[gmx-users] VdW and Coulomb SR contribution to dVdl in thermodynamic integration

[reich at mpikg.mpg.de]

Dear Gromacs-users,

I have been doing a test thermodynamic integration for the conversion of
methanol into a dummy molecule in water. I have been interested in the
contribution of the VdW and Coulomb short range interaction to dVdl.
According to equations 4.108 and 4.104, at lambda=0 (the methanol-state)
the contributions should be the negative Lennard-Jones and Coulomb short
range interaction energies between methanol and the surrounding water. I
have been calculating the interaction energies with g_energy (choosing
LJ-SR:methanol-rest and Coul-SR:methanol-rest as options). For example for
a particular step in my trajectory I get

-7.954321 kJ/mol for the Lennard-Jones interaction

and -56.966656 kJ/mol for the Coulomb intercation between methanol and water.

So the contribution of the combined interactions to dVdl should be

64.921 kJ/mol


I have, however, also used the option sepdvdl with mdrun to get some
detailed information on dVdl during the simulation.
The line from the log-file giving me the contribution of Lennard-Jones and
Coulomb short range interactions for the simulation step in question is:

  VdW and Coulomb SR particle-p. V -1.49467e+04  dVdl  1.01145e+03

So the simulation output gives me a result that is two magnitudes greater
than what I calculated! Is there anything wrong with my reasoning or do
you have any other clue why I end up with two different results?


Thanks

Luther