1: \begin{abstract}

2:   Hamiltonian Monte Carlo (HMC) is a state-of-the-art Markov chain Monte Carlo sampling algorithm for drawing samples from smooth probability

3:   densities over continuous spaces. We study the variant most widely

4:   used in practice, Metropolized HMC with the St\"{o}rmer-Verlet or

5:   leapfrog integrator, and make two primary contributions.  First, we

6:   provide a non-asymptotic upper bound on the mixing time of the

7:   Metropolized HMC with explicit choices of step-size and number of

8:   leapfrog steps.  This bound gives a precise quantification of the

9:   faster convergence of Metropolized HMC relative to simpler MCMC

10:   algorithms such as the Metropolized random walk, or Metropolized

11:   Langevin algorithm.  Second, we provide a general framework for

12:   sharpening mixing time bounds of Markov chains initialized at a

13:   substantial distance from the target distribution over continuous spaces.  We apply this sharpening device to the Metropolized random walk and Langevin

14:   algorithms, thereby obtaining improved mixing time bounds from a

15:   non-warm initial distribution.

16: 

17: \end{abstract}

18: