[lammps-users] PPPM calculation

LAMMPS LAMMPS Mailing List Mirror
1

Hi,

I find the PPPM method gave me a very different answer from the cut-off method when I calculated
the electrostatic potential between two charges in two dimentions. I don't think the result from
PPPM method is correct. Could anyone tell me what's wrong? My results are as following:

From cut-off method:

units lj #default
dimension 3
newton on #default
boundary p p f
atom_style charge
pair_style lj/cut/coul/cut 2.5 7.5
pair_coeff 1 1 0 5.0 5.61230
pair_modify shift yes
pair_write 1 1 23 r 4.5 10.0 pairPotential type11 1 1

The result (from pair_write command) is:

1 4.5 0.222222 0.0493827

This is correct.

From PPPM method:

units lj #default
dimension 3
newton on #default
boundary p p f
atom_style charge
pair_style lj/cut/coul/long 2.5 7.5
pair_coeff 1 1 0 5.0 5.61230
pair_modify shift yes
kspace_style pppm 0.001
kspace_modify slab 3.0

pair_write 1 1 23 r 4.5 10.0 pairPotential type11 1 1

The result (from pair_write command) is:

1 4.5 0.00626722 0.00917555

I don't know why PPPM gave ame such a different answer. Please help me figure it out.

Thank you in advance!

Dongsheng

2

Dongsheng,

The numbers actually don’t look too different. Granted, I don’t know what system you are trying to simulate. Bear in mind, when using long-ranged corrections, that the answers cannot directly be compared to a cutoff version. The cutoff version does not include the long-range tail since it truncates the potential at the requested cutoff. PPPM, as any Ewald-like solution, tries to correct for the long-range effect by summing over replica boxes by means of wave vectors. You can read up on Ewald sums and alike in i.e. “Understanding Molecular Simulations” by Frenkel and Smit and references therein.

Pieter

3

You can't just look at the output of the pair_write command
to compare the two cases. The short-range pair part
of PPPM is Coul times an error function so it is damped.

The correct thing to do is compare Coul cutoff with
eng coul + coul long when running PPPM. The cutoff
value should converge to the PPPM answer for longer
cutoffs.

Steve