Has anyone had experience using SIESTA to calculate degenerate
frequencies?
Even if I force exact symmetry in the unit cell, I get very non-degenerate
frequencies when they should be identical. Could I be doing something
wrong
or is this a numerical limitation with SIESTA?
Thanks,
Ben
Dear Ben:
this is pure numerics (as there are no symmetry constraints imposed
in Siesta), but the fact that you have VERY non-degenerate frequencies
which should be equal means that something is not so good with
your numerics.
Reasons for numerical noise:
- mesh (due to the construction of primitive cell,
X is not exactly equivalent to Y, or similar);
- k-mesh;
- unknown/undetected bugs :-)
When doing phonon calculations, you should check the convergence
of MeshCutoff (eggbox effect), and once this done, increase
the k-mesh until the forces do not noticeably depend on it.
Best regards
Andrei Postnikov
