Dear All,
As there is no implementation of the Anderson thermostat in
LAMMPS I wanted to realize it via drawing the velocities of all the atoms in a giving time intervals (MD steps) from the Gaussian distribution as the lammps runs a simulation.
But unfortunately I cannot find how to make lammps reset the velocities in say 100 MD steps withing a single run of say 10000 MD steps without interrupting this run and starting another one after resetting the velocities using velocity-command.
I would appreciate very much any hint.
Thank You!
Dear All,
As there is no implementation of the Anderson thermostat in
LAMMPS I wanted to realize it via drawing the velocities of all the atoms in
a giving time intervals (MD steps) from the Gaussian distribution as the
lammps runs a simulation.
But unfortunately I cannot find how to make lammps reset the velocities in
say 100 MD steps withing a single run of say 10000 MD steps without
interrupting this run and starting another one after resetting the
velocities using velocity-command.
I would appreciate very much any hint.
in my personal opinion (which is rather biased), it would be best to
use temp/csvr for this. an anderson thermostat gives your system a
random kick every time it is run, which can have all kinds of unwanted
side effects (temp/rescale is even worse, but that is a different
story).
temp/csvr works in a similar fashion, i.e. it creates a new gaussian
velocity distribution, but then rather gently mixes it with the
existing distribution through a linear combination with boltzman
factor based weighting coefficients, similar to a berendsen
thermostat. so through the time constant, you can smoothly adjust how
strongly you want you couple your system.
axel.
Dear Axel,
Thank you for the reply!
I completely agree with you.
I just need the Anderson thermostat in order to compare the results with one paper from other guys.
Sincerely,
Anton
it looks like I can reset the velocities every given number of steps using the keyword “every” of the “run” command.