Re: [lammps-users] Tabulated potential + coulombic long range potential
Meral Sharkas <eng_meral.sh10@...16...>
Fri, 11 Aug 2017 10:10:25 +0000 (UTC)
Thank you very much Axel for your reply and guidance.
May i ask plz what does a 'bad geometry' indicates.does that mean like overlappimg of atoms? Also if I am having the unit cell position from the internet not developed by myself is there a chance to have a bad geometry?
I understand that this is out of the scope of the mailing list but I just want to know if it is a problem in how I am using lammps or another problem..
I am simulating radiation damage in zircon and according to the literature the potential is ZBL for ranges < 1 A , born mayer huggings <10 A and colombic long range.
I tabulated the short ranges potential an joined the two potentials using a fermi function. When i attempted to use a pair_style hybrid/overlay to overlay the colombic long range potential with the tabulated potential, i got an error of "Lost atoms".
N.b: When creating the tabulated potential, i made all the cutoff distances= 10 A(for zbl and BMH potential)
As a troubleshoot to my problem i tried to simulate the system with just pair_style coul/long, i got NAN results, and when i used pair_style coul/cut i got lost atoms.
only using coulomb interactions for point particles cannot result in stable dynamics, since coulomb has no short range repulsion. atoms of opposite charge will just end up on top of each other (also known as "classical coulomb catastrophe"). so NaN (i.e. divergence) or lost atoms (due to extremely strong forces when atoms get close) are to be expected.
when you get lost atoms for your tabulated potential, chances are, that your tabulation is faulty or your geometry not good. this is something that you have to resolve yourself. that is just part of the job of doing simulations. nobody here will do this work for you.
please note, that building tabulated potentials has become much simpler with LAMMPS with the addition of pair style python. you can simply define a piece-wise defined potential function (and check or derive its force through numerical differentiation) in python, plot and compare it with a standalone code and then put it into LAMMPS, create the tables using the pair_write command. there are several examples provided in the examples/python folder.
i am wondering what could be the reason of these problems?
Please find a part of my input and if you can help me that would be really appreciated.