[Pw_forum] Imaginary phonon frequency

[Paolo Giannozzi]

Your description is too vague. If you correctly transform your dynamical
matrix to real space and use them to calculate phonons at the original
wavevectors, you should get exactly the same numbers. Do you?

P,

> I used nq1 = 4, nq2 = 4, nq3 = 1 to do a phonon calculation. The
> resulting inequivalent number of q points is 6 [...]. After a phonon
> calculation, all frequencies at these 6 points are real and reasonable
> [...] However, when I used these data to get a full phonon dispersion
> in the first BZ. I found that some q points have imaginary
> frequencies.

> Can someone tell me the possible reasons?

> 
-- 
Paolo Giannozzi, IOM-Democritos and University of Udine, Italy