[lammps-users] Special Bonds 1-3 Interactions

Sorry - this message was incomplete when I sent it:

Will - see the 5Aug10 patch. Please check that
the logic I added did not break anything - i.e. the
angle option works as you expect in your tests.

Steve

Steve,

I tested the new executable and all seems to be working as expected.

Your first point (a) definitely seems necessary. As for part (b), let me see if I understand the point you are making…

Do you mean the case where atom i and j are involved in both a 1-3 and 1-4 pair based on bonding topology? In this case, it seems the user should be allowed to specify which atoms pairs are included in the 1-3 and 1-4 lists by using the “special_bonds…angle yes and dihedral yes” option. If these options are not chosen (which is the default), or if one option is chosen but not the other, I’m not what sure what the best procedure is. At least it’s not obvious to me…

Will

Do you mean the case where atom i and j are involved in both a 1-3 and 1-4
pair based on bonding topology? In this case, it seems the user should be
allowed to specify which atoms pairs are included in the 1-3 and 1-4 lists
by using the "special_bonds...angle yes and dihedral yes" option. If these
options are not chosen (which is the default), or if one option is chosen
but not the other, I'm not what sure what the best procedure is. At least
it's not obvious to me...

The issue is this:

a) currently, with the changes we just made, with the new "angle"
option, if two atoms in an angle but are not listed in an angle, but
are 1,3 atoms in a dihedral, then the 1-3 weighting factor will
not be applied

b) another option would be to only not apply the 1-3 weighting
if the two atoms do not appear in either the angle list or
the dihedral list (as 1,3 atoms)

I think (b) is probably more consistent with what users would expect.
But I'm not sure what other force fields do.

Steve

Option (b) makes more sense to me. I’m no expert on every commonly used force field out there, but it seems in general a weighting factors are only sometimes applied on 1-4 pairs and rarely on 1-3 pairs…so this might be a pretty isolated case. Nonetheless, option (b) certainly seems a better alternative to (a). In the event that a user does want to use weighting factors for 1-3 or 1-4 pairs and is also using some unusual three-body terms (requiring the use of “dehedral yes” and/or “angle yes” options), option (b) would probably give the user more direct control over the implementation they desire.

Will

I added a 9Aug10 patch that adds this additional dihedral check.
Please run your test with the new "angle" option one more
time to see if it still works?

Thanks,
Steve

Steve, all seems to be working as expected. Thanks, Will