Problem on the ewald(PPPM) algorithms

LAMMPS LAMMPS Mailing List Mirror
1

Dear Prof. Stan,

I want to simulate the process of heterogeneous nucleation (containing ions (40 na+)) of supersaturated water vapor, and PPPM is

used to compute the long-range coulombic interactions. Thus, net charge of my system is not zero, there is a warning in lammps:

system is not charge neutral, net charge=40.

When the net charge of my simulation system is not zero, the ewald(PPPM) algorithms will act as if the simulation cell is surrounded

by an evenly distributed countercharge and thus neutralized. Besides, I find that a correction to the system energy is executed by

LAMMPS (In page 327 of the LAMMPS manual (compute group/group command), it says that “However, one can assume the

net charge of the system is neutralized by a uniform background plasma, and a correction to the system energy can be applied to

reduce artifacts. This correction term does not affect the force calculation and will be zero if one or both of the groups are charge

neutral. This energy correction term is the same as that included in the regular Ewald and PPPM routines.”

I have used the compute group/group command to check whether the correction term is added. I found that the energy is different,

but the force is the same, if the boundary keyword is set to yes and no.

So I want to know whether this energy correction is included into the PPPM method during the the calculation of particles’s

movement or it just work when the compute group/group command is used to calculate the energy (whether the

long-rang contribution to the force acting on the particle is ill-defined?). What is the evidence that I can use to convince

others?

Best regards

Zhang Chao

2

Dear Prof. Stan,

I want to simulate the process of heterogeneous nucleation (containing ions
(40 na+)) of supersaturated water vapor, and PPPM is

used to compute the long-range coulombic interactions. Thus, net charge of
my system is not zero, there is a warning in lammps:

system is not charge neutral, net charge=40.

let me reiterate. no serious reviewer will accept a simulation of such
a system as a serious study.
the implicit neutralization will at best be considered acceptable for
simulations with a net charge <= +/- 1.

in your specific situation of studying a rare event, i suggest you
spend some time to explore the *vast* body of published work
concerning simulation of rare events and the associated free energy
differences. LAMMPS supports several of the associated methods. a
brute force and unphysical model like yours is never a good choice and
no statement like what you have been repeatedly fishing for, will help
you to justify it. semantics and specific wording of the documentation
are no replacement for understanding and accepting the reality of the
physics. what does it help, if somebody here say, :"yes, this is the
proper correction energy". reviewers don't care unless you can provide
*conclusive* proof. and that requires that *you* can justify it and
not just quote some random statement from some random person on a
mailing list.

axel.

3

There is a correction to the energy applied to the Ewald and PPPM methods when the net charge is not zero, similar to compute group/group. For more info about this correction see J. Chem. Phys. 108, 7070 (1998); http://dx.doi.org/10.1063/1.476320 Equation 3 and corresponding references. But as Axel has said, you are on shaky ground, and your results could be totally wrong if you have a net charge of +40. It is up to you to prove if your simulations are meaningful or not.

Stan

4

Dear Prof. Axel Kohlmeyer,

Thanks very much for your help. Now, I describe my problem: I want to simulate the process of heterogeneous nucleation (containing one ion) of supersaturated water vapor, and I use NVT ensemble. I still have two problems on the answer you give me:

  1. LAMMPS manual says that the correction is only for energy, not force and yes it is applied to Ewald and PPPM. Strangely, the force of each particle is calculated by the energy, why the correction is not useful for force, whether it is because that the correction is a constant, and the derivative of the correction is zero?
  2. If I use LAMMPS to simulate my problem (using PPPM), I was asked by a question like that: if the system carries a net charge, the simulation is conducted at infinite energy and infinite pressure. Furthermore, the long-range contribution to the forces acting on the particles is ill-defined. Is the calculation of force is right? What should I do to prove the reliability of the force calculated by PPPM with LAMMPS? Firstly, I will talk about the idea of mine: I can use the “pair_style lj/cut/coul/cut” and “pair_style lj/cut/coul/long” (PPPM) to calculate the Coulomb force, respectively. Then, I calculate some results which I need in this simulation with different cutoff for “pair_style lj/cut/coul/cut”. If the result calculated by the PPPM approximately equal to the result calculated by the “pair_style lj/cut/coul/cut” method (with a very long cutoff), I can prove the of the force calculated by PPPM in LAMMPS. Is my ideal reasonable? I hope you can give me some advices.

Thanks very much for your attention!

Best regards

Zhang Chao

Axel Kohlmeyer 写:

5

Repeating the same questions over and over again with minimal changes won’t help you to get around having to do the “dirty work” yourself. Read the papers, read the source code, make tests, compare. None of us will do it for you.

Axel