[gmx-users] Calculating the electrostatic potential at a point over time

[Mark Abraham]

On 28/03/2012 12:13 PM, Andrew DeYoung wrote:
> Greetings,
>
> Is it possible to determine the electric potential at the location of an
> atom, relative to infinity?  From physics, I think that electric potential
> (relative to infinity) at the position vector \vec{r} DUE TO a point charge
> q located at \vec{r'}, is given by, in SI units:
>
> V_q(\vec{r}) = q/(4*pi*epsilon0*|\vec{r}-\vec{r'}|)
>
> where |\vec{r}-\vec{r'}| is the magnitude of the vector between the position
> of the source charge (\vec{r'}) and the observation location (\vec{r}).
>
> I would like to sum over that formula to determine the electric potential AT
> the position of an atom located at \vec{r} DUE TO all the other partial
> charges in the system:
>
> V(\vec{r}) = \sum_i ((q_i)/(4*pi*epsilon0*|\vec{r}-\vec{r'_i}|))
>
> where the sum is over all charges in the system except the charge at
> \vec{r}.
>
> If a point charge Q is at \vec{r}, then I think that this is just the
> electrostatic energy of Q, divided by Q.
>
> Is there any way to calculate this in Gromacs?  I know that g_potential
> computes the electrostatic potential across the box (by integrating the
> Poisson equation, it seems), but I want to compute just the electrostatic
> potential at the location of a single atom.

Set up an energy monitor index group containing the atom of interest. 
See manual 3.3, 8.1.1, 7.3.8. Now the .edr file will have the potential 
of that atom with respect to each other energy group (i.e. the rest). 
You can do this in your original simulation, or after the fact with 
mdrun -rerun yourtrajectory.

Mark