stellar rotation period, granulations, stellar spots and plages, and oscillations, with varying timescales of years, days, nights, and minutes. Some activity can be avoided by optimizing observation sampling strategies, but even after those considerations, some activity may still overpower and mask the true exoplanet detection. However, we can examine correlations of various stellar activity indicators (i.e. H-alpha lines, CaII triplet lines, wavelength- and time-dependency of periodic signals, and so on) that would enable us to achieve our main goal of determining whether a signal is truly a planet or if it is an artifact by the star.
Therefore for my PhD, I will be exploring methods to improve the radial velocity search by implementing Bayesian methods, Gaussian Process, and the possibility of Machine Learning techniques (i.e. perceptrons). Of course, in the ideal world, having 1000s of data points per star would be greatly beneficial but reality tells a different story which will be a challenge to overcome. Alongside, I will be looking at the CARMENES data for interesting planetary systems to study more in depth, and possibly incorporate TESS results (~2018) to produce intriguing results.
*Observed stars in CARMENES most likely do not have transiting data but follow-up data may be possible to obtain in future exoplanet transiting missions.
Supervisor: Martin Kuerster (MPIA)