Precision of phonon eigenvectors

Hi Dr. Gale,

Apologies if this has been asked before, but I couldn’t find it in the manual. Could you tell me the error in each component of a phonon eigenvector calculated at the Gamma point? When I use the “eigenvector” keyword to write them, I see in the *.eig file they are reported to 6 decimal places each - are they really accurate to six digits?

All the best,


Hi Connor,
I don’t think this is in the manual. The eigenvectors when computed in double precision should be precise to 6 d.p., but it’s highly unlikely that they are accurate to 6 d.p. since this depends on the quality of the force field model. There is also the caveat that if you have degenerate modes then the eigenvectors can mix and so are ambiguous to any number of decimal places.