TB2J rotation

[Dorye Esteras]

Dear developers,

I am doing rotations to calculate noncollinear exchange parameters and I have two questions.

1)  I remember that some years ago, rotating the structure instead of the spin required to update the kpoints of the calculations for example 9 9 1 to 9 1 9 etc... However, in the guide, I see some indications to keep the original kpoints even when rotating the structure:

https://docs.siesta-project.org/projects/siesta/en/school-2021/tutorials/tb2j/
https://tb2j.readthedocs.io/en/latest/src/rotate_and_merge.html

Did something change? Because in the structure rotations, I still see that the cell_matrix is rotated as well.

2) I am using siesta (I found here is a problem to rotate the spins). I have seen there is a tool to rotate the density matrix, but the process is still not clear for me.

If I directly use the new DMs with tb2j I found there is no difference in the tb2j results, but if I run siesta calculations starting from these new density matrices, my spins are still in z. 

Can you give me more details about this approach?

Thanks in advance,

Dorye

[Xu He]

Hello, 

1) The k-point should be kept the same. Indeed in the very early stage of the documentation I made a mistake to suggest the changing of the k-points, but it was soon corrected. This is because the k-point we use are rotated with the cell if reduced-coordinates are used (which is often the default). 

2) The tool to rotate the density matrix is not supposed to be used with the rotation of the strucuture. What it does is to rotate the spin orientation in the density matrix, which can be used as the initial density matrix for the calculation with spin rotation.  

Best regards, 

HeXu

[Satish Kumar]

Hi anyone who is export in TB2J please help me to learn this how to use in siesta if anyone please write to me on 97sat...@gmail.com