1: \begin{abstract}

2:  We derive non-asymptotic quantitative bounds for convergence to 

3:  equilibrium of the exact preconditioned Hamiltonian Monte Carlo algorithm (pHMC) 

4:  on a Hilbert space. As a consequence, explicit and dimension-free bounds for pHMC applied to high-dimensional distributions arising in transition path sampling and path integral molecular dynamics are given.  Global convexity of the underlying potential energies is not required. Our results are based on a two-scale coupling which 

5:  is contractive in a carefully designed distance. 

6:  % Our main new tool is a two-scale coupling that is contractive in a carefully designed distance.

7:  \end{abstract}

8: