[lammps-users] A problem with lammps, How to set right thermostat parameters?

LAMMPS LAMMPS Mailing List Mirror
1

Dear Steve,

Thank you very much for your message. You were right; I was using Anderson method in Discover and Nose-Hoover in Lammps for the thermostat. Below is what you suggested. (I have ~2% difference in the initial pressure and maybe that is why the energy average is different).

Now the important question for me is how to set the thermostat parameter (Tdamp) correctly (the whole reason of these comparisons was that I have realized that a lot of my systems do not get to the eq situation after a long run (4-12 ns) and the block averages have a drift sometimes <10kcal/mol ). I always use 100 for Tdamp which might not be suitable for every system. I have also tried 1000 but have not seen any difference in the results.

Dear Tyler,

Thanks for the message. Lammps09 was that last version available on “Ohio Super Computer”. I will try to install that last version; however, I guess I have to learn first how to find the correct thermostat parameters …

image010.png

image013.png

image002.png

image008.png

image009.png

image011.png

image012.png

2

For Tdamp and Pdamp the rule-of-thumb I use is 100 timesteps
for Tdamp and 1000 for Pdamp. Note this is timesteps, but T and Pdamp
are specified in time, which will be different for different units.

You can also try other thermostats or a Berendsen barostat. Nose/Hoover
is tempermental when your system is not near equilibrium.

Steve

image012.png

image011.png

image009.png

image013.png

image010.png

image002.png

image008.png