Fix addtorque

LAMMPS LAMMPS Mailing List Mirror
1

Dear all,

Recently, I wrote a fix style to calculate the point induced dipole moment for polarization. Since the induced dipole under external efield will add translational force and rotational torque, I need to add them to the system. I find a fix style named fix addtorque which adds torque to atom sites. I wonder if I add torque to atom->torque:

  1. Do I must use fix shake or fix rigid to ensure the molecule move rigidly?

  2. How does LAMMPS use the torque for post fix style?

  3. Is it enough to use full atom style and regular nve or nvt thermostat but not the one with rotation like fix nve/sphere?

Thanks in advance

Han

2

Dear all,

Recently, I wrote a fix style to calculate the point induced dipole moment
for polarization. Since the induced dipole under external efield will add
translational force and rotational torque, I need to add them to the system.
I find a fix style named fix addtorque which adds torque to atom sites. I
wonder if I add torque to atom->torque:

please note that the atom->torque property is only meant to be used
with "extended" particles, i.e. atom style sphere or ellipsoid or body
or line or tri, where the integrator will integrate not only the
center of mass, but also the rotation around them. for most molecular
cases, however, you have point particles and those cannot rotate.

1. Do I must use fix shake or fix rigid to ensure the molecule move rigidly?

the answer to this question is independent from your fix. your fix
should only add forces to molecules. this essentially the inverse of
what fix rigid does (which takes the per-atom forces and converts them
into a center of mass and a rotational force). if there are no other
forces, then the forces from your fix should induce a rotation and
nothing else. this is what fix addtorque does. however, in your case
you cannot directly use the code from fix addtorque, but will have to
do something much more complex. fix addtorque operates on a single
group and the number of groups is limited in LAMMPS, and thus you most
likely would rather make your definition of what constitutes an object
responding to the torque be based on molecule ids. for that you'd have
to compute per molecule center of masses and moments of inertia and
then project out the forces to the constituent atoms. as molecules can
stretch across multiple parallel subdomains, some communication is
needed. most likely, you want to do something similar to what is done
in fix rigid/small to collect this information.

2. How does LAMMPS use the torque for post fix style?

i have no idea what you mean by "post fix style". please reformulate
your question.

3. Is it enough to use full atom style and regular nve or nvt thermostat but
not the one with rotation like fix nve/sphere?

fix nve/sphere is for extended particles. that doesn't seem to be
applicable for your case.

axel.

3

I’ll add that if you have a rigid body, the per-atom torques

are applied to generate torque on the rigid body. However

fix shake does not do that.

Steve