[gmx-users] interaction energies & accuracy of PME ...

[Michael Brunsteiner]

Dear all,

I am interested in the energy of adsorption of a molecule
on a surface and in aqueous solution. To account for
electrostatic LR interactions I applied a recipe proposed
previously in this list (and briefly described further below)

My problem is: the electrostatic long-range contribution to the
interaction energy I get by applying this recipy is positive 
throughout. Intuitively I would expect it to be negative (generally,
unless you are dealing with a rather strange/pathologic system)

Hence my question:
Is this expectation of mine wrong or could it be that 
this way of calculating the term is too inaccurate
(difference between two large numbers, single precision,
default PME parameters) ??

the recipe:

1) perform MD of molecule on surface + water
   and record the average Coulomb-LR (reciprocal Ewald) 
   term, U_Coul_LR(all), every 20, or so, steps.
2) rerun the same simulation twice with
   a) the charges on the molecule set to zero
      to get the average Coulomb-LR (reciprocal Ewald) 
      term U_Coul_LR(Mzero)
   b) the charges in the rest of the system set to zero
      to get the average Coulomb-LR (reciprocal Ewald) 
      term U_Coul_LR(Xzero)
3) Calculate the long-range contribution to the interaction
   energy as <U_Coul_LR(all)-U_Coul_LR(Mzero)-U_Coul_LR(Xzero)>

Thanks in advance for any suggestions,
Michael


__________________________________
Do you Yahoo!?
Yahoo! Finance: Get your refund fast by filing online.
http://taxes.yahoo.com/filing.html