Sunday, 1 May 2016
Background. A major goal of NASA’s exoplanet exploration theme is the ultimate detection and characterization of Earth-mass planets around Solar-like stars. As of March 2008, more than 250 exoplanets are known. About 95% of these, including all those orbiting nearby stars, have been found by precision Doppler surveys, primarily by Co-I Butler’s group and the Geneva group (Butler et al. 2006). Statistics of currently detected planets, again primarily from radial velocity searches, indicate that the number of planets increases at lower masses (Fig. 1). Extrapolation predicts that 11% of solar-like stars have a planet with mass greater than the Earth in an orbit of a year or less (Cumming et al. 2008). Achieving the extraordinarily ambitious goal of actually finding habitable planets requires the detection of nearby stars that are capable of sheltering Earth-like planets, i.e., stars with long-period Jupiter-mass planets – Jupiter intercepts Oort Cloud comets that would otherwise catastrophically impact the Earth (Wetherill 1994a), and rocky planets in the habitable zone.
About 10% of the stars with planets discovered in RV searches actually have multiple planets. By studying the architectures of these systems, we will deduce the dynamical effects that produced them. From the data collected by Co-I Butler, one would like to determine the mass and orbital parameters of each orbiting planet. The dynamical interactions between planets in a multi-planet system, such as the Solar System directly impact habitability. For example, the late heavy bombardment may have occurred because Jupiter and Saturn entered a dynamical resonance (Gomes et al. 2005). During this time, the Moon and Earth received impacts at high rates, which both would have delivered volatiles and reshaped our young planet.
Research to Date. Co-I Butler is conducting long term precision Doppler surveys with the Keck 10-m, Magellan 6.5-m, Lick 3-m, and the Anglo-Australian 3.9-m telescopes. These surveys have found about 160 planets over the past 12 years. Over the past five years, Butler has improved the long-term precision of his survey from 3 m/s radial velocity to 1.5 m/s.
An attempt to fit data for a system of four planets observed in RV by three telescopes would involve twenty-three free parameters. The large number of parameters in multiple-planet systems, the sparse data sets, and the fact that signal-to-noise ratios are low for low-mass planets, makes the fitting process challenging. Conventional approaches involving periodograms often perform poorly when multiple companions are present with comparable orbital periods.
Proposed Work. With measurement precision of 1 m/s and long telescope runs of 2 to 6 months, Butler’s search will be sensitive to terrestrial mass planets in the habitable zone of nearby stars. He will bring on-line several important capabilities in the next year. He will commission the Planet Hunting Spectrometer for Carnegie’s Magellan Telescopes for 1 m/s on the nearest 200 southern hemisphere stars and two 32-inch robotic photometry telescopes at Las Campanas Observatory for assessing the photometric stability of our target stars. Photometry helps vet radial velocity data for signals generated by stellar activity and provides a list of quiet stars for which the highest precision may be obtained. Using the 2.4-m robotic planet finding telescope at Lick Observatory, partially funded by NASA and due to start observations in 2008, Butler will follow interesting low amplitude, terrestrial planet, signals every night. For K and M dwarfs, these new facilities will be capable of detecting rocky planets within the habitable zones.
Co-I Chambers is developing a new approach to obtaining multiple-planet fits to existing Doppler velocity data sets using particle-swarm optimization (PSO). This optimization technique has been used extensively in other fields such as electrical engineering (e.g. Robinson and Rahmat-Samii 2004), but has received little attention in the astronomical community. Preliminary tests have found that PSO outperforms other optimization methods for the problem at hand.
Chambers will continue these studies following a 3-phase approach. In Phase 1, he will design tests to determine the optimal set of parameters for finding good fits in a given amount of computer time. In Phase 2, he will examine systems with known planets to improve model fits and search for additional companions. The great majority of published exosolar systems contain a single planet, yet in many cases, trends suggest additional bodies are present. In Phase 3, he will consider the effects of planetary interactions (e.g. Laughlin and Chambers 2001). Once these methods are incorporated into the PSO technique, Chambers will identify systems in which interactions are necessary to explain the RV data.