# Problem with "fix momentum angular"

Hello

I use “fix ID group-ID momentum N angular” for two condition
1- when I only have one OH in a box and " fix 1 all momentum 500 angular"
2- when I have one OH and argon atoms in the box “fix 1 Hydroxyl momentum 500 angular”
I use that fix to remove the rotation of OH every N step, for first condition that I only have OH in the box when I calculate the angular momentum square , it’s completely zero 0.00000000000E+000 but when I use this fix for the second condition (OH and argon in the box) and then calculate the angular momentum square it’s not completely zero 0.0000155 and the rotational energy is not completely zero.

I don’t know how lammps removes the rotational part, does it first remove the center of mass position and velocity (put the center of mass of molecule at the origin) and then subtract the rotation or I have to use this fix for linear momentum as well ? “fix ID hydroxyl momentum 500 linear 1 1 1 angular”

Best,
Leila

Hello

I use "fix ID group-ID momentum N angular" for two condition
1- when I only have one OH in a box and " fix 1 all momentum 500 angular"
2- when I have one OH and argon atoms in the box "fix 1 Hydroxyl momentum
500 angular"

I use that fix to remove the rotation of OH every N step, for first
condition that I only have OH in the box when I calculate the angular
momentum square , it's completely zero 0.00000000000E+000 but when I use
this fix for the second condition (OH and argon in the box) and then
calculate the angular momentum square it's not completely zero 0.0000155 and
the rotational energy is not completely zero.

I don't know how lammps removes the rotational part, does it first remove
the center of mass position and velocity (put the center of mass of molecule
at the origin) and then subtract the rotation or I have to use this fix for
linear momentum as well ? "fix ID hydroxyl momentum 500 linear 1 1 1
angular"

LAMMPS did what you ask it to do, i.e. it removed the angular
velocity, i.e. velocity that is left over after the center of mass
velocity. since you don't do it in every step, it can build it up
again, if there are sufficient forces. that is just how things work.

it is not quite clear from your description, what exactly you are
after and how you implement this, but it appears to me, that a smarter
way to handle the situation would be to remove angular velocities only
at the very beginning of the simulation, and then subtract the
_forces_ that would initiate a rotation in _every_ step with a
suitably computed atom style variable using fix addforce or fix
setforce. using fix momentum for this is abuse and since you seem very
concerned about small errors, a big problem, since every time fix
momentum is activated, you give your molecule a kick. that is very
unphysical and can ruin whatever you are looking for.

axel.

Axel,

Thank you.

Best,
Leila