Multiscale detection of temporal features in molecular dynamics simulations: Applications to acceleration methods

Extending the time scales accessible through direct molecular dynamics simulations is an area of active research. New acceleration techniques promise to achieve experimentally relevant time scales. A common ingredient in many of these acceleration schemes is the clear identification of the current state of a system in the presence of strong thermal noise. Furthermore, the detection of transitions, and precise determination of transition times, also presents challenges. We exploit a wavelet-based technique for detecting temporal features in molecular dynamic simulations, including the detection of meta-stable structures and transition events. The properties of a wavelet basis allow the detection of features over multiple time scales. The details of our implementation allow the detection of. temporal features to be computationally efficient and possible during an ongoing molecular dynamics simulation.