I’m trying to relax a 2D MOF structure with this published interatomic potential using the given minimal input for minimization. The initial structure is DFT optimized for cell dimensions and atomic positions.
When I switch the pair style from lj/cut/coul/cut to lj/cut/coul/long with PPPM k-space style, the H atoms attached to N atoms—which are in a square planar arrangement around Ni metal atom—tend to attract each other and overlap. This leads to structural distortions after minimization. Can anyone please help me understand this behavior and suggest alternative approaches in LAMMPS for accounting for long-range Coulomb interactions? Data, minimal input files (md_pair1.in: lj/cut/coul/cut and md_pair2.in: /lj/cut/coul/long) and the minimization dumps from both pair styles (dump1.lammpstrj: lj/cut/coul/cut and dump2.lammpstrj: /lj/cut/coul/long) are attached.
Thank you.
Input:
units metal
special_bonds amber
atom_style full
boundary p p p
dielectric 1.0
kspace_style pppm 1.e-4 # comment out this part for lj/cut/coul/cut 14.0 16.0
pair_style lj/cut/coul/long 14.0 16.0 # change pair style to lj/cut/coul/cut 14.0 16.0
bond_style hybrid harmonic morse
angle_style harmonic
dihedral_style charmm
improper_style cvff
pair_modify mix geometric # default
read_data ${path}/DATA
include ${path}/potential.coeff
thermo 100
thermo_style custom step time temp press etotal pe ke evdwl ecoul ebond eangle edihed eimp
minimize 1e-10 1e-10 10000 100000000
You forgot to attach the potential.coeff file.
You also didn’t mention which platform you are running on.
Best capture the output of lmp -h and attach it as well.
Does the same issue happen with the 29 August 2024 stable release?
Unfortunately, I don’t have permission to share this file, which was originally provided to me by the authors of the linked article (I should have better conveyed this information in my original post). I understand if it’s not possible to offer any further help without this file.
You also didn’t mention which platform you are running on.
I am running LAMMPS on Red Hat Enterprise Linux 8.10 with an x86_64 architecture, using GNU C++ 8.5.0 with OpenMP 4.5 and MVAPICH2 MPI 2.3.7 on a cluster with SLURM
Best capture the output of lmp -h and attach it as well.
Sorry, but that doesn’t make sense. The article is published, but the authors are not sharing the parameters? On top of that they are all at Sandia, which is a big supporter of LAMMPS development. So they are using the software that we are sharing openly and do not want to reciprocate? That is hard to believe.
To clarify, the authors didn’t explicitly state that we shouldn’t share the parameters (they were generously provided to us), but I need to ask their permission to do so since the interatomic potential is not included with their published article. This process might take some time, and I didn’t want to leave your questions unanswered in the meantime. I apologize for not foreseeing that the potential file might be an important factor when I posted…
