Hi Katalin, My apologies for very stupid question: did you change nq's in q2r input file?
I still prefer 2.1.5 (rather old) version and there is not too much difference in q2r.x between 2.15 and 3.1.1 versions. So, do not worry about your q2r (and PWSCF) version. I use this procedure last 4 years and never met any problem for any lattice structure I considered (bcc, fcc, hcp, bct, orthorhombic, sc), except ibrav=0 though this case is also no problem. And once for hcp lattice I used quite unusual choice of basis vectors, and forgot about it that led to this erros. Bests, Eyvaz. --- Katalin Gaal-Nagy <katalin.gaal-nagy at physik.uni-regensburg.de> wrote: > Dear Eyvaz, > > thanks for the reply. > > I know that it was explained before, and I generated > the q points exactly > in that way: Running pw.x for some minutes (the q > points have the correct > units, etc, thanks Cesar for the hint!), take the q > points and put them > to the phonon calculation. It worked well with the > 4x4x4 mesh. > For the 4x4x8 mesh I did the same, compared the two > grids and calculated > just the q points which have been missing. And there > the error appeared > for some (not all) of the new q points. > > For curiosity, I just removed the not-allowed q > points from the files, but > then there are some q points missing. At the moment > I guess that due to > some symmetry operation some signs of the q points > are changed, or, > translated to other IBZs, whatever, and I want to > look through the q2r > routine and print out the "reference" q points and > compare them with the q > points I have in the files with the dynamical > matrices. Then I will see > what is going on. (I still use the "old" version of > the q2r). > > Unfortunatly, I am not at the office for one week, > but I will give a short > report as soon as I have some news. > > All the best, > Katalin > > > > Hi, > > > > Sorry interfering in, but it was explained before > why > > this happens. Let me remind again. > > > > In my opinion, there is two solutions of the > problem: > > > > 1) use >v3.0 where you can directly put nq1, nq2, > nq3 > > and everything will go smoothly: > > (something like this:) > > &inputph > > tr2_ph=1.0d-14, > > prefix='C8', > > ldisp=.true., > > nq1=4, nq2=4, nq3=4 > > amass(1)=12.01, > > outdir='$TMP_DIR/', > > fildyn='C8.dyn', > > / > > > > In this case you have to supply enough CPU time to > > complete dynamical matrix calculations. Otherwise > in > > some clusters you will lose your temporary files > and I > > am not sure that an option "recover" will help in > this > > case. I would be happy if this one is not true. In > > fact, I faced this problem. > > > > 2) If you prefer say 2.1 version, again, it was > > explained before, and the best way is to generate > > q-points via pw.x (kill the job in few seconds and > > look at *.scf.out file) and put them into your > input > > file. > > In this case the \Gamma point is calculated > separately > > and, thus, remove the \Gamma point from the list. > > > > Hope they are helpful. > > > > Bests, > > Eyvaz. > > > > --- cesards at msi.umn.edu wrote: > > > >> A common cause for this problem is defining > q-points > >> in reciprocal cell > >> coordinates. q-points must be defined in 2pi/a > >> ccoordinates. > >> > >> Yours, > >> > >> Cesar R.S. da Silva > >> > >> > _______________________________________________ > Pw_forum mailing list > Pw_forum at pwscf.org > http://www.democritos.it/mailman/listinfo/pw_forum > __________________________________________________ Do You Yahoo!? Tired of spam? Yahoo! Mail has the best spam protection around http://mail.yahoo.com
