Free energy surface with PIMD and Metadynamics using i-PI plugged to plumed

[vvassile...@gmail.com]

Dear all,

I would like to compare the free energy surface (FES) projected onto a collective variable between a classical molecular dynamics (n_beads = 1) and a path integral molecular dynamics (PIMD). To this end, I'm using the existing interface between plumed and i-PI to run metadynamics simulations.

In this regard, I would like to know if, for obtaining the FES from PIMD Metadynamics, I only need to use the default procedure of plumed (i.e., use sum_hills on the HILLS output written by plumed) or if I need to do different post-processing of my simulation outputs.

Also, in the zundel example provided with i-PI, a PIMD with n_beads=8 is run but the n_beads within the "<bias>" tag is set to 1. Hence, I would like to know what should be the number of beads within the "<bias>" tag in "<ensemble" if I'm running a 16-bead simulation. 

I will be very thankful for any help you can provide.

Best regards,

Valentin

[marian...@gmail.com]

Hi Valentin,

The bias is applied on the centroid - and I think this is all that is implemented // makes sense to do at the moment. So on the plumed side everything should work as usual, and the only thing is that applying the bias to the centroid may not be what you want to do at very low temperatures, depending on the situation.

Then I think more implementation (but potentially also more derivations) would be needed.

Disclaimer: I am not a power user of the functionality, so maybe Michele can say more or correct me, but this is what I remember from when I used it.

Cheers,

Mariana

[vvassile...@gmail.com]

Hi Mariana,

Thank you for your prompt reply.

Michele reached me by private e-mail and told me practically the same. So, I will do <bias n_beads="1"> to apply the bias only on the centroid as you both suggested.

Best,

Valentin