ReaxFF and qeq/reax with atom_style != charge

LAMMPS LAMMPS Mailing List Mirror
1

Please post to the mail list, not to me directly.

Are you saying that you run a ReaxFF calculation
twice, the only difference is atom_style charge vs
atom_style full, and you get different behavior?

I don't think that should matter. If that is the case,
please post your (simple, small) input scripts.

Steve

2

Please post to the mail list, not to me directly.

Sorry, somehow didn’t notice that…

Are you saying that you run a ReaxFF calculation
twice, the only difference is atom_style charge vs
atom_style full, and you get different behavior?

I don’t think that should matter. If that is the case,
please post your (simple, small) input scripts.

Here are the files (ffield.reax is from the LAMMPS distribution), I’m running it like this:

lmp -var atstyle charge -var infile charge.data < CH.reax.lmp lmp -var atstyle full -var infile full.data < CH.reax.lmp

In the second case, minimization puts all each carbon’s hydrogens in the same place, and then the system explodes. And even with minimization turned off, the system explodes, as well.

Thank you in advance,
Vasilii

full.data (1.34 KB)

charge.data (1.03 KB)

ffield.reax (14.4 KB)

CH.reax.lmp (647 Bytes)

3

Your full.data file has bonds in it. Why would you
do that with ReaxFF? More to the point, the
special_bonds default will turn off Coulombic
interactions between bonded atoms. So you are
not running the same model with the 2 cases. Hence
charge equil does something different.

Steve

4

Your full.data file has bonds in it. Why would you
do that with ReaxFF?

My original intent has been to use QEq with molecular mechanics, ReaxFF was used for debugging, hence bonds were carried over from the original data file, assuming (incorrectly) that it would simply not make any difference with no bond_coeffs specified. I forgot about special_bonds, indeed. Leaving out the bonds or fixing special_bonds helps. Thank you very much!