Trajectory reweighing: Entropy-energy decomposition and dynamical properties from pulling simulations
소속 :
연사 : 정다운 박사님 (University of California, Irvine, USA)
일시 : 2012-05-31 14:00 ~
장소 : 500동 L-305호
일 시 : 2012년 5월 31일, 2:00 PM
장 소 : 500동 L-305호
-Abstract-
The Jarzynski identity can be applied to molecular dynamic simulations in which the system is pulled repeatedly but quickly along some coordinate, allowing the calculation of a free energy profile along the pulling coordinate from a simulation set of independent trajectories that can be run in parallel. By performing this calculation at different temperatures, derivatives of the free energy can in principle be obtained by finite difference. Here we show that, by applying a trajectory reweighting formalism, free energy values can also be recovered exactly at slightly perturbed temperatures without repeating the parallel simulation set. These perturbed free energies can be used to calculate entropy and energy profiles along the pulling coordinate via thermodynamic identities involving temperature derivatives of the free energy. The novel method of reconstructing these profiles is exemplified on a prototypical model system. We also compute the time correlation function to address the kinetics of the system from repeated pulling simulations. Trajectory ensembles at multiple conditions such as different spring constant or pulling speeds are aggregated to obtain an optimal estimate of reweighed dynamical properties at a physical condition. It can be applied to any trajectory ensemble provided the ratio of path weight at different conditions can be computed. Trajectory sampling at multiple conditions allows efficient sampling of trajectory space as well as reduced computational cost.
장 소 : 500동 L-305호
-Abstract-
The Jarzynski identity can be applied to molecular dynamic simulations in which the system is pulled repeatedly but quickly along some coordinate, allowing the calculation of a free energy profile along the pulling coordinate from a simulation set of independent trajectories that can be run in parallel. By performing this calculation at different temperatures, derivatives of the free energy can in principle be obtained by finite difference. Here we show that, by applying a trajectory reweighting formalism, free energy values can also be recovered exactly at slightly perturbed temperatures without repeating the parallel simulation set. These perturbed free energies can be used to calculate entropy and energy profiles along the pulling coordinate via thermodynamic identities involving temperature derivatives of the free energy. The novel method of reconstructing these profiles is exemplified on a prototypical model system. We also compute the time correlation function to address the kinetics of the system from repeated pulling simulations. Trajectory ensembles at multiple conditions such as different spring constant or pulling speeds are aggregated to obtain an optimal estimate of reweighed dynamical properties at a physical condition. It can be applied to any trajectory ensemble provided the ratio of path weight at different conditions can be computed. Trajectory sampling at multiple conditions allows efficient sampling of trajectory space as well as reduced computational cost.
