Force halving

[skfegan]

For correlated forces, the correlation is taken into account in the correlation matrix.

For independent forces, we want to avoid double counting, so force halving should be used. I think this should apply to all the torques (all the rotational motions are approximated as uncorrelated) and translational motion at the whole molecule level.

[skfegan]

This only applies to the vibrational entropy terms.