Does Hamiltonian Monte Carlo mix faster than a random walk on multimodal densities?

Open in new window