Dear Colleagues,
I have a taks to compute local density of states at a single point
for a large homogeneous sample with narrow STS tip potential and in
magnetic field. I have tried several methods in Kwant to do this:
1) Naive diagonalization of H for system with no leads and direct
calculation of LDOS. It works, but not for large systems, so, I see
strong finite-size effects.
2) Sparse Lanczos method with several eigenvalues/functions extracted
near each point in the grid of energy values.
3) KPM method. I am not happy with the energy resolution I can get
with it.
4) Take a smaller main sample (with the size of the tip potential),
but avoid the finite-size effects by attaching a lot of leads pointing
in , e.g. 6 directions. Then, use the kwant ldos function to scan
LDOS over a grid of energies. Use magnetic_gauge to implement B field
in both sample and scattering region.
I see the last possibility as the quickest, but it gives me
something that seems wrong to me.
Does kwant ldos function work correctly if all the leads are insulating
for a given energy? Does it correctly catch the localized states in the
scattering region?
Thank you for any feedback,
Sergey
