Hi, I am simulating a system that consists of spheres with embedded
dipoles. I have been using fix nve/sphere update dipole as an
integrator with fix langevin omega yes as my thermostat.
However, I'd like to use the Nose Hoover thermostat on the same
system. I understand that I need to use fix nvt/sphere and that
includes both integrator and thermostat. Is there an equivalent in
fix nvt/sphere for the "update dipole" part of the fix nve/sphere
integrator or can I not use fix nvt/sphere with dipoles?
I'm afraid I can't find an answer either way in the fix nvt/sphere
part of the manual.
no there isn't - someone would have to add it.
It's unlikely that langevin versus Nose/Hoover
will lead to significantly different dynamics.
Thanks. I realise the choice of thermostat should ideally not make
too much difference - I was hoping to be able to explore whether the
possible bug with the Langevin thermostat that Joachim Stenhammar
raised on 15 Aug and 21 Aug was actually an issue for me without
having to dig too much into the code myself.
(3x3x3 lattice of tiny two-atom dumbell moleculeces with periodic
boundary conditions. Same system simulated using langevin or
nose-hoover. Very dilute system. I wish I could find the lammps
input script I used.)
langevin_dumbell_lattice.avi (637 KB)
nosehoover_dumbell_lattice.avi (608 KB)
yes - I need to fix that bug - hopefully next week before
the holiday break.
I'm thinking Andrew is being tongue-in-cheek here.
Thermostatting a system with 2 particles is not what
I had in mind. Rather I meant that the statistics
of a big system will be indistinguishable with the
two thermostats, not the fine-scale dynamics.