Classical molecular dynamics simulations rely on a set of forcefield parameters in order to produce correct results. When existing forcefields fail to characterize a molecule, systematic improvement of the parameters is difficult.For a correct parameterization, the energy gradient of a variety of conformations of the desired molecule should be the same when evaluated with both quantum and molecular dynamics methods. A program has been developed which implements a genetic algorithm (GA) to minimize the least squares difference between the QM and MD energies for a selection of conformations of a given structure. Implementation of a GA for parameter fitting results in significant improvements in accuracy when compared with default parameters for a number of systems.The algorithm enables the accurate simulation of molecules that were previously difficult or impossible to parameterize, making existing molecular dynamics software considerably more robust and applicable to a wider variety of systems.

Paramfit: A program for automated forcefield parameter generation using a genetic algorithm

243rd National Meeting (2012)

The official archive and repository for all ACS Meetings & Expositions abstracts and select posters or presentations.