[gmx-users] what does the "Coulomb (SR) mean under PME?

Mark Abraham Mark.Abraham at anu.edu.au
Wed Feb 15 08:15:36 CET 2012

On 15/02/2012 2:46 AM, Qiao Baofu wrote:
> Hi all,
> Anybody can help me out! Thanks in advance?
> In my test simulation, there are one Na^+ and one Cl^- (the distance 
> of 1nm) in vacuum in a very big simulation box (10*10*10 nm^3). I 
> calculated the energies under two different conditions, 
> "coulombtype=PME" vs "coulombtype=cut-off".
> 1. "coulombtype=PME"
> Coul. recip.               -139.191         --          0          0  
> (kJ/mol)
> Coul-SR:NA-CL *-0.0322805*         --          0          0  (kJ/mol)
> LJ-SR:NA-CL              -0.00100285         --          0          0  
> (kJ/mol)
> 2. "coulombtype=cut-off"
> Coul-SR:NA-CL *-138.935*         --          0          0  (kJ/mol)
> LJ-SR:NA-CL              -0.00100285         --          0          0  
> (kJ/mol)
> I did some calculations. Under "coulombtype=cut-off", E(Coul-SR) = 
> f*e_i*e_j/r with f=138.935, which is what I expect.
> However under "coulombtype=PME", how does Gromacs get the value of 
> -0.0322805kJ/mol for "*SR*" term? If I understand it correctly, "SR" 
> means "short-range", which is within the cut-off distance. Since the 
> cut-off distance (1.2nm here) is much smaller than the simulation box 
> length (10nm), there is not short-range interaction with the image box.

Your two atoms are within the cut-off distance, so their interaction 
contributes to Coul-SR:NA-CL. The total energy for that interaction is 
distributed over the Coul-SR and Coul. recip terms (which is how PME 
works...), which you will note is approximately equal to Coul-SR for 
coulombtype cut-off.

By construction, your periodic images are too far away to contribute to 
this term in either case (and you'd have been prevented by grompp from 
getting this far, if that were not true).

> Some details:
> Gromacs 4.5.5
> PBC is used.
> distance of Na^+ and Cl^-: 1nm
> Na^+ and Cl^- are frozen in XYZ dimensions.
> Only 1 step is run.

Using mdrun -rerun is often the best way to calculate single-point 

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://maillist.sys.kth.se/pipermail/gromacs.org_gmx-users/attachments/20120215/2f5dfe98/attachment.html>

More information about the gromacs.org_gmx-users mailing list