Retired A Stars And Their Companions. III. Comparing The Mass-Period Distributions Of Planets Around A-Type Stars And Sun-Like Stars

Bowler et al.
The Astrophysical Journal. January 20, 2010

After the introduction and background about many interesting things I need to read more about later, Bowler et al gives updated information about the five stars already shown to possess planets in a 159 star survey. They applied a Bayesian model comparison test to decide whether or not to add in a constant acceleration due to a binary companion (BIC=-2 ln(Lmax) + k ln(N), lower BIC preferred). Parameter uncertainties were estimated using a Markov Chain Monte Carlo algorithm. Even though I understand what they wrote as the process, I am unsure of how they get the actual uncertainties from the chain of steps. Lomb-Scargle periodograms were used to search for additional short-period companions. They also studied all five stars with photometry over an extended time period to tell if the periodicity was a result of stellar activity or if the planet transited.

One of the most interesting parts of the paper for me was when they discussed the Detection Limits. They synthesized fake orbits and determined the highest K that would be detectable with the given noise. In addition to testing the null planets, they tested those with planets for additional ones. Even though they said they could detect less than Saturn mass planets closer in, most of the noise was from the star so detection could not improve much more. The only chance of finding Neptune sized planets would be from transits or micro lensing. Even with the noise, the finding of higher mass planets was probably the result of a real trend.

For this small sample of stars, they compared the statistics to those for stellar-type stars. Approximately 24% of the stars have planets, consistent with other predictions to more planets around higher mass stars. They also tested whether or not the planets could be drawn from the same mass, period distribution as sun-like stars. When including planet occurrence rates, it was less than a 10^-5 probability that they were the same. Even when they accepted different planet occurrence rates, they ruled out them being drawn from the same mass-period distribution with confidence over 4 sigma. Even though they had too few data points to actually find the parameters, they did rule out several regions of the parameter space. There is also a low eccentricity trend for planets around IM subgiants that differs with normal stars.