[gmx-users] non isotropic kinetic energy

[herz]

> It is well known that water at a surface is ordered. How do you choose
> the waters to analyze? If there is a net ordering in the system due to
> external factors your system is not in equilibrium, but rather in a
> steady state. However if you have e.g. vacuum-water-vacuum the
> symmetry makes that equipartition should hold, provided the averaging
> is over a time long enough such that water molecules sample both
> surfaces.
>
> Of course if you are worried about the constraints you could do a
> simulation with flexible water for testing.
"However if you have e.g. vacuum-water-vacuum the symmetry makes that
equipartition should hold, provided the averaging is over a time long
enough such that water molecules sample both surfaces."

I agree 100%! We have averaged rotational and translation kinetic energy
per water molecule for all water molecules in the system over the
complete system (vacuum, water,vacuum) over a long time (so we think we
are in equilibrum). So equipartition should hold, but it appears it
doesn't (rot energy not equipartitioned).
We have tryed with flexible water and here the problem vanishes (so
equipartition works using the same analysis!).  This difference between
constraint and unconstraint water leads us to the concludion that there
is aporblem with the constraints.

Alex