I have done phonon calculation in Titanium in both Alpha (non-ideal HCP) and Omega (hexagonal with 3 atoms per unit cell) phases after relaxation. When I done the phonon calculation with 2 q-points in each direction (Monkhorst-Pack) I have got a reasonable phonon spectrum with q2r.x and PlotPhon, but when I have done a phonon calculation with 3 q-points or 4 q-points in each direction I have got very large negative frequencies. I am working with degauss=0.01 Ry or degauss=0.005 Ry because this is the smearing in which the energy converge to 10^(-5) Ry. The energy is converge to 10^(-6) Ry in respect to all the numerical parameters (ecutwfc,ecutrho,k-points). When I increase the smearing to degauss=0.2 Ry I have got reasonable phonon spectrum with wiggling. From calculations I have done in zero Kelvin (only SCF without phonons) I found the phase transition in zero Kelvin in reasonable pressure only with degauss<0.01 Ry. To sum up, the cold curve looks reasonable with degauss<0.01 Ry but the phonon spectrum looks reasonable only with degauss approximately 0.2 Ry. What is your recommendation?
In addition I would like to ask if the q2r.x and/or PlotPhon apply any ASR and what is the difference from matdyn.x in the sence that plotphon does not use *.freq but only *.fc file? I can put homogenous grid in the input file of matdyn.x. Is it causes any problems with hexagonal lattice? -------------- next part -------------- An HTML attachment was scrubbed... URL: http://pwscf.org/pipermail/pw_forum/attachments/20140105/e33e8e0b/attachment.html