[QE-users] GW calculation, wfreq segmentation fault ocurred

[Lucas Nicolás Lodeiro Moraga]

Dear QE users:
I am a newbie in GW calculations, and i need to improve the band gap with
this type of calculation
I have done a GW calculation for a 2x2x2 supercell of Cu2O material. I use
a GGA functional for the scf calculation with a k points grid, then a nscf
calculation to take the wfc for the gamma point to do the GW calculation.
First the wstat calculation with 256 number of pdep eigenpotentials (its
the first aproximation for the GW expansion) finished without problems,
then the wfreq calculation for 11 states , the Valence Band Maximum (at
gamma point) and 5 bottom and 5 over it. In this calculation, the cluster
report inform (runpw6.1-impi5.0.3-intel15.0.1-dfti.job.e13139803):

forrtl: severe (174): SIGSEGV, segmentation fault occurred
Image              PC                Routine            Line        Source

wfreq.x            0000000003EBABD1  Unknown               Unknown  Unknown
wfreq.x            0000000003EB9327  Unknown               Unknown  Unknown
wfreq.x            0000000003E69E04  Unknown               Unknown  Unknown
wfreq.x            0000000003E69C16  Unknown               Unknown  Unknown
wfreq.x            0000000003DFBD26  Unknown               Unknown  Unknown
wfreq.x            0000000003E04BCD  Unknown               Unknown  Unknown
libpthread.so.0    00002B71366F46D0  Unknown               Unknown  Unknown
wfreq.x            0000000000460302  linsolve_sternhei          99
linsolve_sternheimer_m_wfcts.f90
wfreq.x            000000000041EB9B  solve_wfreq_k_            834
solve_wfreq.f90
wfreq.x            00000000004496D0  MAIN__                     65
wfreq.f90
wfreq.x            0000000000406B2E  Unknown               Unknown  Unknown
libc.so.6          00002B7137A45445  Unknown               Unknown  Unknown
wfreq.x            0000000000406A29  Unknown               Unknown  Unknown

the wfreq.out file finish in the W)-Lanczos stage, but does not inform
anything.
This calculation creates a o-ehf_K00001.tab file, but i do not know if its
the converged values or just an intermediated values for the states.

the files are in the next dropbox link:

https://www.dropbox.com/sh/4hix3fkwgb7i9jl/AADj_EtLQvjbrYZ1SMflp3Q7a?dl=0

I use the QE 6.1 in a cluster with 120 cores and 6 nodes. I use as a probe
a little cluster with 1 node and 62 cores and QE 6.1, and the error is the
same.

Does somebody know what happens and why fix it?

Regards
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.quantum-espresso.org/pipermail/users/attachments/20181023/3bf48f05/attachment.html>