[QE-users] pw.x - understanding magnetization

[Ing. Martin Matas]

Dear Abdul,

starting_magnetization is the initial ratio of the number of polarized electrons to the total number of valence electrons for the respective atomic species. Therefore, it ranges from –1.0 (all spins down) through 0.0 (paired spins) to +1.0 (all spins up). It is used to construct the first guess for the wave function, but the magnetic configuration will likely change during the self-consistent optimization.
Your particular case apparently converged to one given result regardless of the initial magnitude of the homogeneously distributed magnetic moment, but starting_magnetization is generally able to affect the resulting wave function; e.g., you can often force the system to end up in a ferromagnetic state, in a non-magnetic state, or in one of several different anti-ferromagnetic states with a varied distribution of spin-up and spin-down atoms in the lattice.
In addition to starting_magnetization, there is the tot_magnetization keyword, defining the total magnetization of the simulation cell, fixed during the entire calculation.

The reason why the sum of magnetic moments per site is smaller than the total magnetization is, indeed, the fact that the spheres do not fill the entire simulation cell.

Hope that helps.

Martin Matas
University of West Bohemia
Czech Republic


 Čtvrtek, Únor 15, 2024 10:19 CET, Abdul Muhaymin via users <users at lists.quantum-espresso.org> napsal:

> Hello QE users,
> 
> I am trying to understand how to appropriately set the starting_magnetization value and how to interpret the output related to magnetization in the pw.x. I tested for a simple system (BCC Fe) with various starting_magnetization and observed that it doesn't matter. The total energy, total magnetization, and absolute magnetization values remain the same. So, does this mean any starting_magnetization value can be used?
> 
> Secondly, when I observe the "Magnetic moment per site (integrated on atomic sphere of radius R)" values, and sum up all the magnetic moments, it doesn't match with the total magnetization or absolute magnetization which is reported at the end (in Bohr mag/cell). The sum of magnetic moment per site is always less than the sum of total magnetization. What could be the reason for this? Is this because magnetic moment per site is integrated on atomic sphere of radius R where R < 1?
> 
> Sincerely,
> Abdul Muhaymin,
> Graduate(MS) Student, Institute of Materials Science and Nanotechnology,
> Bilkent University, Ankara.