Am I doing something wrong here, or is there a way to force the lattice vectors to remain orthogonal?
I also would prefer the supercell to be as cubic as possible. Can I assign its dimensions beforehand, and let mcsqs populate it?
Thanks for self-answering it makes my job easier!
It’s a perfectly valid way to go if you have reasons to prefer orthogonal lattice vectors.
But of course, imposing a constraint always reduces the optimal point you can reach, so you would end up being able to match fewer correlations to the disordered state values for a given number of atoms in the cell. Most good SQS have "weird" lattice vectors precisely because it helps in simulating a disordered phase.
Thanks for the reply, and glad I could save you some time!
It’s just that orthogonal cells are much more convenient for visualization. It’s also easier to handle in Lammps for instance, or to apply uniform/uniaxial compression.
Anyway, I’d have thought that the results do not depend on the shape of the cell. Or are the PBC not taken into account during the supercell building?
I understand about the advantages of orthogonal cells.
The PBC are fully taken into account during the SQS construction. It’s just that you get better SQS when no cell constraints are imposed. A constrained optimization always yields a worse optimum than an unconstrained optimization.
it makes my job easier!