| Summary: | A possible alternative to running chemical experiments in the laboratory could be to conduct them digitally instead, by running computer simulations. One prominent simulation technique is Molecular Dynamics (MD), which simulates the dynamics of molecular systems at the atomic scale. With sufficiently accurate molecular models and enough computational power, the underlying mechanisms for the initiation and occurrence of chemical phenomena could then possibly be understood, like for chemical reactions, phase transitions, permeation, protein folding, and membrane fusion. In combination with experiments, computer simulations can therefore help drive the development of new drug candidates, vaccines, catalysts, and technologies like energy storage and conversion materials. By running straightforward MD, however, observing one single dynamical event may require thousands of years in wall time, far too long to be of any practical use. By utilizing the Monte Carlo (MC) based Replica Exchange Transition Interface Sampling (RETIS) method instead, we reduce the required wall time converge to months or up to one year. While doable, reducing the required simulation time even more would be highly beneficial. The major work presented in this thesis has therefore been to improve the RETIS method by developing various smart MC-based enhancements, like introducing a new, fast decorrelating MC move called Wire Fencing, reducing MC move rejection by biasing the acceptance probability, enabling the calculation of infinite replica exchange swaps without the factorial cost, and deriving a parallel RE scheme with linear MD scaling. Connecting the various enhancements together forms the infinity RETIS protocol that enables rapid convergence and the calculation of rate constants within days or weeks instead. Together with method development, we have also applied infinity RETIS to study a number of rare events, like water autoionization, the formation of carbonic acid from solvated carbon dioxide, the electron transfer reaction between ...
|
|---|