Compensating mass matrix potential for constrained molecular dynamics (Q1371985)

Rigid internal constraints are often used in molecular models to speed up molecular dynamics (MD) simulations. There has been considerable debate regarding the relationship between the statistical averages obtained from such constrained MD simulations and those obtained from conventional unconstrained Cartesian model MD simulations. In particular \textit{U. Fixman} (Proc. Nat. Acad. Sci. 71, 3050 (1974)] pointed out that ensemble averages obtained from MD simulation using the constrained and unconstrained models will differ due to the presence of a metric tensor dependent term in the partition function for constrained molecular models and this has been verified in simulation. Fixman proposed the augmentation of the standard potential by a compensating metric tensor potential in constrained MD simulations to compensate for the effects of the bias term. The prohibitive complexity of computing the metric tensor potential and its gradient has been a major hurdle on the use of these compensating potentials. This paper analyzes the structure of the compensating potential and its gradient and develops simpler expressions for them for the topology molecular models. These expressions are used to derive computational algorithms for use in constrained MD simulations. The algorithms are straight forward extensions of the recently proposed spatial operators based \(O({\mathcal N})\) algorithm.