Hi,

This is sort of a follow-up to my post from two years ago, so I'm

putting it in the same thread.

http://lammps.sandia.gov/threads/msg15491.html

Previously we had worked out that you could use a massless site with

fix rigid, by modifying atom.cpp to comment out the lines which check

for the zero mass.

I am working with a 3 site N2 model, with the third site at the center

of mass which contains a partial charge.

The simulation runs and appears to be correct, so long as fix rigid is

used on the entire rigid body and the initial velocities are only set

for the non-massless sites.

My questions today are about the temperature compute.

1) Should the temperature compute be only on the non-massless sites,

or on the entire rigid body? I had presumed because doing it either

way gives the same kinetic energy it is ok either way. However the

temperature changes of course because of the change in degrees of

freedom. Checking the velocities, it appears that fix rigid assigns

velocities to all three sites in the rigid body, even though the

initial temperature was set to zero for the massless site. Should

this be the case? If initial velocites are set for the massless site

initially the simulation crashes, but runs with the massless site

initially at v=0. Since fix rigid is assigning velocites to the

massless site, I believe then that the temperature compute needs to be

on the entire rigid body, but might be wrong.

2) When using thermo_style one, setting the compute only on the

non-massless site gives a different temperature compared to using it

on the entire rigid body (although the kinetic energies are the same

as I mentioned). However, when using thermo_style multi, the

temperature is always the same no matter what I choose for the

temperature compute. Is this a bug? I wouldn't think changing the

thermo_style only would change the result.

I have placed example input and output here. Two examples are given.

Both are for the thermo_style one case (as I mentioned thermo_style

multi gives the same result). The thermo_style multi case agrees with

the case which includes the entire rigid body for thermo_style one.

http://gubbins.ncsu.edu/users/joshua/LAMMPS_N2_Question/nitrogen_question_lammps.tar.gz

One case uses only the N atoms

compute new nitrogen temp

thermo_modify temp new

while the other case uses both the N atoms and the massless site

compute new nitrogen_plus temp

thermo_modify temp new

Thanks in advance for any advice about the compute temperature you can give.

Josh