I wan to model CSL grain boundaries，but there is atomic overlap at the grain boundary. Through literature review，Several successive rigid body translations, followed by an atom-deletion technique can remove the overlap atoms. So I want to know how to realize the successive rigid body translations?
These kinds of transformations can be done by defining a group and then apply the displace_atoms command. However, your description sounds more like the system preparation was done with an external tool that specializes in such processing like Atomsk.
The LAMMPS way of preparing such systems would be to define the simulation cell and then regions for the different grains and then filling those regions with atoms using the create_atoms command after defining a suitable lattice and using the “orient” keyword with corresponding Miller indices to define the orientation of the lattice for that region. Potential overlaps can then be removed with the “overlap” keyword of the delete_atoms command.
Yes, I established the grain boundary model by atomsk software. There are atoms overlapping at the grain boundary. By deleting the atoms and then minimizing the energy, it is found that the arrangement of grain boundary atoms is irregular.
It is extremely frustrating for people responding in forums when they are confronted with incomplete and vague information. Please keep in mind that people here are volunteering their time and thus you should make the best effort to get specific responses and not information you already know.
Atoms go where the physics tells them to go. So if you don’t get the results you expect, you have to check on your input. It is rarely the case that this can be resolved with a general answer. For the kind of system you describe, it can easily happen that symmetry and regularity depend on exact cancellation of forces and that can be difficult to achieve when you are using floating point math and input data where positions are stored with finite precision. Thus an otherwise theoretically metastable geometry can easily change into an irregular, unsymmetric one. This is usually not an issue of the software in use, but a question of proper planning and checking of the steps to produce the geometry.