Hybrid Monte Carlo (HMC) methods

This is joint work with Elena Akhmatskaya from Fujitsu Laboratories Europe (FLE) on algorithmic improvements for the hybrid Monte Carlo (HMC) method. We use modified energies to increase the acceptance rate of the molecular dynamics sub-step of HMC. Furthermore, a choice from various modified momentum refreshment steps makes the newly proposed generalized shadowing HMC (GSHMC) method suitable for sampling purposes as well as stochastic dynamics simulations. In a dynamics context, the method can be made close to stochastic Langevin dynamics as well as dissipative particle dynamics (DPD). See the publications E. Akhmatskaya and S. Reich, GSHMC: An efficient method for molecular simulations and E. Akhmatskaya, N. Bour-Rabee and S. Reich, A comparison of generalized hybrid Monte Carlo methods with and without momentum flip for further details. Other related contributions can be found under my publications page.

Back to the homepage of Sebastian Reich