[Pw_forum] Convergence issues with Bulk Magnesium

[MMS]

Hi,

I am trying to simulate bulk Mg in the hcp structure. Before beginning my
structural relaxation I wanted to test convergence of the energy values
with respect to ecutwfc. This is one of the input files.

&control
        calculation='scf'
        title='Mgscf'
        verbosity='high'
        restart_mode='from_scratch'
        prefix='Mg'
        tstress=.true.
        tprnfor=.true.
        pseudo_dir='./pseudo/'
        outdir='./'
/
&system
        ibrav=4
        celldm(1)=6.0
        celldm(3)=1.623
        nat=2
        ntyp=1
        nbnd=10
        ecutwfc=30
        ecutrho=240
        occupations='smearing'
        degauss=0.015
        smearing='mv'
/
&electrons
        diagonalization='david'
        mixing_mode='plain'
        mixing_beta=0.7
        conv_thr=1.0D-9
/
ATOMIC_SPECIES
Mg 24.305 Mg.pz-n-vbc.UPF
ATOMIC_POSITIONS {crystal}
Mg 1/3 2/3 1/4
Mg 2/3 1/3 3/4
K_POINTS {automatic}
6 6 4 0 0 0

I kept varying ecutwfc, and ecutrho was parametrized to be eight times
ecutwfc. I had used the perdew-zunger LDA pseudopotential.

Now the ecutwfc values versus the energy values in Rydberg are shown below
20 -4.28664244
30 -4.28693764
40 -4.28699724
50 -4.28701233
60 -4.28704753
70 -4.28717529
80 -4.28739243
90 -4.28771080
100 -4.28803666
110 -4.28853887
120 -4.28941450
130 -4.28990855
140 -4.29054764
150 -4.29142411
160 -4.29309040

What I notice is that the change in energy values (different between
successive values) initially start decreasing, but slowly start increasing
and oscillating but always positive, meaning that the energy was
continually decreasing. This doesnot help me conclude ecutwfc's convergence
point.

Now, I tried varying the degauss values and even tried increasing the k
points, but I still see the same behavior. Can I get some pointers in this
regard?

Meenakshi Sundaram
Graduate Student
Mechanical and Aerospace Engg.
Cornell
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.quantum-espresso.org/pipermail/users/attachments/20120705/a13d1834/attachment.html>