PLUMED/QMMM-CP2K

[SAhmadkhani]

Dear PLUMED developes,

I am using PLUMED to connect a machine-learning-trained surface potential to the MM region in a CP2K QM/MM simulation. My understanding is that, in this setup, PLUMED effectively applies an additional bias potential to the MM subsystem.

However, I am unsure how this bias potential is incorporated into the CP2K Hamiltonian. In particular, since my ML model is coarse-grained and coupled to the QM region, I am not clear how the corresponding energy terms are consistently included, especially given the intractable components of the full Hamiltonian.

I would greatly appreciate any clarification or suggestions.

Best regards,
Somayeh Ahmadkhani