PAW dataset and spin polarized calculations  [SOLVED]

External optional components of Abinit (BigDFT, Wannier90...) and codes using/providing data from/to Abinit (AtomPAW, ONCVPSP, EXC, DP, Yambo...)

Moderators: ebousquet, bxu

Locked
rajaraman
Posts: 7
Joined: Sat Sep 15, 2018 7:08 pm
Location: Kalpakkam, India

PAW dataset and spin polarized calculations

Post by rajaraman » Sun Sep 23, 2018 8:26 am

I was testing spin polarised tutorials as I need to work with Fe based systems.
while the abinit provided normconserving PSP is working exactly as given in tutorial, paw potential does not give spin polarisation. Am I missing additional settings?
here is the "occ" info from outputs.
PSPNC (26fe.pspnc)
occopt = 3 xclevel = 1
occopt = 3 xclevel = 1
occ1 2.000000 2.000000 2.000000 2.000000 0.000000 0.000000
occ2 1.000000 1.000000 1.000000 1.000000 0.000000 0.000000
occopt 3
occupation numbers for kpt# 1
occupation numbers for kpt# 1
occupation numbers for kpt# 1
occ1 2.000000 1.999891 1.999891 1.229151 1.229151 0.286756
occ2 1.000000 0.999995 0.999995 0.983960 0.983960 0.694675
occopt 3

PAW (Fe.GGA_PBE-JTH.xml)
occopt = 3 xclevel = 2
occopt = 3 xclevel = 2
occ1 2.000000 2.000000 2.000000 2.000000 2.000000 2.000000
occ2 1.000000 1.000000 1.000000 1.000000 1.000000 1.000000
occopt 3
Augmentation waves occupancies Rhoij:
occupation numbers for kpt# 1
Augmentation waves occupancies Rhoij:
occupation numbers for kpt# 1
occupation numbers for kpt# 1
occ1 2.000000 2.000000 2.000000 2.000000 2.000000 2.000000
occ2 1.000000 1.000000 1.000000 1.000000 1.000000 1.000000
occopt 3
Attachments
fe.in
paw input
(945 Bytes) Downloaded 274 times
fe.out
paw output
(50.74 KiB) Downloaded 286 times

rajaraman
Posts: 7
Joined: Sat Sep 15, 2018 7:08 pm
Location: Kalpakkam, India

Re: PAW dataset and spin polarized calculations

Post by rajaraman » Sun Sep 23, 2018 11:56 am

Found that the spin tutorial potential file (26fe.pspnc) had 8 valance electrons, while PAW treats 16 valance electrons.
generated new paw potential with 8 valance electrons (4s2-3d6) (see attached file).
This one atleast started working for spin polarisation, though the magnetisation value is still off.
Mya be I have to tune the PAW potential?
Is it still possible to use the 16 valance electrons case for paw potential and tune other paramaters to work with spin polarised calculations?

Fe_prim_111/spin/paw_8e$ grep Magnet ../pspnc/fe.out
Magnetization (Bohr magneton)= 1.96743463E+00

Fe_prim_111/spin/paw_8e$ grep Magnet fe.out0001
Magnetization (Bohr magneton)= 2.32388264E+00

ebousquet
Posts: 469
Joined: Tue Apr 19, 2011 11:13 am
Location: University of Liege, Belgium

Re: PAW dataset and spin polarized calculations  [SOLVED]

Post by ebousquet » Tue Oct 09, 2018 5:53 pm

Dear rajaraman,
Yes you can use the 16 electrons PAW dataset (JTH I guess?), no problem. However, adapt the number of band of your calculation to be sure. Be careful that the grid of k-point is not converged at all for this kind of system, please check how it goes with k-point grid.
For the pseudopotentials I would advise you to use the ones from the pseudodojo website:
http://www.pseudo-dojo.org/
Best wishes,
Eric

rajaraman
Posts: 7
Joined: Sat Sep 15, 2018 7:08 pm
Location: Kalpakkam, India

Re: PAW dataset and spin polarized calculations

Post by rajaraman » Sat Oct 27, 2018 12:25 pm

Thank you Eric.
Sure I will check convergence.
regards'
-rajaraman

Locked