The ABINIT Users Mailing List ( CLOSED )

["Anglade Pierre-Matthieu" <anglade@gmail.com>]

Hi,

To both :
Please after compiling with gfortran, run the abinit self-tests (cd
tests; make tests_acc;) then could you please pack all report files
(tar jcf report.tar.bz2 `find . -name '*report*'` )and share them on
the dedicated file sharing area of the mailing list (
https://listes-2.sipr.ucl.ac.be/abinit.org/d_read/forum/ ).  This is
almost the only way we can help you. By the way, could you also
include your config.log ?

It is very strange to find problems with standard off-the-shelf
hardware like Sara's. However I would not be surprised at all of some
bugs in a dual CPU machine. I had one on wich openBSD, DSL, and suse
would run fine but no other distro would start. After checking with
the motherboard vendor it happens that although the memory was
certified for this mobo it was not 100% compatible (for whatever
reasons I can not undersand) and was triggering the problem (memtest86
runs perfectly however). All in all this makes me think that some non
standard hardware may contain bugs that Abinit could unearthed...

To sara :
On my computer intel core2 duo on intel G35 everything run fine for
your run Sara. At least until broyden step 13 here Abinit find that
you have set a too low value for dilatmx. Once we have discovered what
is going wrong with you case, you may restart from the following. It's
0.08Ha below the original positions.
  acell=  2.0719239500E+01  2.0719239500E+01  9.7833147784E+00
  rprim=  1.0000000000E+00  1.6858543111E-33 -2.2851135397E-35
         -5.0000000000E-01  8.6602540378E-01  2.2851135397E-35
          6.1232339957E-17  1.0605752387E-16  1.0000000000E+00
  ucvol=  3.6371756927E+03 Bohr^3
  lengths=  2.0719239500E+01  2.0719239500E+01  9.7833147784E+00 Bohr
  angles (23,13,12)=  90.00000000  90.00000000 120.00000000 degrees
 Cartesian coordinates (bohr)
  3.65666858009936E+00  6.33353576717282E+00 -1.01507514530747E-01
  2.13357016937630E+00  8.97166911773326E+00  4.79014987467790E+00
  3.65671146417107E+00  1.16098272274240E+01 -1.01507514530747E-01
  6.70295116968889E+00  1.16098519865544E+01  4.79014987467790E+00
  8.22604958041193E+00  8.97171863599395E+00 -1.01507514530747E-01
  6.70290828561718E+00  6.33356052630317E+00  4.79014987467790E+00
  1.65844224799601E+00  8.97166280138445E+00  1.32442186183298E+00
  3.41915297359945E+00  1.20213032355611E+01  6.21607925104164E+00
  6.94052060049757E+00  1.20213343110402E+01  1.32442186183298E+00
  8.70117750179221E+00  8.97172495234277E+00  6.21607925104164E+00
  6.94046677618878E+00  5.92208451816614E+00  1.32442186183298E+00
  3.41909914929069E+00  5.92205344268699E+00  6.21607925104164E+00

regards

PMA

On Sat, Oct 4, 2008 at 5:10 AM, sara yazdani <sarayazdani@yahoo.com> wrote:
> Dear Sir,
>
> Thanks for running my input file.
>
> I have checked with memtest86, but it does not show anything wrong with the
> memory. I am using INTEL(R) CORE TM 2 DUO Processor.
>
> Since you told me the problem may arise from my computer I have written
> details for the manufacturer to see if they can find any conflict !!! If I
> found anything solvable, I will write for all.
>
> Thanks again.
>
> Regards,
> Sara
>
> --- On Fri, 3/10/08, jimprtlnd01@gmail.com <jimprtlnd01@gmail.com> wrote:
>
> From: jimprtlnd01@gmail.com <jimprtlnd01@gmail.com>
> Subject: Re: Re: [abinit-forum] Haulting of the run
> To: anglade@gmail.com, forum@abinit.org
> Date: Friday, 3 October, 2008, 2:38 PM
>
> THIS IS ALMOST EXACTLY the behavior I'm seeing.  Please continue to look
> into
> this.  I get constant hangs (where it stops updating the log files) and
> SEGMENTATION FAULTS galore.
>
> I am running on a 8-core Mac Pro.  I have tried gfortran, and intel
> compilers.
> I have tried 5.4.4, 5.5.4, and 5.6.2.  I have tried multiple input files.
>
> All exhibit similar behavior.  Hang or segmentation fault.  Usually if I let
> the hang run for a couple of hours, it eventually crashes with a seg fault.
>
> I can sometimes get it through Broyden Step 0, but only by changing my input
> file, lowering ecut to 20 or something like that.
>
> Please please let me know what I can do to help track this down.  I'm
> running
> very similar molecules, like Fe3O4, Cu2O.
>
> I will include an input file for completeness.        I can provide any more
> information, and will be HAPPY to work with someone to track this down and
> get
> it fixed.  I am an experienced software developer, so I think I can be a lot
> of
> help with running tests and tracking this bug.
>
> It's preventing me from using ABINIT to accomplish anything.
>
> One more thing.  In this particular instance, a dilatmx of 1.2 caused it not
> to
> crash.        But, instead it exited with an error saying it needed at least
> dilatms
> of 1.58.  So you see my dilemma...
>
> ******* Input file
>
>
>   kptopt   1          # Automatic generation of k points, taking
>                            # into account the symmetry
>   tolvrs   1.0d-12    # SCF stopping criterion
>   iscf         5            # Self-consistent calculation, using algorithm 5
>
>
> #######################################################################
> #Common input variables
>
>  chkprim  0
>
> #Definition of the atom types
> acell 3*4.252 angstrom
>  ntypat   2
>  znucl        8 29
>  natom         9
>  typat 1 2 2 1 1 2 1 2 1
>  xangst             1.265   1.054   -0.917
>              2.604   0.846   0.438
>              -0.211   2.047   -0.208
>              3.944   0.638   1.793
>              -1.69   3.043   0.503
>              0.658   -0.672   -1.487
>              0.05   -2.403   -2.059
>              2.005   1.989   -2.403
>              2.752   2.932   -3.903
>
>
>
> #Gives the number of band, explicitely (do not take the default)
>   nband  72
> #  iprcel 142
>
> #Exchange-correlation functional
>
>   ixc 1             # LDA Teter Pade parametrization
>
> #Definition of the planewave basis set
>
>   nstep  60
>   ecut        70.0           # Maximal kinetic energy cut-off, in Hartree
>   optcell 2
>   ecutsm 0.5
>   ionmov 2
>   ntime 80 # number of steps of calculations IF ionmove = 1 or 2
> # toldff 5.0d-5
> # toldfe 1.0d-4
> # occopt 3   # **** may need to change for metals
>   tolmxf 5.0d-4  #**** this is to speed up geometry optimization
>   dilatmx 1.8
>   strprecon 0.1
>
>    diemac   9.0
>
>



-- 
Pierre-Matthieu Anglade