what do the three SCF calculations mean?

[Roc]

Dear everyone,

    I have not the complete understanding about the two-probe calculations. In the output file(Voltage=0), there is three SCF calculations, as following,
the first one is about the Electrodes Calculation, second is Equivalent Bulk Calculation (Initial Density for TwoProbe), last one is TwoProbe Calculation.

anyone could give an explanation,respectively?

thanks a billion!

[Nordland]

1) The first SCF calculation is performed for on the electrode. It is a normal DFT calculation and it is used to determine the hamiltonian for the electrode as it was a normal bulk system. This hamiltonian is used for calculating the surface greens function, so we can calculate self-energies. The self-energies are needed in order to describe the interaction between the (semi-)infinite electrodes and finite scattering region.

2) The Equivalent Bulk Calculation is a convergence trick and I will get back to it in 5 lines time

3) In the last SCF calculation we will try to obtain the self-consistent electron density for the entire system with electrons flowing in
from the infinite electrodes through the scattering region. This will determine the "ground"-state for this system, and from this we will be able to derive all the transport properties and other physical quantities. However finding this ground-state is often quite hard, and therefore it is needed to have a good guess for the starting density, which brings us back to the Equivalent Bulk Calculation.

2b) If the user has requested it, ATK will automattcally start a Equivalent Bulk Calculation that will setup the scattering region with periodic boundary conditions, and will perform a normal DFT calculation on this in order to get a self-consistent density for this system. This density will then be used as the starting density of the TwoProbe calculation.

[Roc]

thank you very much, Norland.

the second SCF that convergence trick is quite interesting.

your explanation is quite detailed. great job!!!! 

Come on!

[yangzw1985]

I also interested in the SCF calculation. As far as I know, calculations should be carried out in two steps. first, the SCF calculation; second, the physical properties caoculation.

For example, if we want calculate the transmission spectra of a li-h2-li model, we can first perform the SCF calculation, and got the corresponding nc file, and then calculate the TRANS SPECTRA.

Now, if we have setted all the parameters in the method menu, and performed the SCF calculation, and got the nc file, How can we know is the self-consistent calculation convergenced or not?  Should we determined it by the nc file or the log out results, or /and any other parameters?

thanks!

[zh]

Now, if we have setted all the parameters in the method menu, and performed the SCF calculation, and got the nc file, How can we know is the self-consistent calculation convergenced or not?  Should we determined it by the nc file or the log out results, or /and any other parameters?

Obviously, the convergence of self-consistent should be controlled by a parameter which is actually specified in the input file. See the explanation in the manual for more details.
http://quantumwise.com/documents/manuals/ATK-2008.10/ref.iterationcontrolparameters.html
Also, the information about the convergence self-consistent calculation is printed out the log file.