TM pseudopotential segmentation fault with 7.0.3

External optional components of Abinit (BigDFT, Wannier90...) and codes using/providing data from/to Abinit (AtomPAW, ONCVPSP, EXC, DP, Yambo...)

Moderators: ebousquet, bxu

Locked
yurtesen
Posts: 15
Joined: Sun Nov 11, 2012 9:42 pm

TM pseudopotential segmentation fault with 7.0.3

Post by yurtesen » Thu Nov 29, 2012 12:06 pm

Hello,

I am not sure if this is the right place to post this. But here it goes:

I recently built abinit 7.0.3 and it is crashing with a specific dataset. I normally build it with ATLAS, FFTW, GSL and MVAPICH2 but I also tried with OpenMPI and without any special libraries (using the defaults). I also tried running 1 MPI process and 12 MPI processes, also tried to build without mpi-io. I couldnt build it without MPI, I guess one needs MPI right? The result was the same.

The input files
diamond.pdf
Change the extension to .tgz
(80.05 KiB) Downloaded 398 times
are attached. Change the extension to .tgz ( forum does not seem to allow uploads of any useful extensions :D )

Code: Select all

Program received signal SIGSEGV: Segmentation fault - invalid memory reference.

Backtrace for this error:
#0  0x2AC3D877AEF7
#1  0x2AC3D877B4C4
#2  0x343843291F
#3  0xE7DA83 in kpgsph_ at kpgsph.F90:364
#4  0xE7B98D in kpgio_ at kpgio.F90:163
#5  0x500F70 in inwffil_ at inwffil.F90:318
#6  0x4984A4 in gstate_ at gstate.F90:589
#7  0x424A75 in gstateimg_ at gstateimg.F90:421
#8  0x41988B in driver_ at driver.F90:586
#9  0x40FD4C in abinit at abinit.F90:435


The last output I see from the program is:

-inwffil : will read wavefunctions from disk file co_DS1_WFK

This used to work with v6.12.3 fine. How can I help to get this issue fixed?

Thanks,
Evren

karu
Posts: 7
Joined: Tue Sep 25, 2012 10:50 am

Re: TM pseudopotential segmentation fault with 7.0.3

Post by karu » Thu Nov 29, 2012 1:56 pm

Hi all,

Couple of more bits of info on this (I'm not sure if this is relevant):

-Another input that uses HGH pseudos does not exhibit the same behavior.
-But that input does not try to read output wave functions from a previous calculation.
BR,
Kari

Emile
Posts: 15
Joined: Tue Dec 04, 2012 12:54 pm

Re: TM pseudopotential segmentation fault with 7.0.3

Post by Emile » Wed Dec 05, 2012 11:22 am

Dear abiniters

As Yurtesen , I am also encountering a problem with abinit 7.0.3.
I make a response function calculation according to the tutorial and it systematically failed at DATASET 3.
Al GGA PAW atomic data were used in the first time. Then I came back to the use Al LDA PAW atomic data and I made lot of tests, same results. I also tested the parameter pawxcdev without more success.


Tests were done on:

UBUNTU 12.04 LTS
.Version 6.12.3 of ABINIT
.(MPI version, prepared for a x86_64_linux_gnu4.4 computer)
---> works

UBUNTU 12.04 LTS
.Version 7.0.3 of ABINIT
.(MPI version, prepared for a x86_64_linux_gnu4.6 computer)
---> bug after DATASET 3, after reading files from DATASET 1

[reccar197070:32198] *** Process received signal ***
[reccar197070:32198] Signal: Segmentation fault (11)
[reccar197070:32198] Signal code: Address not mapped (1)
[reccar197070:32198] Failing at address: 0xb261edb8
[reccar197070:32198] [ 0] /lib/x86_64-linux-gnu/libpthread.so.0(+0xfcb0) [0x7f43166efcb0]
[reccar197070:32198] [ 1] ../../Prog/abinit-7.0.3/src/98_main/abinit(pawdensities_+0xde3) [0xa37fd3]
[reccar197070:32198] [ 2] ../../Prog/abinit-7.0.3/src/98_main/abinit(pawdenpot_+0x1834) [0xa2e644]
[reccar197070:32198] [ 3] ../../Prog/abinit-7.0.3/src/98_main/abinit(respfn_+0x1139b) [0x4901bb]
[reccar197070:32198] [ 4] ../../Prog/abinit-7.0.3/src/98_main/abinit(driver_+0x5401) [0x464b21]
[reccar197070:32198] [ 5] ../../Prog/abinit-7.0.3/src/98_main/abinit() [0x45d8d1]
[reccar197070:32198] [ 6] ../../Prog/abinit-7.0.3/src/98_main/abinit(main+0x1d) [0x45cddd]
[reccar197070:32198] [ 7] /lib/x86_64-linux-gnu/libc.so.6(__libc_start_main+0xed) [0x7f431612c76d]
[reccar197070:32198] [ 8] ../../Prog/abinit-7.0.3/src/98_main/abinit() [0x45ce0d]
[reccar197070:32198] *** End of error message ***

###################################################################

UBUNTU 12.10
.Version 6.12.3 of ABINIT
.(MPI version, prepared for a x86_64_linux_gnu4.7 computer)
---> Works

UBUNTU 12.10
.Version 7.0.3 of ABINIT
.(MPI version, prepared for a x86_64_linux_gnu4.7 computer)
---> bug after DATASET 3, after reading files from DATASET 1

Program received signal SIGSEGV: Segmentation fault - invalid memory reference.

Backtrace for this error:
#0 0x7FC73EEDB0F7
#1 0x7FC73EEDB6D4
#2 0x7FC73E62749F
#3 0xA73D5D in pawdijfr_ at pawdijfr.F90:173
#4 0x5F3D40 in dyfnl3_ at dyfnl3.F90:311
#5 0x48569A in respfn_ at respfn.F90:932
#6 0x45E1E0 in driver_ at driver.F90:595
#7 0x455ACC in abinit at abinit.F90:435

UBUNTU 12.10
.Version 7.0.3 of ABINIT
.(sequential version, prepared for a x86_64_linux_gnu4.7 computer)
---> bug after DATASET 3, after reading files from DATASET 1

Program received signal SIGSEGV: Segmentation fault - invalid memory reference.

Backtrace for this error:
#0 0x7FB9DF9830F7
#1 0x7FB9DF9836D4
#2 0x7FB9DF0CF49F
#3 0xA1742D in pawdijfr_ at pawdijfr.F90:173
#4 0x59EC1D in dyfnl3_ at dyfnl3.F90:311
#5 0x438121 in respfn_ at respfn.F90:932
#6 0x411240 in driver_ at driver.F90:595
#7 0x408C7C in abinit at abinit.F90:435

UBUNTU 12.10
(BINARIES FROM ABINIT WEBSITE)
.Version 7.0.3 of ABINIT
.(MPI version, prepared for a x86_64_linux_gnu4.5 computer)
---> bug, after DATASET 3, after reading files from DATASET 1

Calculation stopped without any comments

I have no idea of what is happening. If you want additional informations, please, let me know, I will do my best. Best regards,
Emile

####################################

Configure.ac file:

Code: Select all

prefix="/usr/local"

enable_mpi="yes"
enable_mpi_io="yes"
with_mpi_prefix="/usr"

with_fft_flavor="fftw3"
with_linalg_flavor="atlas"
with_dft_flavor="atompaw+bigdft+libxc+wannier90"
with_trio_flavor="netcdf+etsf_io+fox"
with_math_flavor="gsl"

with_fft_incs="-I/usr/include/"
with_fft_libs="-L/usr/lib -lfftw3"
with_linalg_libs="-L/usr/lib -llapack -lf77blas -lcblas -latlas"
with_libxc_incs="-I/usr/local/include"
with_libxc_libs="-L/usr/local/lib -lxc"
with_netcdf_incs="-I/usr/include"
with_netcdf_libs="-L/usr/lib -lnetcdf -lnetcdff"
with_etsf_io_incs="-I/usr/include"
with_etsf_io_libs="-L/usr/lib -letsf_io_utils -letsf_io"
with_math_incs="-I/usr/include"
with_math_libs="-L/usr/lib -lgsl -lgslcblas"

enable_gw_dpc="yes"
enable_maintainer_checks="no"
enable_test_timeout="yes"
enable_gui_build="no"


response function input file:

Code: Select all

# Crystalline Al : computation of the phonon spectrum

   ndtset   10
#Set 1 : ground state self-consistency

  getwfk1   0            # Cancel default
  kptopt1   1            # Automatic generation of k points, taking
                         # into account the symmetry
    nqpt1   0            # Cancel default
  tolvrs1   1.0d-20      # SCF stopping criterion (modify default)
  rfphon1   0            # Cancel default
 
#Q vectors for all datasets

#Complete set of symmetry-inequivalent qpt chosen to be commensurate
# with kpt mesh so that only one set of GS wave functions is needed.
#Generated automatically by running GS calculation with kptopt=1,
# nshift=0, shiftk=0 0 0 (to include gamma) and taking output kpt set
# file as qpt set. Set nstep=1 so only one iteration runs.

     nqpt   1            # One qpt for each dataset (only 0 or 1 allowed)
                         # This is the default for all datasets and must
                         #  be explicitly turned off for dataset 1.

     qpt2   0.00000000E+00  0.00000000E+00  0.00000000E+00
     qpt3   0.00000000E+00  0.00000000E+00  0.00000000E+00
     qpt4   2.50000000E-01  0.00000000E+00  0.00000000E+00
     qpt5   5.00000000E-01  0.00000000E+00  0.00000000E+00
     qpt6   2.50000000E-01  2.50000000E-01  0.00000000E+00
     qpt7   5.00000000E-01  2.50000000E-01  0.00000000E+00
     qpt8  -2.50000000E-01  2.50000000E-01  0.00000000E+00
     qpt9   5.00000000E-01  5.00000000E-01  0.00000000E+00
     qpt10 -2.50000000E-01  5.00000000E-01  2.50000000E-01

#Set 2 : Response function calculation of d/dk wave function

    iscf2   -3         # Need this non-self-consistent option for d/dk
  kptopt2   2          # Modify default to use time-reversal symmetry
  rfphon2   0          # Cancel default
  rfelfd2   2          # Calculate d/dk wave function only
  tolwfr2   1.0d-23    # Use wave function residual criterion instead

#Set 3 : Response function calculation of Q=0 phonons and electric field pert.

  getddk3   2          # d/dk wave functions from last dataset
  kptopt3   2          # Modify default to use time-reversal symmetry
  rfelfd3   3          # Electric-field perturbation response only

#Sets 4-10 : Finite-wave-vector phonon calculations (defaults for all datasets)

   getwfk   1          # Use GS wave functions from dataset1
   kptopt   3          # Need full k-point set for finite-Q response
   rfphon   1          # Do phonon response
  rfatpol   1 1        # Treat displacements of all atoms
    rfdir   1 1 1      # Do all directions (symmetry will be used)
   tolvrs   1.0d-10     # This default is active for sets 3-10

#######################################################################
#Common input variables

#Definition of the unit cell
    acell   3*7.5204497364  # This is equivalent to   10.61 10.61 10.61
    rprim   0.0  0.5  0.5   # In lessons 1 and 2, these primitive vectors
            0.5  0.0  0.5   # (to be scaled by acell) were 1 0 0  0 1 0  0 0 1
            0.5  0.5  0.0   # that is, the default.

#Definition of the atom types
   ntypat   1         # There are two types of atom
    znucl   13        # The keyword "znucl" refers to the atomic number of the
                      # possible type(s) of atom. The pseudopotential(s)
                      # mentioned in the "files" file must correspond
                      # to the type(s) of atom. Here, type 1 is the Aluminum,
                      # type 2 is the Arsenic.

#Definition of the atoms
    natom   1         # There are two atoms
    typat   1       # The first is of type 1 (Al), the second is of type 2 (As).
                       
     xred   0.0  0.0  0.0

#Gives the number of band, explicitely (do not take the default)
    nband   40         
    occopt   3
    tsmear  300.00 K

#Definition of the planewave basis set

     ecut   40.0           # Maximal kinetic energy cut-off, in Hartree
pawecutdg  120.0
pawxcdev     0

#Definition of the k-point grid
    ngkpt   4  4  4         
  nshiftk   4              # Use one copy of grid only (default)
   shiftk   0.0 0.0 0.5    # This gives the usual fcc Monkhorst-Pack grid
            0.0 0.5 0.0
            0.5 0.0 0.0
            0.5 0.5 0.5

#Definition of the SCF procedure
     iscf   7          # Self-consistent calculation, using algorithm 5
    nstep   2000         # Maximal number of SCF cycles
   diemac   9.0        # Although this is not mandatory, it is worth to
                       # precondition the SCF cycle. The model dielectric
                       # function used as the standard preconditioner
                       # is described in the "dielng" input variable section.
                       # The dielectric constant of AlAs is smaller that the one of Si (=12).
# add to conserve old < 6.7.2 behavior for calculating forces at each SCF step
 optforces 1

Emile
Posts: 15
Joined: Tue Dec 04, 2012 12:54 pm

Re: TM pseudopotential segmentation fault with 7.0.3

Post by Emile » Fri Dec 07, 2012 12:28 pm

Hello,

More informations. Al being a metal, dataset 2 and 3 are not required.
If I remove them, then response function calculations work correctly without any bugs.

So for Al (metal):
-abinit 6.12.3 RF calculations work with or without dataset 2&3.
-abinit 7.0.3, RF calculations work only without dataset 2&3.

I still have problem with my phonon dispersion curves but I will open an other topic concerning this.

Also I believed that with abinit 7.0.3 it would be possible to make response function calculations with GGA PAW atomic data.
I now think it not possible.

Best regards,
Emile

yurtesen
Posts: 15
Joined: Sun Nov 11, 2012 9:42 pm

Re: TM pseudopotential segmentation fault with 7.0.3

Post by yurtesen » Wed Dec 12, 2012 12:45 pm

Problem appears to be continuing with 7.0.4

yurtesen
Posts: 15
Joined: Sun Nov 11, 2012 9:42 pm

Re: TM pseudopotential segmentation fault with 7.0.3

Post by yurtesen » Mon Jan 28, 2013 12:03 pm

Problem appears to be continuing with 7.0.5.
Please let us know how to aid further in fixing the issue!

Locked