PME is most reliable for a system that has no net charge, but there are provisions for cases where that is not the case, esp. for free energy simulations where the charge changes over the course of the simulation. However, I'm not certain exactly how it's done; it involves the QCOR keyword, and the default value is 0.0 for neutral systems, but it is not clear to me what to use for other cases.
BTW, the Ewald docs appear to incorrectly state the default for QCOR is 1.0; the source code now sets it to 0.0, since a solvated system is the usual case.
Last edited by rmv; 07/19/21 09:58 PM.