Using separate thermostat groups in non-equilibrium MD

GROMACS version: 2024.4
GROMACS modification: No

I understand that multiple thermostat groups are generally considered unnecessary for most equilibrium MD simulations, especially with modern algorithms like V-rescale. However, I am curious about the best practices for non-equilibrium MD simulations, particularly for steered molecular dynamics (SMD) of peptides in lipid bilayers.

In an SMD setup where a peptide is being pulled through a membrane, would it be advisable to define the peptide as a separate temperature coupling group (tc_grps), while the rest of the system (membrane, water, and ions) forms another group? The idea is to maintain accurate temperature control for the peptide, which is experiencing direct external force and might be prone to local heating, without affecting the temperature of the bulk membrane or solvent.

At the same time, I am aware that using too many small thermostat groups can introduce artifacts, and GROMACS even warns against any group having fewer than 10% of the total atoms. The peptide in this case would be less than 10% of the total system, which triggers the warning. But since this is a non-equilibrium setup (with the peptide being actively perturbed), would the benefit of independent temperature control outweigh the risk of artifacts?

I would appreciate any insights.

1 Like

To get meaningful results with steered MD, the pull rate needs to be so low that you are near equilibrium, so there should not be an issue there either.

Thank you so much for your helpful response.