[QE-developers] Seg fault errors in QE 6.8 for jobs using smearing

[Paolo Giannozzi]

Many tests (also with smearing) are compiled with different compilers
(including gfortran 10.2) and run during the night (CET time):
http://bionano2.dimi.uniud.it:8010/#/console. Unless you have more
evidence, what you report seems to indicate that the problem is in the
compiler, not in the code.

Paolo

On Tue, Nov 23, 2021 at 8:11 AM Bainglass, Edan <Edan.Bainglass at unt.edu>
wrote:

> *Hello,*
>
>
>
> *For gfortran(10.3.0)-compiled QE 6.8, I’m getting the following error for
> jobs that include smearing:*
>
>
>
> *Program received signal SIGSEGV: Segmentation fault - invalid memory
> reference.*
>
>
>
> Backtrace for this error:
>
> #0  0x146941220d21 in ???
>
> #1  0x14694121fef5 in ???
>
> #2  0x1469414dd3bf in ???
>
> #3  0x7ffd48ddd9cc in ???
>
>
>
> *Job terminates right after this line:*
>
>
>
>      iteration #  1     ecut=    35.00 Ry     beta= 0.70
>
>
>
> *This does not occur with the latest Intel compiler. It also does not
> occur with either compiler on QE 6.7.*
>
>
>
> *For completeness, here are a few jobs that fail:*
>
>
>
> *1.*
>
>
>
> &CONTROL
>
>    !
>
>    calculation = 'scf'
>
>    restart_mode = 'from_scratch'
>
>    tprnfor = .TRUE.
>
>    !
>
> /
>
> &SYSTEM
>
>    !
>
>    ibrav = 8
>
>    celldm(1) = 10.6881
>
>    celldm(2) = 0.866025
>
>    celldm(3) = 3.95422
>
>    nat = 10
>
>    ntyp = 3
>
>    ecutwfc = 35
>
>    ecutrho = 300
>
>    occupations = 'smearing'
>
>    degauss = 0.03
>
>    smearing = 'mv'
>
>    nbnd = 80
>
>    tot_charge = 0
>
>    !
>
> /
>
> &ELECTRONS
>
>    !
>
>    conv_thr = 5.D-6
>
>    !
>
> /
>
> K_POINTS gamma
>
> ATOMIC_SPECIES
>
> C 1 C.pbe-rrkjus.UPF
>
> O 1 O.pbe-rrkjus.UPF
>
> Pt 1 Pt.pbe-n-rrkjus_psl.0.1.UPF
>
> ATOMIC_POSITIONS (bohr)
>
> C        5.335084148   4.646723426  12.901029877
>
> O        5.335009643   4.619623254  15.079854269
>
> Pt       8.061327071   0.098057998   8.992142901
>
> Pt       2.608989366   0.098058283   8.992140585
>
> Pt       0.000036609   4.720846294   8.968756935
>
> Pt       5.335159557   4.721612729   9.380196435
>
> Pt       0.000041121   7.802951963   4.604626508
>
> Pt       5.335161233   7.697749113   4.753489408
>
> Pt       2.697860636   3.152173889   4.688412329
>
> Pt       7.972463687   3.152174491   4.688415209
>
>
>
> *2.*
>
>
>
> &CONTROL
>
>    !
>
>    calculation = 'scf'
>
>    restart_mode = 'from_scratch'
>
>    tprnfor = .TRUE.
>
>    !
>
> /
>
> &SYSTEM
>
>    !
>
>    ecutrho = 300
>
>    ecutwfc = 30
>
>    ibrav = 1
>
>    celldm( 1 ) = 20
>
>    nat = 2
>
>    ntyp = 2
>
>    nbnd = 6
>
>    tot_charge = 0
>
>    nspin = 2
>
>    starting_magnetization(1) = 1
>
>    occupations = 'smearing'
>
>    smearing = 'gaussian'
>
>    degauss = 0.01
>
>    nosym = .true.
>
>    !
>
> /
>
> &ELECTRONS
>
>    !
>
>    conv_thr = 5.D-3
>
>    !
>
> /
>
> K_POINTS (gamma)
>
> ATOMIC_SPECIES
>
> C   12      C.pbe-rrkjus.UPF
>
> H    1        H.pbe-rrkjus.UPF
>
> ATOMIC_POSITIONS (angstrom)
>
> C        0.00000000      0.00000000      0.16007400
>
> H        0.00000000      0.00000000     -0.96044600
>
>
>
> *3.*
>
>
>
> &CONTROL
>
>     calculation = 'scf',
>
>     tprnfor = .true.,
>
> /
>
> &SYSTEM
>
>     nspin = 1,
>
>     ecutwfc = 35,
>
>     nat = 2,
>
>     occupations = 'smearing',
>
>     ibrav = 0,
>
>     degauss = 0.01,
>
>     smearing = 'marzari-vanderbilt',
>
>     tot_charge = 0,
>
>     ntyp = 1,
>
>     ecutrho = 350,
>
> /
>
> &ELECTRONS
>
>     mixing_beta = 0.3
>
>     conv_thr = 1e-06,
>
> /
>
> &IONS
>
> /
>
> ATOMIC_SPECIES
>
> Ag 107.868000 Ag.pbe-n-rrkjus_psl.1.0.0.UPF
>
> ATOMIC_POSITIONS {bohr}
>
> Ag 2.771763  2.771763  18.897260
>
>  Ag 0.000000  0.000000  22.740340
>
> K_POINTS {automatic}
>
>    1 1 1   0 0 0
>
> CELL_PARAMETERS {bohr}
>
> 5.543526 0.000000 0.000000
>
> 0.000000 5.543526 0.000000
>
> 0.000000 0.000000 65.08548
>
>
>
>
>
> *Thanks in advance for any info,*
>
>
>
> *Dr. Edan Bainglass*
>
> *Postdoctoral Research Assistant*
>
> *Environ Developer*
>
> *MATERIALab Group*
>
> *Physics Department*
>
> *University of North Texas*
>
>
> _______________________________________________
> developers mailing list
> developers at lists.quantum-espresso.org
> https://lists.quantum-espresso.org/mailman/listinfo/developers
>


-- 
Paolo Giannozzi, Dip. Scienze Matematiche Informatiche e Fisiche,
Univ. Udine, via delle Scienze 206, 33100 Udine, Italy
Phone +39-0432-558216, fax +39-0432-558222
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.quantum-espresso.org/pipermail/developers/attachments/20211123/afd9d80f/attachment-0001.html>