fix qeq/slater electrostatic energy

Dear Lammps users,

I have a potential coupled with Streitz-Mintmire type QEq developed in GULP. I would like to use it with LAMMPS.

In LAMMPS to check the qeq part, I ran static single shot simulation in bulk periodic system. I used the options

pair_style coul/long
fix qeq/slater

I could get the same atomic charges on atoms in both GULP and LAMMPS with the same QEq parameters. The energy calculated from charge interactions are also same, however, the total electrostatic energy was different between two codes. GULP computes also QEq Self-energy and Coulomb correction for Streitz-Mintmire terms.

I was wondering if these energy terms were implemented in LAMMPS. Has anyone had a similar problem? I appreciate if you can share your experience with using fix qeq/slater to compute charge interactions and the computed electrostatic energy.

Thanks in advance.

Fatih

Fatih Sen
Postdoc
Argonne National Lab

Ray can answer this better than I. Also, he has recently
developed a LAMMPS implementation of the Streitz/Mintmire
potential which we will soon release in LAMMPS.

Steve

Dear Lammps users,

I have a potential coupled with Streitz-Mintmire type QEq developed in
GULP. I would like to use it with LAMMPS.

In LAMMPS to check the qeq part, I ran static single shot simulation in
bulk periodic system. I used the options

pair_style coul/long
fix qeq/slater

I could get the same atomic charges on atoms in both GULP and LAMMPS
with the same QEq parameters. The energy calculated from charge
interactions are also same, however, the total electrostatic energy was
different between two codes. GULP computes also QEq Self-energy and Coulomb
correction for Streitz-Mintmire terms.

Since you are using pair_style coul/long, the electrostatic energy is
divided in to ecoul and elong. Are you outputting both?

Fix qeq/slater includes what you call QEq self energy and S-M Coulomb
correction terms, but only in the charge equilibration calculations, not in
ecoul and elong. This is because the pair_style is coul/long, not S-M. So
even if you add elong and ecoul together, the total energy will be
different from the S-M implementation in GULP.

As Steve said, a LAMMPS implementation of the S-M potential will be
released soon, and I'd appreciate if you can let me know if you get same
results then.

Cheers,
Ray

Hello Ray,

I printed only pe and ecoul. When I had only pair_style of coulomb, the potential energy was composed of only the coulomb interactions. The pe value was same as the charge interactions energy printed in GULP. I wish there was a way to print out the self energy and S-M coulomb correction terms of the charge equilibration since they are already calculated.

I will look forward to the release of the LAMMPS with Streitz-Mintmire potential and will let you know how the energies compare with GULP. Is there an estimate of timeframe for the release?

Do you also have plans to implement the original Rappe and Goddard’s QEq that considers ns Slater orbitals?

Thanks for the help.

Fatih

I stand corrected: fix qeq/slater does not calculate S-M energies but only charge derivatives of the energy terms. So there is no way that fix qeq/slater can output these energies.

When S-M potential is released, you can slightly modify the src code to output the energy terms you want. However, there is no timeline for the release. We currently don’t have plans to implement the R-G ns Slater orbitals.

Cheers,
Ray