[Pw_forum] difference in output from dynmat.x and matdyn.x

[Sridhar Sadasivam]

Dear espresso users and developers,

I am trying to obtain the phonon dispersion of alpha-quartz (SiO2) which
has a 9 atom unit cell. I ran ph.x on a 3 x 3 x 3 q-point grid. I am seeing
some strange behavior where there seems to be a difference in the
frequencies output from dynmat.x and matdyn.x. I will explain my problem
below.

I diagonalized the dynamical matrix at the gamma point using dynmat.x with
acoustic sum rule set to crsytal. Since SiO2 is a polar material, LO-TO
splitting will need to be considered at the gamma point but I am ignoring
it currently (I did not specify any direction for LO-TO splitting in the
dynmat.x input). This gave me the following 27 frequencies (in cm^-1) at
the gamma point: (0,0,0, 152.24, 175.57, 175.57, 256.32, 256.32, 341.09,
354.41, 376.86, 376.86, 419.01, 419.01, 435.37, 455.70, 659.12, 659.12,
744.54, 754.62, 1034.38, 1034.38, 1040.20, 1051.12, 1139.60, 1139.60)

I also used q2r.x (with zasr = 'crystal') to obtain the force constants and
then used matdyn.x (with asr = 'crystal') to obtain the frequencies at the
gamma point. However with matdyn.x, I obtain the following frequencies at
the gamma point: (-29.4088,-22.2535,-22.2535,152.2412,175.8141,175.8141,
256.3373, 256.3373, 341.0892, 355.4227, 376.0495 , 376.0495, 418.6344,
418.6344, 435.3710, 453.6862, 659.7607, 659.7607, 746.4843, 754.8867,
754.8867, 1034.1751, 1034.1751, 1039.7590, 1051.1207, 1139.8191, 1139.8191)

Could someone please explain why I am getting imaginary frequencies from
matdyn.x while dynmat.x gives real frequencies at the same q-point (gamma
in this case)?

Thanks,
Sridhar
Purdue University
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.quantum-espresso.org/pipermail/users/attachments/20141115/43c427e9/attachment.html>