What is it about ewald/n that is different than PPPM for
this problem? Are you including long-range LJ terms to
get a better answer? If not, I don't see why ewald/n
would be preferable.
Steve
Hi Steve,
in my standard production runs I use P3M (for efficiency reasons). However,
it doesn't allow for tilted sim boxes which is required to get e.g. c44 in
the "elastic" example included with LAMMPS. That's where ewald/n comes into
play.
Actually I did try the buck/coul/cut style instead of buck/coul/long, but
the elastic constants hardly resembled those from literature. I guess I
would have to refit the potential entirely to get a good agreement when
using no k-space solver at all.
For determining the bulk modulus and my initial problem, however, it
shouldn't matter which solver I use.
Best regards,
Patrick