Hello again,
I've tried (with success) to achieve goal using nspin=2 with combined
with total_magnetization=2, yet the second approach using
occupations='from_input' and OCCUPATIONS card is still without success.
I clearly understand wrong this card, I was using this (for hydrogene
molecule):
OCCUPATIONS
1
0
1
0
----------------------
OCCUPATIONS
2
----------------------
OCCUPATIONS
2
0
----------------------
OCCUPATIONS
1 1
----------------------
...and few more. The documentation is not excessively helpful but if I
understand the scheme in OCCUPATIONS card is:
OCCUPATIONS
x x x x <-number of UP spins in 1s 2s 2p 3s respectively (for
example for first (eg. sodium) atom)
y y y y <-number of DOWN spins in 1s 2s 2p 3s respectively (for
first atom)
z z z z <-number of UP spins in second atom etc..
but always I got: "strange occupations: number of electrons from
occupations is wrong."
What is the correct input for OCCUPATIONS card for two hydriogene atoms?
Konrad
W dniu 12.06.2018 o 21:46, Lorenzo Paulatto pisze:
On 06/12/2018 06:09 PM, Konrad Gruszka wrote:
OCCUPATIONS
1
1
nspin=2; and tot_magnetization=2
Are those two inputs equivalent?
No, the first one means "one electron in the first band spin up and
one in the first ban spin down", which means no magnetization, or
tot_magnetization=0.
Also note that you need to specify nspin=2 in both cases. Also, in the
first case, setting nspin=1 would have exactly the same effect (for
cheaper, not that it matters in H2 molecule) as the occupations are
the same for both spin polarizations.
Also, will this give proper results when including spin-orbit with
full relativistic pseudopotential or do I need to do it another way?
I'm not sure if occupations="from_input" works with non-colinear
magnetization. I'm quite sure that tot_magnetization does NOT, instead
you have to use fixed_magnetization="total" and set the
tot_magnetization vector. This method is powerful, a bit complicated,
but quite well documented.
_______________________________________________
users mailing list
[email protected]
https://lists.quantum-espresso.org/mailman/listinfo/users
