On Tue, 2013-01-15 at 18:10 -0500, David Strubbe wrote: > Dear QE developers, > > > I have encountered an internal error regarding symmetry with QE 5.0.2, > for a calculation with 8 atoms of Si. The scf calculation with pw.x > runs fine, but then ph.x crashes immediately with the following > message. This happened in serial and parallel with ifort, as well as > in serial with gfortran. Input files scf.in and ph.in are below also. > I tried with the patch to 5.0.3 emailed out by Dr. Gianozzi a few > weeks, and got the same error. > > > David Strubbe > MIT > > > %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% > %%%%%%%%% > Error in routine divide_class (1): > Wrong classes for C_3v > %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% > %%%%%%%%% > > > $ cat scf.in > &control > prefix = 'silicon' > calculation = 'scf' > restart_mode = 'from_scratch' > pseudo_dir = '/home/dstrubbe/raman_Si/pseudo' > / > &system > ibrav = 0 > celldm(1) = 1.88972613 > nat = 8 > ntyp = 1 > nbnd = 32 > ecutwfc = 50.0 > / > &electrons > / > CELL_PARAMETERS > 5.427632 0.0 0.0 > 0.0 5.42763 0.0 > 0.0 0.0 5.42763
This is a well known problem. pw.x finds more symmetries than the routines that analyze the point group. This happens because the difference between 5.427632 and 5.42763 is too small and pw.x finds symmetries that are not there. If this is a mistake in your input than please correct it an use three equal values, if instead this is really what you want, than the threshold for comparing symmetries (accept parameter in symm_base module) should be decreased, otherwise pw.x ignore the difference in any case. HTH, Andrea > ATOMIC_SPECIES > Si 28.086 Si.UPF > ATOMIC_POSITIONS angstrom > Si 0.542763 0.542763 0.542763 > Si 3.256579 3.256579 0.542763 > Si 0.542763 3.256579 3.256579 > Si 3.256579 0.542763 3.256579 > Si 1.899671 1.899671 1.899671 > Si 4.613487 4.613487 1.899671 > Si 1.899671 4.613487 4.613487 > Si 4.613487 1.899671 4.613487 > K_POINTS automatic > 1 1 1 0 0 0 > > > $ cat ph.in > phonons of Si at Gamma > &inputph > tr2_ph=1.0d-12, > prefix='silicon', > epsil=.true., > trans=.true., > amass(1)=28.0855, > / > 0.0 0.0 0.0 > > > _______________________________________________ > Pw_forum mailing list > Pw_forum at pwscf.org > http://pwscf.org/mailman/listinfo/pw_forum
