A question about LDOS parameters

[BandTheory]

Hello,

I am trying to calculate LDOS for a two probe system and I had a question about the parameters.  For 'quantum numbers'  if I do not expect there to be a difference in the spin up/down DOS is there some way that I can just say something like spin.both or should I just calculate it twice for spin up and down?  Also what is the best way to pick the k point for a nanotube?

Thanks very much.

[zh]

Yes, if you want to obtain the total LDOS (here "total" means the sum of up-spin and down-spin) of a spin-polarized system, you should do it twice, one for up-spin and the other one for the down-spin. And then, sum up them.

For a non-spin-polarized system, only the k point should be specified for the quantum number. The more details of quantum number are already explained by the manual: http://quantumwise.com/documents/manuals/ATK-2008.10/ref.calculatelocaldensityofstates.html.

Since CNT is an one-dimensional system, only the Gamma point (0.0, 0.0) could be more meaningful for the k point in the quantum number of LDOS.

[BandTheory]

Thanks for your previously reply.  I missed the part in the manual where you have the option to not include the spin as a parameter.  One more question and perhaps this is a silly one: but why do you supply only one energy for Local Density of States instead of a list of energies like Density of States ?

Thanks again.

[Anders Blom]

For consistency I agree it could have been a list instead of an energy... The thinking is basically that you are typically interested in the LDOS for a particular energy. Also, it's a large quantity (in memory and to save in the file), so usually it's done one by one. If needed, one can always loop it in Python

[BandTheory]

I see.  I will just loop it in python.  Thanks again

[zh]

I think that you should understand the physics meaning of LDOS and why this quantity is introduced.  The LDOS is indeed the square of wave function for a given eigenstate (the quantum number includes n, k, [tex]\sigma[/tex], where n for the index of eigenvalue, k for the index or coordinate of k-point,  and [tex]\sigma[/tex] for the spin).  Therefore, it is no necessary to give a list of energies.  If you are interested in many eigenstates, their LDOS can be calculated one by one when the corresponding eigenvalues are given.

[ipsecog]

It has been argued sometimes that the LDOS for a particular position, as a function of the energy, is useful. Hence, I think, the request for the list of energies.

However, I guess what is more relevant is the DOS projected on a particular atom, and this functionality is now included in ATK 10.8.