[gmx-users] Total energy with different energy groups

[Mark Abraham]

Daniel Cheong wrote:
> well...  the bonded energies are the same, but the 1-4 and nonbonded 
> interactions are different, especially the coulomb-sr energy.  I am 
> using a direct summation method with no cutoffs to calculate the 
> electrostatic energy.  But the method is consistent between the runs.  
> So that shouldn't matter should it?

Your observations are inconsistent. 1-4 interactions should be just as 
reproducible as bonded ones, since they too depend only on the topology 
and configuration. If the other non-bonded interactions aren't 
reproduced then I can only suppose you are also changing your cut-offs, 
or neighbour list generation frequency or some such. Use diff on your 
.mdp files. Use mdrun -rerun.

Mark