[QE-users] pw.x - choosing starting_magnetization value for specific expected total magnetization

[Abdul Muhaymin]

Thanks Guido. I understand this starting_magnetization switch now. I 
wasn't using DFT+U since I first wanted to see what happens without any 
U correction and planned to use U correction later. But this is a good 
suggestion to set a near 0 U value to have the occupation matrix in the 
output. I will try this. Thank you

Sincerely,
Abdul

On 2/21/24 18:37, Abdul Muhaymin wrote:
> Hello QE users,
>
> As I understood that thestarting_magnetization is used to break the 
> spin symmetry and can affect the solution. When I did a simple 
> calculation changing starting_magnetization, it always converged to 
> the same ground state. However, I noticed that for other structures, 
> this is not the case. Since there are many local minima, the 
> starting_magnetization value determines on what magnetic solution the 
> calculation converges to. So my question is then how we should choose 
> it? For example, I am trying to simulate a Cu dopant in ZnS. Since 
> Cu2+ has 9 electron in its 3d subshell, a neutral dopant then only has 
> 1 unpaired electron. So, should I expect 1 Bohr mag/cell? When I set 
> tot_charge=-1 to simulate a charged defect, I got everything as 0.0000 
> as expected since there will be no unpaired electron. But simulating 
> the neutral dopant, I am getting 0.00 total and absolute 
> magnetization. Inspecting the projwfc.x output, I see that the dopant 
> atom has 0.0018 polarization. Why is this the case? Why not the 
> polarization is 1.00 or close to 1?
>
> My second question is how I can correctly convert my expected 
> magnetization to starting magnetization value. For example, if I 
> expect the magnetization per cell to be 3 Bohr magneton, what should I 
> choose as the starting_magnetization? It seems total magnetization, 
> absolute magnetization, polarization etc. are in Bohr magneton unit 
> but not the starting_magnetization. I cannot choose 
> starting_magnetization = 3 since it should be between -1 to +1. So, 
> how to choose starting_magnetization in this case?
>
> Sincerely,
> Abdul Muhaymin,
> Graduate(MS) Student, Institute of Materials Science and Nanotechnology,
> Bilkent University, Ankara
>