Hi all:

I have a question regarding kspace modify in lammps, specifically about “ kspace modify slab” option. Considering the following 2 simulations with the same atomic configurations. The first one uses non-periodic boundary condition in z and kspace modify slab 3.0. The second one uses PBC in all directions but has a vacuum space above the sample in z direction so that lz_new = lz_old*v_factor. To let the two simulations yield the same energy, v_factor should be 3 or 4 ? Our result indicate that v_factor = 4. But looking at the source code we guess v_factor should be 3.

Thanks

Deena

Hi all:

I have a question regarding kspace modify in lammps, specifically about “

kspace modify slab” option. Considering the following 2 simulations with

the same atomic configurations. The first one uses non-periodic boundary

condition in z and kspace modify slab 3.0. The second one uses PBC in all

directions but has a vacuum space above the sample in z direction so that

lz_new = lz_old*v_factor. To let the two simulations yield the same energy,

v_factor should be 3 or 4 ? Our result indicate that v_factor = 4. But

looking at the source code we guess v_factor should be 3.

those two simulations are two different physical models, so the

energies are not comparable.

the requirement for tripling the volume in the non-periodic dimension

is a requirement for solving the poisson equation to an acceptable

accuracy to compute the dipole correction for decoupling the slabs.

making it larger can somewhat improve then accuracy in exchange for

additional computational effort, but the relative total energies have

no meaning in that respect.

axel.

Did you add the shape correction term 2\pi*M/(3V) to your 3d test case?