In my experience, increasing also the convergence for forces and total
energy of the scf cycle you perform before the neb.x run, helps a lot to
clean negative frequencies.
Then, all other cutoffs and settings that increase amount of memory and
cpu time, can helps, sometimes significantly, sometimes their not enough.
Before the phonon, a proper lattice relaxation with higher parameters
than default value for forces, total energy and eventually even electron
convergence (as also wavefunction, charge density, etc..), helps much
more to bring your model to the highest converged condition than just
increasing parameters thresholds on phonon that can make the phonon a
jurassic waiting.
These are the reflections helped me more
Francesco
Strathclyde
On 25/08/16 15:03, Elio Physics wrote:
Dear all,
I have calculated the frequencies for a structure. All frequencies
were positive except for a few "extrapolated" points very close to the
Gamma point which were negative in the range of -1.45 cm^-1. Due to
this the dispersion curve near the Gamma point looked a bit funny (but
looks pretty good elsewhere). I tried the calculations with 2
different phonon grids. Any reason why this is happening? Do i have to
lower tr2_ph more (it is already 10^(-19)!)
Any help would be appreciated.
Elio
Fed Uni of Rondonia
Brazil
_______________________________________________
Pw_forum mailing list
Pw_forum@pwscf.org
http://pwscf.org/mailman/listinfo/pw_forum
_______________________________________________
Pw_forum mailing list
Pw_forum@pwscf.org
http://pwscf.org/mailman/listinfo/pw_forum