[gmx-users] user_defined_potential

[Sang-Min Park]

Dear all,


I have a question concerning the calculation of the Coulomb interaction 
energy with a user defined table.

I generated a table where o n l y the coulomb interaction is multiplied by 
a factor a=2 and furthermore I used the following settings in the mdp 
file:

energygrps= Protein SOL
energygrp_table = Protein SOL
rlist   = 1.0
rcolumb = 1.0
rvdw    = 1.4
vdwtype = cut-off
coulombtype = PME-User
fourierspacing = 0.12
pme_order = 4
optimize_fft = yes


I expected that the coloumb interaction Protein-SOL (E_{user})is 2 times 
bigger than the coloumb interaction Protein-SOL (E_{ref.})which result from a 
reference trajectory with coulombtype = PME.

But actually the energies deviate like :
  min( E_{user}-2*E_{ref})/E_{user} )  ~ 10%
  max( " "                          )  ~ 25%

When I use a factor of a=1 then the coloumb interaction Protein-SOL 
resulting from the mentioned mdruns are exactly the same.


Can someone explain me the origin of this deviations ?