[EXTERNAL] PPPM virial issue

Yes of course, with accuracy 1e-9 I get:

analytical virial 2171039680
numerical virial 2052555954.
With accuracy 1e-10:
analytical virial 2171039591
numerical virial 1997392351
And with accuracy 1e-11:

analytical virial 2171039380
numerical virial 1933876509.

So basically you are saying that the virial of Ewald and PPPM aren’t converging, even when you increase the accuracy? That would be really surprising (and unlikely). Can you post more info, like LAMMPS version and input script?