About release notes v0.10.0 September 1, 2024

[Jaime Garrido]

Hi!

I have read in the release notes v0.10.0 September 1, 2024:

"Allow computing anisotropic J and DMI without three or more calculations within ABACUS and SIESTA interfaces."

Does this mean that from v.0.10.0 and above, we do not need the 3 spin orientations to calculate J and DMI with SIESTA? Is there any limitation or downside for doing this instead of the three calculations+averaging? I am concerned about if this shortcut implies any loss of accuracy.

That is all

Thanks in advance

Jaime

[Xu He]

Hello Jaime, 

Yes that can avoid the using of three spin orientaions. 

The merging of the result is based on the approximation that the electron state does not change too much with the different spin orientations (within the three DFT SCF calculation). 

The alternative way, is to simply rotate the spin and do a non-scf calculation within TB2J, and then do the merging with the same method. It is based on the same approximation but in a different way. 

Both of them will fail when the approximation is invalid.  In that case, I don't think there is a universal J and DMI with the spin rotated. 

Best regards, 

HeXu