I have several systems of a single small aromatic molecule in a box of solvent. I equilibrated the system using a regular MD simulation (all atoms treated with MM). The box is well equilibrated and has a reasonably constant volume for the 20 ns simulation. However, if I then take a snapshot from that simulation and start a QM/MM dynamics simulation, the volume immediately starts to decrease. Sometimes, this results in such large energy changes that CHARMM exits. Is this typical for a QM/MM dynamics simulation? I thought since the system was well equilibrated with regular MM dynamics, it would remain approximately the same volume during the much shorter QM/MM dynamics simulation.

My current theory is that when you change the Hamiltonian to QM/MM, when the pressure is calculated, it thinks the QM atoms have zero energy (or it doesn't realize they are there at all), so it thinks the pressure is way too low and that it needs to decrease the volume. I don't see a way around this if this is what is happening, so since I know the simulation is well equilibrated, I'm going to try doing an NVT simulation now. But, if someone else knows for sure how the pressure calculation treats the QM atoms, that would be very helpful.