Segmentation fault p2y - qe 5.2.0, yambo 4.0.1

[Michael.Friedrich]

Hello,

by importing the databases from quantum espresso 5.2.0 (also tested 5.1.1 and 5.0.3) to yambo 4.0.1 I get the following error.

<---> DBs path set to .
<---> Index file set to data-file.xml
<---> Header/K-points/Energies... done
<---> Cell data... done
<---> Atomic data... done
<---> Symmetries...[SI no]......[-I no]...[TR yes][dhcp-n3physik-93:86516] *** Process received signal ***
[dhcp-n3physik-93:86516] Signal: Segmentation fault: 11 (11)
[dhcp-n3physik-93:86516] Signal code: (0)
[dhcp-n3physik-93:86516] Failing at address: 0x0
[dhcp-n3physik-93:86516] *** End of error message ***
Segmentation fault: 11

The same holds for qe 5.0.3 and yambo 3.4.1 installed on our cluster, though with a different error message

<---> P(W) 2 Y(ambo) Ver. 5.0
<---> DBs path set to .
<---> Index file set to data-file.xml
<---> Header/K-points/Energies...done
<---> Cell data...done
<---> Atomic data...done
<---> Symmetries...[SI no]......[-I no]...[TR yes][fe1:mpi_rank_0][error_sighandler] Caught error: Segmentation fault (signal 11)
Segmentation fault (core dumped)

Are the ONCV pseudopotentials which I (try to) use for my calculations a problem? I attached my calculation (input files, psps, data-file.xml).

Here's also my configuration file.

./configure \
PFC="/Users/michaelf/Tools/openmpi-1.8.8/intallation/bin/mpif90" \
MPICC="/Users/michaelf/Tools/openmpi-1.8.8/intallation/bin/mpicc" \
\
CC=/sw/lib/gcc4.9/bin/gcc-fsf-4.9 \
CPP="/sw/lib/gcc4.9/bin/cpp-fsf-4.9 -E -ansi" \
FC="/sw/lib/gcc4.9/bin/gfortran-fsf-4.9" \
F77=/sw/lib/gcc4.9/bin/gfortran-fsf-4.9 \
FCCPP=/sw/lib/gcc4.9/bin/cpp-fsf-4.9 FCFLAGS=-m64 CFLAGS=-m64 \
\
--enable-msgs-comps \
--enable-dp \
--enable-open-mp \
--with-mpi-libs="-L/Users/michaelf/Tools/openmpi-1.8.8/intallation/lib" \
\
--with-fft-libs="-L/sw/lib/ -lfftw3 -lfftw3f" --with-fft-includedir="-I/sw/include/ autoreconf -vif" \
\
--enable-iotk --with-iotk-path="/Users/michaelf/tools/espresso-5.2.0/iotk/" \
--with-p2y-version="5.0" \
\
--enable-netcdf-hdf5 \
--enable-netcdf-LFS \
--with-netcdf-path=/usr/local/ \
--with-hdf5-libdir=/usr/local/ \

Thank you for your help
Michael Friedrich

[myrta gruning]

dear Michael

I do not think it is related to the PP since it went through the atomic data part.
It may be related with the XC which is the next thing after symmetries. I remember the problem with the 3.4.1 though I thought that was fixed.
I will look at that. Thank you for signalling the seg fault.

Best,
m

[andrea.ferretti]

Dear Michael,

I think Myrta is right, I had a similar problem and it was caused by the following lines of data-file.xml

Code: Select all

  <EXCHANGE_CORRELATION>
    <DFT type="character" size="1" len="20">
PZ
    </DFT>
  </EXCHANGE_CORRELATION>

Basically, PZ is used as a kind of jargon within QE, but it is not recognised by Yambo/libxc.
You can fix the problem by changing PZ to LDA or

Code: Select all

    <DFT type="character" size="1" len="20">
 SLA  PZ   NOGX NOGC
    </DFT>

Since PZ is allowed by QE, this has to be fixed within Yambo in one of the next releases.
Thanks for reporting

Andrea

BTW: note that the string bearing the XC name (PZ, LDA, or "SLA PZ NOGX NOGC") should be 20 digits long, as stated by the len attribute.
attention must be paid when replacing values

[Michael.Friedrich]

Thank you very much for your very fast reply!
For the installation on the cluster the problem is solved with the suggested change in the xml file.
The problem for the local installation on my mac remains.
I copied the qe output LN.save to the cluster, imported the database with p2y and copied everything back to my mac and yambo works fine (so I can live with that ).
I just wanted to let you know that If I use p2y on my mac I still obtain the segmentation fault posted above...

Best regards
Michael