about openmx

371 views
Skip to first unread message

chao zhou

unread,
Dec 23, 2021, 7:09:45 AM12/23/21
to TB2J
Dear HeXu
with the openmx.dat I set, I want to run the GGA+U calculation ,but it report the error 
Invalid flag for LDA+U (Atoms.SpeciesAndCoordinates)  0
Please check your input file
the coordinates is wrong ,why is it? I went to the forum and couldn’t find the answer, so I want some help.
zhouchao
openmx (2).dat

Xu He

unread,
Dec 23, 2021, 12:07:10 PM12/23/21
to chao zhou, TB2J

Dear zhouchao,

The openmx input need some extra terms in the Atoms.SpeciesAndCoordinates when LDA+U is turned on.

See this page in the manual:

http://www.openmx-square.org/openmx_man3.9/node109.html

HeXu

--
You received this message because you are subscribed to the Google Groups "TB2J" group.
To unsubscribe from this group and stop receiving emails from it, send an email to tb2j+uns...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/tb2j/fcf4d9d9-4090-45aa-9de6-1dc3879222aan%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

chao zhou

unread,
Dec 23, 2021, 9:14:29 PM12/23/21
to TB2J
I see,but i set the spinpolarization  "On" not ‘’NC‘’,why I should set the Noncolinear calculation,but with the NC calculation I set,it still cannot run.
So what's wrong  with it.

Xu He

unread,
Dec 25, 2021, 4:46:03 AM12/25/21
to TB2J

Hi,

I think you can find both the format for the input of both the collinear and non-collinear on that webpage. There is one example of the noncollinear input here:

https://github.com/mailhexu/TB2J-OpenMX/blob/master/examples/SrMnO3_FM_SOC/openmx.dat

Perhaps you can check the difference in the format. I am not an experienced OpenMX user, but sometimes the DFT codes are quite strict about the format, like the number of spaces.

Best regards,

HeXu

chao zhou

unread,
Dec 27, 2021, 9:24:05 AM12/27/21
to TB2J
Dear HeXu
I have run openmx successfully.After the same structure is optimized and self-consistent under OpenmX and Vasp, the magnetic moment information is different, and therefore there is also a qualitative difference in the description of the  exchange interaction. Of course, one is the local orbit basis vector and the other is the plane wave basis vector.How to get a balance theoretically and experimentally, for example, consider U, but the determination of U seems to be less sophisticated.Also with differences in the results of different  software, I want to ask  your opinion, I have been disturbing you all the time, thank you very much.
zhouchao

Xu He

unread,
Dec 27, 2021, 9:53:17 AM12/27/21
to chao zhou, TB2J

Dear zhouchao,

Regarding the difference between the two codes, you can check:

- How large is the difference between the magnetic moments? The local magnetic moment is defined as the sum inside the paw sphere in VASP and the projection to the atomic orbitals in OpenMX. So there could be some small differences.

- Check if the basis set in OpenMX is good enough by comparing e.g. the forces in a slightly distorted structure in the two codes. You can do that by using the same functional/kpoints/smearing WITHOUT +U.  Also you can

try with different basis sets, to see if the results vary.

- With +U, if you use the same U value, the result can be different, as the orbital occupations are defined differently (paw sphere vs projection to atomic orbitals).

- If the DFT results are close, there can still be some differences in the exchange values from TB2J. As the perturbation of spin rotation is different (Wannier functions vs. atomic orbitals).

- In many systems indeed +U needs to be well tuned. Many (if not all) magnetic systems are not well described by the functionals at the LDA/GGA level, as magnetism is often related the strong correlation.

Best wishes,

HeXu

Reply all
Reply to author
Forward
0 new messages