[Wien] Non-Collinear Magnetism (NCM) and Wien2k

Robert Laskowski rolask at theochem.tuwien.ac.at
Fri Aug 24 16:27:17 CEST 2012


On Tuesday 21 August 2012 22:37:55 Jerry Bettis wrote:
> All,
> Questions 1) When setting up a non-collinear calculation there is no need
> to distinguish up and down spin magnetic ions. Therefore, like magnetic
> ions share identical election configurations in case.inst. In the
> file case.inncm the relative arrangement of a pair spins are designated. Is
> this statement true?

in non collinear case the wavefunctions are not pure spinors, furthemore 
up and down part of the spinor vary in space. In order to define a spinor you 
need to define the coordinate frame which tells you what is up and down, this 
is the quantization axis. case.inncm file allows to specify the initial 
magnetic structure just by specifying the Euler angles which define the 
direction of the quantization axis for each atomic sphere. It is then assumed 
that the initial spin density matrix is diagonal for each atomic sphere. 
During SCF the non diagonal elements of the spin density matrix can evolve, 
and the direction of each quantization axis can be also altered, depending on 
the optimization switch in inncm. If you choose to optimize the direction of 
the magnetic moments the axis will always points to the direction of the 
average magnetic moment, so can rotate from SCF cycle to cycle. Or if someone 
does not want to optimize the directions of the magnetic moments, they stay 
constant but instead constrain field is calculated for each sphere, and 
included into the Hamiltonian.   

> Question 2) The program LSTART requires the file case.inst. The program
> DSTART requires the files case.in0, case.in1, and case.in2. After obtaining
> all the inputs files required for a successful non-collinear calculation
> can I use these "Start-up files" as a starting point for calculations which
> differ only by the relative arrangement of magnetic ions in the case.inncm
> file?

LSTART calculates the atomic density (up and down parts), or you may call it 
diagonal elements of the spin density matrix, using information about atomic 
configuration from inst file. It is exactly like in wien2k, but there the 
quantization axis points always to z for each sphere.

> Of course I would re-run ncmsymmtry for calculations with different
> case.inncm files. Ultimately, I am investigating the energy differences
> between magnetic structures with identical chemical unit cells. I would
> like to know if this type of logic is reasonable.

In this case you may do it that way, or select the lowest common symmetry 
settings and use it for all calculations.



Dr Robert Laskowski
Vienna University of Technology, Institute of Materials Chemistry, 
Getreidemarkt 9/165-TC, A-1060 Vienna, Austria
tel. +43 1 58801 165303               Fax  +43 1 58801 15698

More information about the Wien mailing list