Shell molecular dynamics

GROMACS can simulate polarizability using the shell model of Dick and Overhauser 43. In such models a shell particle representing the electronic degrees of freedom is attached to a nucleus by a spring. The potential energy is minimized with respect to the shell position at every step of the simulation (see below). Successful applications of shell models in GROMACS have been published for \(N_2\) 44 and water45.

Optimization of the shell positions

The force \(\mathbf{F}\)\(_S\) on a shell particle \(S\) can be decomposed into two components

\[\mathbf{F}_S ~=~ \mathbf{F}_{bond} + \mathbf{F}_{nb}\]

where \(\mathbf{F}_{bond}\) denotes the component representing the polarization energy, usually represented by a harmonic potential and \(\mathbf{F}_{nb}\) is the sum of Coulomb and van der Waals interactions. If we assume that \(\mathbf{F}_{nb}\) is almost constant we can analytically derive the optimal position of the shell, i.e. where \(\mathbf{F}_S\) = 0. If we have the shell S connected to atom A we have

\[\mathbf{F}_{bond} ~=~ k_b \left( \mathbf{x}_S - \mathbf{x}_A\right).\]

In an iterative solver, we have positions \(\mathbf{x}_S(n)\) where \(n\) is the iteration count. We now have at iteration \(n\)

\[\mathbf{F}_{nb} ~=~ \mathbf{F}_S - k_b \left( \mathbf{x}_S(n) - \mathbf{x}_A\right)\]

and the optimal position for the shells \(x_S(n+1)\) thus follows from

\[\mathbf{F}_S - k_b \left( \mathbf{x}_S(n) - \mathbf{x}_A\right) + k_b \left( \mathbf{x}_S(n+1) - \mathbf{x}_A\right) = 0\]

if we write

\[\Delta \mathbf{x}_S = \mathbf{x}_S(n+1) - \mathbf{x}_S(n)\]

we finally obtain

\[\Delta \mathbf{x}_S = \mathbf{F}_S/k_b\]

which then yields the algorithm to compute the next trial in the optimization of shell positions

\[\mathbf{x}_S(n+1) ~=~ \mathbf{x}_S(n) + \mathbf{F}_S/k_b.\]