Choice of thermostat during production MD: Nosé-Hoover vs. V-Rescale

User discussions
1

GROMACS version: 2020.4
GROMACS modification: No

Dear GROMACS User Forum,

Is the combination of Nosé-Hoover thermostat / Parrinello-Rahman barostat preferable over the combination of V-Rescale thermostat / Parrinello-Rahman barostat, during production MD?

Do you have any suggestions on how I should proceed if my system is stable with V-Rescale / Parrinello-Rahman, but not with Nosé-Hoover / Parrinello-Rahman?

Thank you so much for any comments!

Happy New Year & Best Regards,
Stefan

1 Like
2

Both thermostats produce valid canonical ensembles. NH allows for greater fluctuations so if your system is unstable with it, use V-rescale, though with sufficient equilibration, either thermostat should produce a stable simulation.

2 Likes
3

@jalemkul sorry to revive an old topic, but I want to ask the same question specifically in the context of a lipid bilayer simulation. In the KALP15 / DPPC tutorial you use v-rescale during nvt equilibration, then switch to Nose-Hoover for npt equilibration and production. The comment says “More accurate thermostat”. Is this still your recommendation for lipid bilayers?

4

That tutorial reflected the state of the art about 15 years ago. Its methods are somewhat dated and algorithms exist now that didn’t then. Decide based on best practices from the literature, not the tutorial (which no longer even works with files being removed from Peter Tieleman’s website).