[Pw_forum] pwscf 'Relax' problem

[Paolo Giannozzi]

Thank you for reporting this. Many problems with QE in 
parallel execution derive from the mixup of different MPI 
libraries and runtime environments. Sometimes I wonder if
some machines have more MPI libraries and launchers installed
than processors.

Paolo

On Wed, 2015-04-01 at 16:31 +0200, Konrad Gruszka wrote:
> Dear Community,
> 
> I've finally managed to solve this problem, so in general I write this 
> if someone will be dealing with same/similar case.
> 
> The whole problem was in launch command. On the machine on which I work 
> are installed two communication environments mpi and openmpi. So the run 
> command (mpiexec) was wrong, and in my case I have to use 'mpirun.mpich'.
> 
> After that, any of mentioned communicates disappeared.
> 
> Regards,
> Konrad Gruszka
> 
> Assistant prof.,
> Czestochowa University of Technology, Poland
> 
> W dniu 30.03.2015 o 22:50, Konrad Gruszka pisze:
> > Dear Community,
> >
> > I'm new in QE. I'm using the 5.1.1 version on linux platform.
> > While trying to do the relax computation, I'm experiencing the following
> > problem:
> >
> >     Error in routine bfgs (1):
> >         step_old is NOT normalized
> >
> > (this is not from the start, it appears after some time, e.g. after 3
> > hours)
> >
> > then a little later (couple minutes), I've get about 50 lines similar
> > like those below:
> >
> >        c_bands:  2 eigenvalues not converged
> >        c_bands:  2 eigenvalues not converged
> >        c_bands:  2 eigenvalues not converged
> >        c_bands:  2 eigenvalues not converged
> >        c_bands:  2 eigenvalues not converged
> >        c_bands:  1 eigenvalues not converged
> >        c_bands:  1 eigenvalues not converged
> >        c_bands:  1 eigenvalues not converged
> >        c_bands:  1 eigenvalues not converged
> >
> > and at the same time in console:
> >
> > darnok at darkstar:~/qe/espresso-5.1.1/moje/YFe5$ mpirun -np 8 pw.x -in
> > yfe5.geopt.in -npool 1 > yfe5.geopt.out
> > application called MPI_Abort(MPI_COMM_WORLD, 1) - process 0
> > application called MPI_Abort(MPI_COMM_WORLD, 1) - process 0
> > application called MPI_Abort(MPI_COMM_WORLD, 1) - process 0
> > application called MPI_Abort(MPI_COMM_WORLD, 1) - process 0
> > application called MPI_Abort(MPI_COMM_WORLD, 1) - process 0
> > application called MPI_Abort(MPI_COMM_WORLD, 1) - process 0
> > application called MPI_Abort(MPI_COMM_WORLD, 1) - process 0
> >
> > Then only one process (out of 8) remains and continues to run. My
> > questions are:
> > 1. If this computation will terminate normally at the end, can I trust
> > the results?
> > 2. How to solve this problem?
> > 3. For relax calculations I've read that the K_points section probably
> > should be more like 1 1 1 0 0 0 (but why?)
> > 4. Any other tips to speed up the calculations of below physical problem?
> >
> > This is my input file:
> >
> >    &control
> >       prefix='YFe5',
> >       restart_mode='from_scratch'
> >       calculation='relax'
> >       pseudo_dir='/home/darnok/qe/espresso-5.1.1/pseudo/'
> >       outdir = '/home/darnok/qe/espresso-5.1.1/tempdir/'
> >    /
> >    &system
> >       ibrav=  4, celldm(1) =9.505, nat=  6, ntyp= 2, celldm(3)= 0.77,
> >       ecutwfc = 80.0, smearing = 'gaussian', occupations = "smearing",
> > degauss=0.02,
> >    /
> >    &electrons
> >    mixing_mode = 'plain'
> >    mixing_beta = 0.7
> >    /
> >
> >    &ions
> >    /
> >
> > ATOMIC_SPECIES
> >    Fe  55.845  Fe.pbe-mt_fhi.UPF
> >    Y  88.906  Y.pbe-mt_fhi.UPF
> >
> > ATOMIC_POSITIONS {crystal}
> >    Y 0.00 0.00 0.00
> >    Fe 0.33 0.66 0.00
> >    Fe 0.66 0.33 0.50
> >    Fe 0.50 0.00 0.50
> >    Fe 0.00 0.50 0.50
> >    Fe 0.50 0.50 0.50
> > K_POINTS {automatic}
> >      9 9 9 0 0 0
> >
> > Thanks in advance
> > Konrad Gruszka
> >
> >
> > _______________________________________________
> > Pw_forum mailing list
> > Pw_forum at pwscf.org
> > http://pwscf.org/mailman/listinfo/pw_forum
> 

-- 
 Paolo Giannozzi, Dept. Chemistry&Physics&Environment, 
 Univ. Udine, via delle Scienze 208, 33100 Udine, Italy
 Phone +39-0432-558216, fax +39-0432-558222