1: \begin{abstract}

2: %

3: Precise radial velocity measurements have led to the discovery of

4: $\sim170$ extrasolar planetary systems.  Understanding the

5: uncertainties in the orbital solutions will become increasingly

6: important as the discovery space for extrasolar planets shifts to

7: planets with smaller masses and longer orbital periods.  The method of

8: Markov chain Monte Carlo (MCMC) provides a rigorous method for

9: quantifying the uncertainties in orbital parameters in a Bayesian

10: framework (Ford 2005a).  The main practical challenge for the general

11: application of MCMC is the need to construct Markov chains which

12: quickly converge.  The rate of convergence is very sensitive to the

13: choice of the candidate transition probability distribution function

14: (CTPDF).  Here we explain one simple method for generating alternative

15: CTPDFs which can significantly speed convergence by one to three

16: orders of magnitude.  We have numerically tested dozens of CTPDFs with

17: simulated radial velocity data sets to identify those which perform

18: well for different types of orbits and suggest a set of CTPDFs for

19: general application.  Additionally, we introduce other refinements to

20: the MCMC algorithm for radial velocity planets, including an improved

21: treatment of the uncertainties in the radial velocity observations, an

22: algorithm for automatically choosing step sizes, an algorithm for

23: automatically determining reasonable stopping times, and the use of

24: importance sampling for including the dynamical evolution of multiple

25: planet systems.  Together, these improvements make it practical to

26: apply MCMC to multiple planet systems.  We demonstrate the

27: improvements in efficiency by analyzing a variety of extrasolar

28: planetary systems.

29: %

30: \end{abstract}

31: