Page 1 of 1

Spin Polarised Calculations

Posted: Thu Jun 29, 2023 2:10 am
by satvik_kashyap
Hello,

To perform Spin Polarised Calculations on a Non-Magnetic System, which among the two tags- MAGMOM and NUPDOWN should be set to zero? Also, do I set the values on the individual atoms as zero or set them to zero collectively on all atoms? E.g. If there are 6 atoms of A and 6 atoms of B, do I set 6*0 6*0 or 12*0?

How should I set the two tags for a magnetic system? What about a system whose magnetic property is unknown?

What is the standard practice regarding these two tags while performing spin-polarised calculations? When is one used and not the other?

From my relatively little experience, setting NUPDOWN to a certain number forces the calculation to yield that magnetization; however, when MAGMOM is set, it allows for the magnetisation to relax and converge on its own, and it does not always yield the number set initially. Is this of any relevance?

Thank You

Re: Spin Polarised Calculations

Posted: Thu Jun 29, 2023 8:11 am
by fabien_tran1
To perform Spin Polarised Calculations on a Non-Magnetic System, which among the two tags- MAGMOM and NUPDOWN should be set to zero?
MAGMOM is used to set the atomic magnetic moments of the starting density, however the magnetic moments are allowed to change during the SCF iterations. NUPDOWN is used to fix the difference between the numbers of spin-up and spin-down electrons (i.e., the total magnetic moment of the cell), but NUPDOWN=0 does not necessarily mean that the magnetization will be zero at every point of space in the unit cell.

Why would you like to do a spin-polarised calculations on a non-magnetic system? If the system is non-magnetic, then ISPIN=1 is the way to have no magnetization everywhere in the unit cell.
Also, do I set the values on the individual atoms as zero or set them to zero collectively on all atoms? E.g. If there are 6 atoms of A and 6 atoms of B, do I set 6*0 6*0 or 12*0?
"6*0 6*0" and "12*0" are equivalent. You can test it yourself.
How should I set the two tags for a magnetic system?
It depends what you want to do.
What about a system whose magnetic property is unknown?
You have to guess what magnetic configuration(s) are reasonable from the physical point of view, set MAGMOM accordingly and then compare the total energies of the various magnetic configurations. The one with the lowest (most negative) total energy could possibly be considered as the ground-state structure.
What is the standard practice regarding these two tags while performing spin-polarised calculations? When is one used and not the other?
MAGMOM is quite commonly used. NUPDOWN probably less often.
From my relatively little experience, setting NUPDOWN to a certain number forces the calculation to yield that magnetization; however, when MAGMOM is set, it allows for the magnetisation to relax and converge on its own, and it does not always yield the number set initially. Is this of any relevance?
See my answers above. In general, reading relevant literature helps to understand the concepts and details.