Is the newton force pair calculate correctly in tip4p water model when periodic boundary condition exists?

Dear lammps users,

I am worried about it because I have done some tests with pbc.I put one water molecule in the center of the cell, and at the boundary of the cell. The force calculated are different, which I think should be the same.

Can you explain it?

I print out the force acted on the atoms after the reverse_comm().

With the water in the middle of the box, the result I get is

Dear lammps users,

I am worried about it because I have done some tests with pbc.I put one
water molecule in the center of the cell, and at the boundary of the cell.
The force calculated are different, which I think should be the same.

Can you explain it?

there are two components:
1) pppm is not completely position invariant.
2) the parameter estimator for kspace does not work well for low charge density.

so:
- use a tighter convergence for pppm
- use the estimator for pppm with a normal density water system and
look up the resulting parameters (grid, gewald) and then use those.

axel.

> Dear lammps users,
>
> I am worried about it because I have done some tests with pbc.I put one
> water molecule in the center of the cell, and at the boundary of the
cell.
> The force calculated are different, which I think should be the same.
>
> Can you explain it?

there are two components:
1) pppm is not completely position invariant.
2) the parameter estimator for kspace does not work well for low charge
density.

Thank you for your help. The tighter convergence can solve the problem. But
do you know is there a way to change these parameters, or I have to coding
for it?

> Dear lammps users,
>
> I am worried about it because I have done some tests with pbc.I put one
> water molecule in the center of the cell, and at the boundary of the
> cell.
> The force calculated are different, which I think should be the same.
>
> Can you explain it?

there are two components:
1) pppm is not completely position invariant.
2) the parameter estimator for kspace does not work well for low charge
density.

Thank you for your help. The tighter convergence can solve the problem. But
do you know is there a way to change these parameters, or I have to coding
for it?

just follow the second suggestion in combination with the first.

axel.