[QE-users] Non-converging vc-relaxation calculation

[Stephen Zhang]

Hi everyone,

I've encountered the issue of my vc-relax calculation not converging. Here
is my input,
############################################################
&control
    calculation = 'vc-relax'
    prefix = 'CoSnS'
    outdir = './outdir'
    pseudo_dir = '/home/stephen/qe-6.5/SSSP_precision_pseudos'
    etot_conv_thr = 1e-6
    forc_conv_thr = 1e-5
/
&system
    ibrav=5, celldm(1) =10.16896836537393, celldm(4) = 0.4980629252775897,
    nat=7, ntyp=7,
    ecutwfc=50,
    ecutwfc=500,
    occupations='smearing',smearing='gaussian',degauss=0.1,
    nspin=2,
    starting_magnetization(1) = 0.24,
    starting_magnetization(2) = 0.25,
    starting_magnetization(3) = 0.3,
    starting_magnetization(4) = 0.02,
    starting_magnetization(5) = 0.04,
    starting_magnetization(6) = 0.09,
    starting_magnetization(7) = 0.01,
/
&electrons
    conv_thr=1e-9
/
&ions
/
&cell
    cell_dofree='ibrav'
/
ATOMIC_SPECIES
 Co1  58.933195 Co.pbe-spn-kjpaw_psl.0.3.1.UPF
 Co2  58.933195 Co.pbe-spn-kjpaw_psl.0.3.1.UPF
 Co3  58.933195 Co.pbe-spn-kjpaw_psl.0.3.1.UPF
 Sn4  118.71 Sn.pbe-dn-kjpaw_psl.1.0.0.UPF
 Sn5  118.71 Sn.pbe-dn-kjpaw_psl.1.0.0.UPF
 S6  32.065 S.pbe-n-kjpaw_psl.1.0.0.UPF
 S7  32.065 S.pbe-n-kjpaw_psl.1.0.0.UPF
ATOMIC_POSITIONS {crystal_b}
 Co1 0.5000 0.0000 0.0000
 Co2 0.0000 0.5000 0.0000
 Co3 0.0000 0.0000 0.5000
 Sn4 0.5000 0.5000 0.5000
 Sn5 0.0000 0.0000 0.0000
 S6 0.7195 0.7195 0.7195
 S7 0.2805 0.2805 0.2805
K_POINTS (automatic)
  3 3 3 0 0 0
#####################################################################3
A recurring problem I've seen is that the scf accuracy will decrease to a
certain point and then jump up again. According to the pw_user_guide
<https://www.quantum-espresso.org/Doc/pw_user_guide.pdf> this is due to the
fact that I am using few k-points and have a metallic material. The guide
recommends the following:

Usually one can solve the problem by adding a few empty bands and a small
broadening

However, I'm not sure on how to do either. Do i just increase the nbnd
number and increase the degauss value?

Thank you,
Stephen

-- 
*University of California, Berkeley*
*Department of Letter and Sciences*
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.quantum-espresso.org/pipermail/users/attachments/20200716/2fdebac0/attachment.html>