[gmx-users] interaction energy of bulk TIP3P

[Mark Abraham]

> Hi All
>
> I simulated a cubic box of 2180 TIP3P water molecules using gmx 3.3 and PME
> for 400 ps and extracted the interaction energy of a randomly picked
water
> molecule with rest of the system.
> Simulation was done at 298K and 1bar and usual procedure for generating
a
> water box and equilibration were used.
>
> Average interaction energy reported in the log file was
> Epot (kJ/mol)        Coul-SR          LJ-SR
>       WAT-rest   -8.13828e+01    1.21886e+01
>
> giving a total of          -8.13828e+01  + 1.21886e+01 = -69.194 kJ/mol
>
> Comparing with published numbers using other MD programs this should be
in
> the range of -81 +/- 1 kJ/mol i.e. ~11 kJ/mol less.
>
> any clues on origin of this discrepancy ?  relevant .top , .mdp files
are
> below.

The PME algorithm does an approximation to the full periodic Coulomb
summation, but does so with a calculation in direct space, as above, and
another in reciprocal space, where your energy groups are not meaningful.
So you're only getting the direct-space component of the interaction
energy.

Mark