I have a question about why the Coulombic pairwise energy is so different for two kinds of rigid water implementations.
I simulated 4200 SPC/E water molecules in a periodic cubic box.
(1) fix npt, fix shake and fix pppm.
(2) fix rigid/npt/small molecule + fix pppm with the same LJ parameters and cutoff.
On step 0, I found that the long-range kspace energy for (1) and (2) are exactly the same (-219023.25 kcal/mol) while there is a huge difference in the Coulombic pairwise energy: 219648.08 kcal/mol for (1) and -628914.35 kcal/mol for (2).
Maybe it is not appropriate to simulate SPC/E water using fix rigid. If these two kinds of implementations are equivalent, one can use the second method to simulate tip5p water in LAMMPS with a
almost massless charge. Since fix shake can be used only for up to 4 atoms.
Many thanks in advance!