Hi,
In the lammps mailing list, I found that, when using ewald method, lammps can successfully handle non-neutral systems by considering a uniform background charge, and also by automatically applying a correction to the energy.
I am interested to know more about this correction.
Can you please tell me some direct references? Also, I am interested to see how the code applies this correction (line number of the code?)
Thanks
Hi,
In the lammps mailing list, I found that, when using ewald method, lammps can successfully handle non-neutral systems by considering a uniform background charge, and also by automatically applying a correction to the energy.
I am interested to know more about this correction.
this is not an explicit correction that is applied. it is a consequence of ignoring the (divergent) term for k=0 in the reciprocal space part of the lattice sum.
technically speaking, the energy of a lattice with a unit cell having a net charge is infinitely large. however, for a neutral unit cell, the energy contribution at k=0 is zero. so if you ignore that contribution, the physical effect is the same as if you have the unit cell embedded in a neutralizing jellium. the rest is just successful marketing.
Can you please tell me some direct references? Also, I am interested to see how the code applies this correction (line number of the code?)
just read up on lattice sums. this is knowledge that goes back almost a hundred years. there is plenty of material available. 
axel.
Thanks Axel.
As I want to cite the method in my document, I need to know exactly how LAMMPS does the process, and refer to it. Only in the page of compute group/group I could find some related text. I appreciate if you could tell me the corresponding references. Is this one of the references?
Bogusz, Stephen, Thomas E. Cheatham III, and Bernard R. Brooks. The Journal of chemical physics 108.17 (1998): 7070-7084.
Thanks
Thanks Axel.
As I want to cite the method in my document, I need to know exactly how LAMMPS does the process, and refer to it. Only in the page of compute group/group I could find some related text. I appreciate if you could tell me the corresponding references. Is this one of the references?
no. LAMMPS does what every MD code does. as i explained, there is no specific active step, but rather what people call this “compensation” happens automatically when you omit the divergent term at k=0. this should be explained (and referenced) in every reasonable MD textbook (Allen/Tildesley, Frenkel/Smit, etc.) and the original ewald paper from the early 19-hundreds, but also in papers describing the various kspace styles that LAMMPS implements. the kspace_style pages lists most of them.
it is rather sad to see, that you worry about this only now after you have done your simulations and are preparing a paper. you should have studied this before even starting a simulation, so you know what you are doing.
Bogusz, Stephen, Thomas E. Cheatham III, and Bernard R. Brooks. The Journal of chemical physics 108.17 (1998): 7070-7084.
also, it is very bad practice to reference papers without having read them. if you read those, you will see which ones you need to cite.
axel.