Handling 1-4 scaling with charmm style

Dear LAMMPS users,

I am simulating the linear/cyclic alkane using Trappe united atom model whose 1-4 scaling is 0.0

For sometime, please assume it has full 1-4 scaling. ( I have a valid reason behind it, and do not want you to deviate from the actual problem)

I am using pair_style lj/charmm/coul/charmm 14.0 15.0 ( using coul/charmm for chargeless is nevertheless stupidity)

Now, Full scaling can be handled in two ways

(i) special bonds lj 0.0 0.0 1.0

(ii) Charmm pair routine has explicit evaluation of 1-4 using dihedral weighting factor. I can have a dihedral weighting factor of 1.0 with no special_bonds.

I have checked for linear alkanes, this is working.

But when I used same principle for cyclic alkanes it breaks, cannot figure it out after long time.

I have attached the files to produce the difference in energy for cyclic molecules. Can anyone suggest me, whats going wrong?


files.tar (70 KB)

I’m not clear on what you are asking.
(i) vs (ii) are separate issues. If you set
the 1-4 weights to 0.0 via special_bonds,
then that turns off pairwise interactions
between all 1-4 atoms, whether they
are listed in dihedral interactions or not,
so that the calculation of the pair style ignores them.

The 1-4 weight in dihedral_style charmm
applies to a calculation done by
the dihedral style of a pairwise interaction
which is calculated inside the dihedral method.
It applies only to the 1-4 atoms in the
dihedrals you list in topology interactions.

For cylclic molecule, e.g. a 6-ring, there are
2 possible dihedrals that have the same 2 end
atoms, e.g. 1-2-3-4 and 1-6-5-4. The pairwise
calculation will only consider the 1-4 pairwise
interaction once and weight it via special_bonds.
You will likely want to list both dihedrals in
your topology list. If the dihedral weight is non-zero,
that means the 1-4 pairwise interaction (within
the dihedral) will thus be computed twice, once
by each listed dihedral.


Thanks steve.

This helped in figuring out the solution I wanted.