Density of states for SrMnO3 doesn't start

Elk Users
Natalie
2013-06-25
2013-07-02
  • Natalie
    Natalie
    2013-06-25

    Hello everyone,

    When I try to run a DOS calculation on spin polarised SrMnO3, the code stops after a minute without producing any output or error message. Other materials don't show this problem.

    I've re-run the SCF calculation with a high k-point density (10x10x10 k-points) and subsequently started a DOS calculation (task 10) in the same folder. This worked for other elements and oxides as well, and also when I only do Sr, the TDOS.OUT and partial DOS are created. I've attached the elk.in.

    Thanks a lot for the help and best wishes,

    Natalie

    tasks
    10

    nwrite
    2

    !number for exchange correlation functional: 3 = LSDA and 21 (20) = revised (nonrevised) PBE
    xctype
    21

    spinpol
    .true.

    !cut-off energy
    rgkmax
    7.0

    !set .true. if rgkmax is high
    trimvg
    .true.

    beta0
    0.025

    betamax
    0.1

    epspot
    0.00001

    ! fairly large number of empty states required for magnetic cases
    nempty
    10

    avec
    0.8660254038 -0.5000000000 0.0000000000
    0.0000000000 1.0000000000 0.0000000000
    0.0000000000 0.0000000000 1.6663114965

    scale
    10.2865350527

    atoms
    3 : nspecies
    'Sr.in' : spfname
    4
    0.000000000000000 0.000000000000000 0.000000000000000 0.00000000 0.00000000 0.00000000
    0.000000000000000 0.000000000000000 0.500000000000000 0.00000000 0.00000000 0.00000000
    0.333333333333333 0.666666666666667 0.250000000000000 0.00000000 0.00000000 0.00000000
    0.666666666666667 0.333333333333333 0.750000000000000 0.00000000 0.00000000 0.00000000
    'Mn.in' : spfname
    4
    0.333333333333333 0.666666666666667 0.612200000000000 1.00000000 0.00000000 0.00000000
    0.666666666666667 0.333333333333333 0.112200000000000 1.00000000 0.00000000 0.00000000
    0.333333333333333 0.666666666666667 0.887800000000000 -1.00000000 0.00000000 0.00000000
    0.666666666666667 0.333333333333333 0.387800000000000 -1.00000000 0.00000000 0.00000000
    'O.in' : spfname
    12
    0.500000000000000 0.000000000000000 0.000000000000000 0.00000000 0.00000000 0.00000000
    0.000000000000000 0.500000000000000 0.500000000000000 0.00000000 0.00000000 0.00000000
    0.500000000000000 0.000000000000000 0.500000000000000 0.00000000 0.00000000 0.00000000
    0.500000000000000 0.500000000000000 0.500000000000000 0.00000000 0.00000000 0.00000000
    0.000000000000000 0.500000000000000 0.000000000000000 0.00000000 0.00000000 0.00000000
    0.500000000000000 0.500000000000000 0.000000000000000 0.00000000 0.00000000 0.00000000
    0.819300000000000 0.638600000000000 0.250000000000000 0.00000000 0.00000000 0.00000000
    0.638600000000000 0.819300000000000 0.750000000000000 0.00000000 0.00000000 0.00000000
    0.180700000000000 0.361400000000000 0.750000000000000 0.00000000 0.00000000 0.00000000
    0.180700000000000 0.819300000000000 0.750000000000000 0.00000000 0.00000000 0.00000000
    0.361400000000000 0.180700000000000 0.250000000000000 0.00000000 0.00000000 0.00000000
    0.819300000000000 0.180700000000000 0.250000000000000 0.00000000 0.00000000 0.00000000

    sppath
    '/home/elk-1.4.22/species/'

    ngridk
    10 10 10

    vkloff
    0.5 0.5 0.5

     
  • Markus
    Markus
    2013-06-25

    Hi Natalie,

    this is quite a big calculation. Does the code eventually run out-of-memory?

    By the way, your nempty is a bit low. For a magnetic system you need something like nempty = something * natoms * nspin, where something is 3 to 5 or so. This depends on the system. Remember that elk treats magnetism second-variationally, so the empty states are used to construct the basis for the second variation. Try at least 40 empty states for good results.

    Regards,
    Markus

     
  • Natalie
    Natalie
    2013-07-02

    Dear Markus,

    Thanks a lot for your quick reply! I've run the calculation with nempty = 40 and on 48 CPUs and it finished. Thanks for your help!

    Best wishes,
    Natalie