[gpaw-users] Mixer in Transport Calculator

[Sebastian Caicedo Davila]

Hello GPAW users,

I have found a small problem with GPAW’s transport calculator. I’m working with amorphous material sandwiched between two gold electrodes. The system itself is difficult to converge, but setting the mixer to mixer=Mixer(0.01, 5, 100) helps the system to converge neatly (I tested with the GPAW calculator).

When I input this setting into the Transport calculator, it is not reflected in the calculation, and makes the sc convergence really slow. The guess steps won't converge (the only difference with the normal GPAW calculation is the Mixer, which in the output is different than the input). I first tried it with version 0.11.0 which is the one I usually use, but even one I move to the latest version, the problem persists. In the Lead calculation, you can see the configuration of the Mixer mixer=Mixer(0.1, 5, 100) is not the one in the input. The scattering region, doesn’t even come close to converge.

The problem can also be seen in a non self consistent transport calculation, where the system will never reach convergence.

If I make a non-sc calculation using separate calculators for the scattering region and the leads, where I can correctly set the mixer, the calculation goes well (and even really fast). I would really like to have a sc- converged calculation.

I don’t know if any of you have had the same problem, and it will be really helpful for me if it can get fixed.

Best,
Sebastian Caicedo-Davila
Research Assistant
Department of Electrical Engineering
University of Andes



Input:

t = Transport(xc='PBE',
     h=0.56,
              basis={'Au': 'dzp', 'C':'dzp'},
              kpts=(4,4,1),
              occupations=FermiDirac(0.1),
              mode='lcao',
              poissonsolver=PoissonSolver(nn=2, relax='GS', eps=1e-12),
              txt='aC_lcao.txt',
              mixer=Mixer(0.01, 5, 100),
              pl_atoms=[pl_atoms1, pl_atoms2],
              pl_cells=[pl_cell1, pl_cell2],
              pl_kpts=(4,4,4),
              edge_atoms=[[0, 0], [edge1, edge2]],
              mol_atoms=mol_atoms,
              guess_steps=10,
              fixed_boundary=False)

Output (Leads):

Mixer Type: Mixer
Linear Mixing Parameter: 0.1
Mixing with 5 Old Densities
Damping of Long Wave Oscillations: 100

Convergence Criteria:
    Total Energy Change: 0.0005 eV / electron
    Integral of Absolute Density Change: 0.0001 electrons
    Integral of Absolute Eigenstate Change: 4e-08 eV^2
Number of Atoms: 12
Number of Atomic Orbitals: 180
Number of Bands in Calculation: 108
Bands to Converge: Occupied States Only
Number of Valence Electrons: 132


-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listserv.fysik.dtu.dk/pipermail/gpaw-users/attachments/20160426/d24432aa/attachment.html>