[gmx-users] Total energy of a group of atoms

[Mark Abraham]

On 11/06/2012 8:22 PM, Markus Kaukonen wrote:
> Dear All,
> Sorry to trouble you with the same question, but my question from last
> friday was probably badly formulated, so I'll try again.
>
> Assume our system consists of ten atoms (atom indexes 1-10).
> It has two groups,
> group A has atoms 1, 7, 9
> group B has atoms 2-6, 8, 10
>
> Is it possible (either during the run or by postprocessing)
> to get the total energies associated to groups A and B (thus E_tot =
> E_tot(A) + E_tot(B) )?
>
> If not, what would be easiest strategy to hack this out? Would it be
> g_energy or g_enemat or something else?
>

This is straightforward to do as far as non-bonded quantities go, but 
requires planning when you construct your .mdp for either the original 
run, or an mdrun -rerun. Look for energy groups in the manual.

Decomposing the bonded quantities will require you do some more tedious 
work, e.g. zeroing parameters or removing molecules in the .top and 
doing a rerun.

As Justin said, you will not be able to easily decompose the 
reciprocal-space term for PME, and whether these sums mean anything is 
doubtful.

Mark