Dear lammps users,
Hello,
I have a question about the effects of damping parameter in nvt command.
When we use “fix nvt” to control temperature, we can make
fix ID all nvt temp (T_ini) (T_fin) (t_damp)
, where t_damp is the temperature damping parameter.
I know that this parameter is the timescale of thermal evolution.
When t_damp is small, the thermal evolution is fast. (Is this correct?)
I found some websites that say “t_damp should be about 100 times as large as Δt”.
Are these websites correct?
I want to know how to take the value of t_damp, and what effects exist when I change t_damp.
I worry that the precision is bad for too small t_damp.
Thank you.
For the exact details of what this parameter does you should read the related publications (not an easy read, though).
A gross simplification is that it controls how strictly the temperature of your system is coupled to the thermostat heat bath which in turn is represented by a chain of (fictional) harmonic oscillators. Since the coupling is done via modification of the acceleration in the time integration, it has to be chosen in such a way that a) the time integration of the fictional degrees of freedom is stable and b) that the coupling is meaningful and not distorting your trajectory too much. Thus, in most cases, using a larger t_damp will lead to a weaker coupling, but it will also take a longer time until the target temperature is reached in case there is a difference between the current and the target. For typical cases (e.g. condensed, liquid molecular systems at ambient conditions with > 100000 atoms) a choice of 100-1000x the length of a timestep is a often a meaningful choice. You have to avoid feedback with the Nose-Hoover chains.
Please note that in most cases a thermostat algorithm in MD is used to reduce the gradual increase of kinetic energy (i.e. Temperature) due to the numerical noise from floating-point math and inconsistencies from using cutoffs etc.
A much bigger concern for accuracy is actually the overall energy conservation of the simulation. That is, when your system has reached equilibrium, it should conserve total energy to a sufficient degree. This needs to be tested with fix nve, since using fix nvt will mask it; in fact, if you still have an increase of temperature with a thermostat, then you have a serious problem. For details on the importance of energy conservation, I suggest to revert to your MD text book. The optimal choice for the damping constant should be determined only after energy conservation with fix nve has been established. And then it should be chosen to maintain he time and ensemble average temperature as needed.
Please note that fix nvt also allows to monitor the (net) amount of energy exchanged between the heat batch and the simulated system, so you can monitor if there are issues.
1 Like