Does metadynamics on PLUMED work in an NPT ensemble?

This is a conceptual, statistical mechanics question I have about metadynamics: does metadynamics, as implemented on PLUMED, work in an NPT ensemble? Specifically, using the Parinello-Rahman barostat as in fix npt?

I am currently running a polymer in solvent simulation, and I will be running a free energy calculation over radius of gyration. My vanilla MD simulations were in the NPT ensemble. I hope to use metadynamics to get a free energy profile over Rg.

PRINT ARG=rg,restraint.bias STRIDE=1000 FILE=COLVAR

When performing metadynamics, we introduce additional energies in the system in the form of Guassian potentials. The way I am thinking about it, this will likely corrupt the computation of the virial – leading to unsound evolution of the system. Is there a flaw in my thinking, or this is a real problem but the PLUMED devs have figured this out?

I would appreciate any advice you have for me!

It is a conceptual, statistical mechanics question indeed. Please repost it in the Science Talk section (Science Talk - Materials Science Community Discourse), where it might also attract more diverse expert opinion.

Your chance to get a competent and useful response would be much larger in a forum about Plumed. This is not really a LAMMPS question…

Understood. I did that and I got a satisfactory answer from Gareth Tribello:

You can run your simulation with NPT. The bias makes a contribution to the virial but we calculate that contribution in plumed and pass it back to the md code so it can incorporate it when it integrates the equations of motion.

1 Like