## Thursday, April 9, 2015

### LAMMPS temperature increase nvt : Noose-Hoover thermostating

The format of the most basic NVT instruction in LAMMPS is:
fix fix-ID group-ID nvt temp Tstart Tstop Tdamp
Example:
fix 1 all nvt temp 20.0 1200.0 0.05
[the units used here is 'units metal', i.e., T is in kelvin, time is in picosecond].
Here, the highlighted texts are the keywords one needs to use. In LAMMPS, 'fix' is a task, and thus fix-ID is the task ID. This ID can be used later to extract information from the output of the task (i.e., fix). The group-ID is the ID of the group of atoms on which the task will be applied. In the example, by 'all' it means the fix is applied to all the atoms.
Tstart is the starting temperature, and Tstop is the 'target' temperature (indeed not 'stopping' temperature. 'Tstop' is a confusing name used in LAMMPS manual). If the simulation is not run for for sufficient no. of timesteps, the temperature may not reach the targeted temperature. In that case, one can restart a run (for that one has to generate a restart file while doing the nvt run with the command 'write_restart').
The parameter Tdamp is related to Noose-Hoover thermostating that takes place during nvt run. The value of Tdamp has to be optimized: too small a value will cause erratic temperature fluctuation, and a large value will delay the equilibration. LAMMPS manual says a good choice for many models is Tdamp = 100 * timesteps. Thus, if the timestep is 0.001 ps (= 1 fs), then Tdamp = 100 * 0.001 ps = 0.1 ps. Note that Tdamp is not equal to 100 * time units, which for 'units metal' would be 100 ps.

Ref: http://mycomputinglogs.blogspot.fr/2013/04/nvt-run-with-lammps.html