[Pw_forum] Fortran runtime error during scf in 4.0cvs3

[Axel Kohlmeyer]

On Wed, 16 Apr 2008, Yaser Rehem wrote:

hi yaser,

YR> I got the following runtime error will attempting a scf on my system:
YR> 
YR> Cut & Paste from my Shell below:
YR> >At line 1288 of file cegterg.f90
YR> >Traceback: not available, compile with -ftrace=frame or -ftrace=full
YR> >At line 1288 of file cegterg.f90
YR> >Traceback: not available, compile with -ftrace=frame or -ftrace=full
YR> >Fortran runtime error: Array section out of bounds
YR> >Fortran runtime error: Array section out of bounds
YR> >*

what compiler? what version?
when _exactly_ does this happen?

this looks a lot like a miscompiled file. 
that array is dynamically allocated after all.

i'm currently running your input on my desktop and it has 
already completed the two iterations...

does it change when you use 6 pools instead (should be
much more efficient unless you run out of memory).

cheers,
   axel.

YR> 
YR> I'm running pwscf 4cvs3  on OS X 10.5.2 (9C7010), openmpi 1.2.4.
YR> 
YR> I was running the input file below on 6 procs, 1 pool, when I got the above
YR> error msg in my shell, and 2 of the 6 pw.x instances quit.
YR> 
YR> P.s. I realize I've got a ridiculously large number of k-points---I'm in in
YR> the process of converging the stress tensor with respect to ecutwfc and
YR> ecutrho, and wanted to ensure k-points were absolutely not an issue.  My
YR> plan was to converge k-points next.
YR> 
YR> Input File:
YR> 
YR> &control
YR>    calculation = 'scf'
YR>    restart_mode='from_scratch'
YR>    prefix='ConvTest'
YR>    outdir = './ConvergenceTests'
YR>    pseudo_dir = '../pseudo'
YR>    tstress = .true.
YR>    tprnfor = .true.
YR> /
YR> &system
YR>    ibrav=  0
YR>    celldm(1) = 8.207702028
YR>    nat=  14
YR>    ntyp= 2
YR>    ecutwfc = 38
YR>    ecutrho = 380
YR>    occupations = 'smearing'
YR>    degauss = 0.03
YR>    smearing = 'cold'
YR> /
YR> &electrons
YR>    mixing_beta = 0.7
YR>    conv_thr =  1.0d-8
YR> /
YR> CELL_PARAMETERS {alat}
YR>   1.010363552   0.000000012   0.000000000
YR>   0.000000021   1.750048386  -0.000024105
YR>   0.000000000  -0.000012933   0.925
YR> ATOMIC_SPECIES
YR>  C  12.0107  C.pz-rrkjus.UPF.txt
YR>  Li 6.941	  Li_VDB_LDA_SEMI.UPF.txt
YR> ATOMIC_POSITIONS {angstrom}
YR> Li       0.000000030   2.506788198  -0.000140635
YR> Li       2.171198117   6.271103089  -0.000356521
YR> C        1.467363622   2.526363526   2.006309099
YR> C        0.733070280   1.265423702   2.006682218
YR> C        0.738585262   3.794292056   2.006616072
YR> C        1.464475654   0.009242556   2.006460051
YR> C        2.923841993   0.009242573   2.006460051
YR> C        3.649732475   3.794292091   2.006616072
YR> C        3.655247398   1.265423737   2.006682218
YR> C        2.920954087   2.526363543   2.006309099
YR> C        3.657326610   6.344930347   2.006451947
YR> C        0.730991187   6.344930313   2.006451947
YR> C        1.464091389   5.063798994   2.006936383
YR> C        2.924226378   5.063799011   2.006936383
YR> K_POINTS {automatic}
YR>  14 8 16 0 0 0
YR> 
YR> -Yaser Rehem
YR> Rehem Research & Consulting
YR> 
YR> 

-- 
=======================================================================
Axel Kohlmeyer   akohlmey at cmm.chem.upenn.edu   http://www.cmm.upenn.edu
   Center for Molecular Modeling   --   University of Pennsylvania
Department of Chemistry, 231 S.34th Street, Philadelphia, PA 19104-6323
tel: 1-215-898-1582,  fax: 1-215-573-6233,  office-tel: 1-215-898-5425
=======================================================================
If you make something idiot-proof, the universe creates a better idiot.