An Application of Heuristic Algorithms to Radial Velocity Data from Multiple-Planet Extrasolar Systems
M. Baloković, M. Kürster · Max Planck Institute for Plasma Physics · 2008
We present our work on the application of heuristic global optimization algorithms to radial velocity data from multi-planet extrasolar systems. The aim is to find an optimum fit for the whole system simultaneously, thus avoiding any assumptions about the correctness of previously found planets. The complexity of the task calls for algorithms efficient in global optimization problems with many degrees of freedom. We explored the capabilities of the basic genetic and simulated annealing (Metropolis) algorithms and added more features to improve their performance. They were then combined for optimum performance on this specific problem, utilizing the morphology of the parameter space in a new search strategy. Here we present the programs and their application to artificial and real radial velocity data in which multiple planets were previously identified.