Radial Velocity Data Analysis with Sparse Recovery Techniques
In the search for exoplanet signal in radial velocity measurements, an efficient algorithm to extract the planet period is a fundamental step. Various phenomena complicate the period search: the period found can be due to a combination of aliases of other periodicities, stellar noise can mimic planetary signals, correlated noise can bury planetary signals etc. Most recent approaches privilege fitting a complete model instead of fitting planets one by one, since global searches reduce both rates of false positives and false negatives. Unfortunately, the performance improvement of these methods comes at the cost of their greater computational complexity.
We will present a new method based on sparse recovery algorithms that allows to search for several planets at once but that is much faster than existing ones, based on random searches. The final tool can be used as a Lomb-Scargle periodogram, but with the advantage of avoiding most cases where the tallest peak of the periodogram is spurious. Furthermore, the new method has demonstrated its efficiency to detect small signals of planets orbiting active stars. We will present the latest improvements of the method, its application to some HARPS systems, as well as the status of its public version.