My forcefield actually “wants” me to turn off ALL intramolecular pair interactions. This is where the problem lies, as its the atoms beyond 3 bonds away that still have a strong enough effect such that the dihedral angle ends up being 90 degrees rather than 0 degrees.
2 thoughts:
- I actually can’t use the method 3 from above (making a fully flexible molecule but with high harmonic constants on all bonds, angles, and dihedrals except the one dihedral I’m interested in), because I would still have the same problem of atoms from beyond 3 bonds away having an overpowering effect on the dihedral.
Thus, to my eye, it appears that there is no way for LAMMPS to handle this type of molecule (I’d be very happy if somebody could prove me wrong).
- However, I’ve come up with the following “hack,” which I’d like your thoughts on. Before this hack, I had 1 bond (CH3-O), 1 angle (CH3-O-C), and 1 dihedral (CH3-O-C-C), with the first 2 having high harmonic constants and the dihedral having the form given by OPLS. The “hack” involves adding 11 additional bonds, in which CH3 gets a “bond” to EVERY atom in the molecule that it wasn’t already bound to (the 6 Cs and the 5 Hs), each of which have a harmonic constant of 0.0. This way, “special_bonds lj/coul 0 0 0” turns off intermolecular interactions between CH3 and all other atoms in the molecule, making the only interactions that get counted the original 3 (the bond and angle with high harmonic constants and the dihedral).
I’ve tried this method out, and it sort of appears to do the trick; the dihedral angle oscillates around 0 degrees as it should and the rest of the molecule remains rigid, BUT my NPT simulation comes to a density rather far off from Jorgensen’s (1037 kg/m^3 instead of Jorgensen’s simulated 992). I’m curious as to whether this hack won’t severely mess up the long-range kspace solver (or something else) in some unanticipated manner; I don’t think it will, since I’m using “special_bonds” instead of “neigh_modify exclude,” but since this is a really strange hack, I can’t be sure that something is going on that I don’t see.
If anyone has a more proper way that LAMMPS should handle this type of molecule or has a comment on this hack, I’d much appreciate it.
CH3
/
O
H C H
\ / \ /
C C
C C
/ \ /
H C H
H