The orbital fits of multi-planetary systems from radial velocity data has proved to be a complex task. In some cases, different orbital solutions provide similarly good fits, especially when two planets are near mean-motion resonances. Ferraz-Melloet al(2005) and Goździewskiet al(2005) showed that the published best fits of systemsHD82932andHD160691are dynamically unstable, and re-determined their orbital parameters with Monte Carlo and genetic algorithms. In both cases dynamically stable

more »

... were found with RMS similar to the published orbits. It was also shown that uncertainties in the stellar mass Ferraz Melloet al(2005) and the stellar jitter Gozdziewskiet al(2005) can significantly affect the orbital determination. Ford (2005) used a Markov chain Monte Carlo technique to quantify the orbit uncertainties. For some planetary systems he found a strong correlation between the orbital elements and/or significant non-Gaussian error distribution in the parameter space. As a consequence, the actual uncertainties in the orbital fits can be much larger (or smaller) than those published.