[lammps-users] Help with using lithium aluminate potentials

LAMMPS LAMMPS Mailing List Mirror
1

Hello,
I am writing to you to request for help with using the LiAlO2 potential.
I am working on lithium aluminate and wanted to know how to use the potentials in Lammps as in which pair style should I opt for to use the potential described below.

Should I use a hybrid of coulomb and table styles?

U(r) = (q1.q2)/r + exp (-r) x Σ (a_kn x 1/r^n)

I would appreciate your help on how I could enter these parameters into the appropriate pairstyle in lammps. I hope I was able to explain my question clearly.

2

You have multiple options:

  • write a custom pair style in C++. this may be the most effort, but also will result in the fastest running simulations, especially if you can include a version of OPENMP, GPU, INTEL, or KOKKOS
  • use the “python” pair style for testing. This is significantly slower, but allows you to work with analytical expressions and it should be sufficient enough to test if the potential parameters you have work well enough at all. from there you can then choose to either implement the previous or the next option.
  • use the pair_write command to write out a table from the python pair style. and then use pair style table. This already has accelerator support, too. The downside is that creating a well working table can be tricky and require some testing for optimal settings that produce accurate and sufficiently fast results.
  • create the table directly. this is the most difficult to debug, since you have no analytical version to compare to. and thus you are “running blind”.
3

This looks like a Buckingham potential:

https://docs.lammps.org/latest/pair_buck.html

The core and shell part is confusing. That could refer to a core/shell model for a polarizable potential: https://docs.lammps.org/latest/Howto_coreshell.html
If the potential was parameterized for that, you cannot simply ignore it.

At this point, please note that you are no longer really asking for LAMMPS advice but asking the kind of question you should ask your adviser or some suitably experienced colleagues for assistance in how to set up your calculations. I am happy to point you in the right direction, but the rest is up to you.

4

Thank you Dr. Kohlmeyer. I am giving it a try. I am also working on another similar model for another LiAlO4 potential. The two body potential looks like

phi = a_ij X exp (-r_ij / rho_ij) - C_ij / (r_ij)^6)

It has a core and a shell part. I only want to compile the core into the potential.
Would you recommend a similar table method for this sort of a potential as you recommended in the last email?

Sincerely,
Ankit Roy
Graduate Research Assistant,
Packard Lab 375,
Mechanical Engineering and Mechanics,
Lehigh University
Contact: +1(484)-935-9320