Hi!
Following up on this question (Pair Coefficients), which is the possibility of changing the pair_style/pair_coeff arguments based on atom positions. Is there a leeway around this without the need to create my own pair_style?
Essentially, I am using a Yuakawa potential in which I want to define a different screening length for different regions of the simulations box (defined by a single coordinate—say z [edge to 0] and [0 to edge]). This is intended to simulate regions with different dielectric constants where I am not simulating the solvent atoms explicitly.
I came across the the use of fix adapt, however, it seems that it does not apply for the Yukawa potential. It says in the manual that I may add a pair_style to this fix “easily”. So, it is one option to pursue.
Another idea is to define a dynamic atom type (akin to a dynamic group) by changing the internal LAMMPS variables through accessing the LAMMPS class through a Python code or so, and by that, I can define multiple pair_coeff lines with the desired screening length for each type. That is, I would have two atom types where each type encompasses the atoms within a specific region that I define, and those atom types are updated through a Python wrapper. I am skeptic on what potential issues the change of atom types amid the simulation could cause, and whether this is advisable at all.
What are some thoughts about this, and whether there are any suggestions? The question referenced above is from 2018, and perhaps there is a LAMMPS command that I am not aware of that I can exploit.
Thanks!