[gmx-users] Re: [gmx-developers] positive Coulomb LR potential for 100 A molecules system, A is positive molecules!

[Mark Abraham]

On 4/27/2011 4:02 PM, huzhongqiao at gmail.com wrote:
> Dear all,

This question does not address issues of GROMACS development, and so 
does not belong on gmx-developers. Please direct queries to the 
appropriate forum. I am replying on the gmx-users list.
> I am calculating the interaction between A and B, where A is positive 
> molecule, B is negative molecule. What I have done is as 
> follows. First I run MD for the system A+B, which consists of 100 A 
> molecules and 100 B molecules respectively, and the whole system is 
> neutral. After MD, I exact trajectory of only 100 A molecules and only 
> 100 B molecules. So now I have three trajectories, 100A, 100B, 
> 100A+100B. Then I use -rerun flag in mdrun command to get potential E 
> of these three systems respectively. The interaction dE between A and 
> B is calculated using the equation dE = E(100A+100B)-E(100A)-E(100B). 
> The value obtained is seemingly reasonable. But when I check the 
> potential compoents, i.e, LJSR, CoulombSR, CoulombLR (or Coul. recip.) 
> terms, one problem is arising. I find the CoulombLR is positive for 
> system 100A. According to my understanding, A is positive and A 
> molecules are repulsive to each other, and so the system of 100 A 
> molecules should have positive CoulombLR potential. But mdrun -rerun 
> for 100A system give negative CoulombLR potential. In MD run, I used 
> PME for coulomb LR interadtion. Is this the reason? Anybody can help 
> explain this?

The sum of the two parts of the PME approximation to the electrostatics 
of an infinite periodic system has a connection to a physical 
observable. There's no reason why one should expect the individual parts 
to have one also.

Mark