Can tune_pme change the Coulomb cutoff to be larger than the simulation box?

GROMACS version: 2022.2
GROMACS modification: No

The optimized Coulomb cutoff found by tune_pme for my simulaiton is 1.6 nm, but my simulation box has only a 1.5 nm buffer (probably slightly less after equilibration). This sounds like a problem to me. Should I disable tune_pme, make my simulation box larger, or box, or do something else?

PME always computes the full, periodic electrostatic interactions, independent of which cut-off is used. So there is no problem with using tune_pme. It could be that a buffer of 1.5 nm is not sufficient, but that is independent of the parameters for PME.

If the Coulomb cutoff determined by tune_pme is larger than the solvent buffer inside the simulation box, should I increase the solvent buffer to match the Coulomb cutoff?

Said another way: How can I tell if the solvent buffer size needs to be increased?

Note that my system is highly charged (140 K+ counterions for neutrality).

Best,

Matthew

Electrostatic interactions decay as 1/r. With PME the electrostatic cut-off is irrelevant, except for performance considerations.

What buffer you need if very difficult to determine. That comes own a lot to experience. If you have a high counterion density, there will be a lot of screening and the buffer might not be so critical. Another question is what ion concentration is reasonable.