Possible bug in Spin-Texture Calculation

exciting
1

Dear Sir/Madam,

I am writing to you regarding the spin-texture calculation for GaAs compound using the EXCITING code.
I have followed the instructions on the EXCITING website (Spin-Texture Calculation - exciting) to perform the spin-texture calculation for GaAs compound. However, when I try to reproduce the results using other DFT codes such as Quantum Espresso or VASP, I notice a discrepancy in the direction of the Sx component of the spin texture plot. It seems that the Sx component has to be multiplied by “-1” to obtain the same texture as the EXCITING code.

I wonder if this is due to a small bug in the calculation of spin texture by the EXCITING code, or if there is another explanation for this difference. Could you please investigate this possibility and let me know your findings?
Thank you for your time and attention
.
Sincerely yours,
Mohammad Sandoghchi

2

Dear Mohammad,

That is an interesting result. I went through the math and the implementation and can not see where a bug might be.

Maybe it is how spin dependent states and spin orbit coupling is treated in the codes? In exciting we use a second variational ansatz. We first calculate the solution for a spin independent system and use this to solve for the spin dependent system with spin orbit coupling. For the details I kindly refer you to our references.

Cheers
Benedikt