tip4p pair_style with triclinic cell

Dear lammps developers

The tip4p pair_style does seem to have a problem with a triclinic cell.
I used the following commands

pair_style lj/cut/tip4p/long 2 1 1 1 0.1546 8.5
kspace_style pppm/tip4p 1.0e-6

when I use the tip4p pair_style, I get a weird energy, while for the same structure, tip3p and spc pair_styles do a good job.

I would be delightful if anybody can check and solve this problem.

With best regards
Soroush

Dear lammps developers

The tip4p pair_style does seem to have a problem with a triclinic cell.
I used the following commands

pair_style lj/cut/tip4p/long 2 1 1 1 0.1546 8.5
kspace_style pppm/tip4p 1.0e-6

when I use the tip4p pair_style, I get a weird energy, while for the same
structure, tip3p and spc pair_styles do a good job.

I would be delightful if anybody can check and solve this problem.

​that is very unlikely to happen unless you have done a better ​job of
documenting your issue and make a more convincing and reproducible(!) case
that this isn't an issue of you using the tip4p style incorrectly.

also, please see: http://lammps.sandia.gov/guidelines.html

​axel.​

The problem of tip4p force field seems to get rise when I use the triclinic cell, with the following reasons:
1- At first I simulated an orthogonal cell without triclinic parameter, then I simulated the same structure with the same tip4p parameters except this time I added zero triclinic parameters to the geometry file. When I compare the energy between these two simulations, I found a gap of energy about 4 Kcal/mol.
2- I used the following references to reproduce energy of water clusters in an orthogonal cells with tip4p potential.

Are you using the latest version of LAMMPS? We fixed a bug with tip4p and triclinic not too long ago that sounds similar: https://github.com/lammps/lammps/issues/299

Yes, I am using the last version of LAMMPS.
Also in tip4p force field parameters, I used the long range electric potential with tip4p kspace solver, which was only implemented in the recent versions of LAMMPS.