Dear fellow lammpers,
We have been developing a novel approach (we hope) to calculate the chemical potential of various systems, and I have been testing it for simple lennard jones systems. When I use pair_style lj/cut with a suitable large cut-off (e.g. Rc > 4.5 in lj units) I get essentially exact results. However, when I use
pair-style lj/long/coul/long to as to treat the vdw long range interaction with PMME - I systematically get a larger value of the excess chemical potential than I expect, even when the estimated errors are negligible. I had a hunch that the
lj/long/coul/long is using the minima of the lennard jones as its lj scale paramater instead of sigma.
And if I assume this is the case, I get the correct results. Is it possible that the code developers made a bobo/mix-up in their description of this parameter. I note that they declare "Note that sigma is defined in the LJ formula as the zero-crossing distance for the potential, not as the energy minimum at 2^(1/6) sigma ", but if I am correct, they meant the opposite.
Is my hunch correct, or is there something else gone awry ?
The relevant part of the lammps script is below ( for a short test run) but the real script is considerably longer and more involved.
Thanks for any insight to this mystery,
Best Regards,
Donal
Dear fellow lammpers,
We have been developing a novel approach (we hope) to calculate the
chemical potential of various systems, and I have been testing it for simple
lennard jones systems. When I use pair_style lj/cut with a suitable large
cut-off (e.g. Rc > 4.5 in lj units) I get essentially exact results.
However, when I use
pair-style lj/long/coul/long to as to treat the vdw long range interaction
with PMME - I systematically get a larger value of the excess chemical
potential than I expect, even when the estimated errors are negligible. I
had a hunch that the
lj/long/coul/long is using the minima of the lennard jones as its lj scale
paramater instead of sigma.
And if I assume this is the case, I get the correct results. Is it possible
that the code developers made a bobo/mix-up in their description of this
parameter. I note that they declare "Note that sigma is defined in the LJ
formula as the zero-crossing distance for the potential, not as the energy
minimum at 2^(1/6) sigma ", but if I am correct, they meant the opposite.
Is my hunch correct, or is there something else gone awry ?
that is difficult to verify from the source code, because it was
written by a person with a very unique programming style, that i have
a very hard time making sense of unless i am of under the influence of
mind altering drugs.
however, i modified the melt example from:
atom_style atomic
pair_style lj/cut 2.5
pair_coeff 1 1 1.0 1.0 2.5
to
pair_style lj/long/coul/long cut off 2.5
pair_coeff 1 1 1.0 1.0 2.5
kspace_style pppm/disp 1.0e-5
and they both result in the same same energies for lj/cut with charges
turned off. this suggests, that the LJ sigma seems ok.
axel.