Hiya,
My name is Alexander Petkov, I am attempting to run thermal simulations on gallium oxide using LAMMPS, with zero success.
I managed to troubleshoot some of the issues I had, but now, while I’m fairly certain I’m simulating the correct structure, the run gets killed almost instantly, with an error message of lost atoms.
I have no idea why this is happening or how to fix it.
Here’s the beginning of my script (it doesn’t go on to run further than the first run command)
a) you obviously have a bad initial structure with overlapping atoms as indicated by the very high initial energy. that will “explode” atoms through your system and when atoms move too fast, LAMMPS will “lose” them.
b) having only a coulomb pair style is not going to work. atoms will simply collapse. you need to look up what is a suitable pair style and what are suitable pair_coeff parameters for that pair style to properly model your system.
axel.
Hiya,
As I said, I understand there’s an issue, but I can’t see the source, as much as I may look at my input.
In reality, there shouldn’t be an issue with the structure. At least nothing that I’ve noticed in the .dat file output at the end.
the problem is the pair style. since you have a classical model with point particles there is no repulsion between positively charged and negatively charged atoms and thus the minimization will lead to those atoms being nearly on top of each other and thus leading to gigantic forces and energies which in turn will lead to the “explosion” and lost atoms.
just add a “thermo 1” line before the minimization.
on my machine that input leads to exact overlap of oppositely charged atoms and thus the simulation will get stuck with infinite forces after about 30 iterations.
Obviously, overlapping atoms would cause a severe problem - but I cannot see that being the case here.
The lattice constants are defined in angstrom units, and the bases in fractional coordinates, right? If that’s the case, there really shouldn’t be anything wrong with the structure, unless there’s something else I’m missing.Would using only coulombic potentials be enough to break the simulation? I would ideally like to keep that as part of the interaction.
What would be the best way to overlap, or use several separate types of potentials, acting simultaneously on same groups of atoms?
you are missing the point I was trying to make completely. you first need to find out what is a correct (classical) model for your system. once you know that, you know what kind of pair style you need and then you can look up how to specify those correctly in the input. if you would add charges to a model that has their effect already included, you make a mistake and will get bogus results, same as you get them now (only with less drastic results, so even worse).
I suggest you spend some time studying textbook content on classical force fields and how they are parameterized and applied before continuing.
even the best simulation input and geometry will lead to garbage results, if you don’t model your system of interest correctly.
axel.
also, you are wasting a huge amount of CPU resources by running a simulation of a rather large system with over 350000 atoms, where a few hundred would be sufficient for initial testing and equilibration.
axel.
Hiya,
As I said, I understand there’s an issue, but I can’t see the source, as much as I may look at my input.
In reality, there shouldn’t be an issue with the structure. At least nothing that I’ve noticed in the .dat file output at the end.
Obviously, overlapping atoms would cause a severe problem - but I cannot see that being the case here.
The lattice constants are defined in angstrom units, and the bases in fractional coordinates, right? If that’s the case, there really shouldn’t be anything wrong with the structure, unless there’s something else I’m missing.
Would using only coulombic potentials be enough to break the simulation? I would ideally like to keep that as part of the interaction.
What would be the best way to overlap, or use several separate types of potentials, acting simultaneously on same groups of atoms?
Thanks in advance for your advice,
Alexander Petkov