[lammps-users] Will "fix nvt + fix temp/rescale" cause any artifact?

Hi Guozhen,

What Xiaopeng says seems like a good advice.
Also, please note the documentation regarding the "Tdamp" keyword. If this is not chosen properly, it can cause really bad thermostatting.

From the documentation regarding NPT/NVT etc.:

"IMPORTANT NOTE: A Nose-Hoover thermostat will not work well for arbitrary values of Tdamp. If Tdamp is too small, the temperature can fluctuate wildly; if it is too large, the temperature will take a very long time to equilibrate. A good choice for many models is a Tdamp of around 100 timesteps. Note that this is NOT the same as 100 time units for most units settings. "

For instance, for my simulations, I use a timestep of 0.001 (in picoseconds). Then, I use a value of Tdamp that is 0.1 - which is 100 times 0.001. Previously I used a Tdamp of 100 (I misunderstood the documentation), and it caused the thermostat to work very poorly (big fluctuations, and it took very long time to reach desired temperature).

The same goes for the Pdamp value (if you use NPT in your simulations). It should be about 1000 timesteps, so if you have (for example) a timestep size of 0.001 picoseconds, your Pdamp should be 1.0.

Hope this helps you.

Sincerely,
Christer H. Ersland.

Hi Christer,

Thanks a lot. I indeed set Tdamp as “100”, mainly because this is the value given in the example in the manual, which may be misleading to some extent. However, the fluctuation of T looks reasonable, which may be attributed to the using of “fix temp/rescale” coincidently.

Sincerely,
Guozhen

the manual, which may be misleading to some
extent.

Tdamp is in units of time. If you use different units (e.g. real
vs metal vs si) then time units change, so you have to think
about what Tdamp should be. The manual doesn't list a version
of the command for every units choice LAMMPS offers.

The fix nvt doc page also says this:
IMPORTANT NOTE: A Nose-Hoover thermostat will not work well for
arbitrary values of Tdamp. If Tdamp is too small, the temperature can
fluctuate wildly; if it is too large, the temperature will take a very
long time to equilibrate. A good choice for many models is a Tdamp of
around 100 timesteps. Note that this is NOT the same as 100 time units
for most units settings.

Steve

Hi, Steve

Thanks for the comment. I just checked the time unit of the “real” style, it is in fs.
So 100 fs is proper. And the ps is the time unit of the “metal” style. That’s probably what Christer has used in his work.

Regards,
Guozhen