some questions about the newest version

[John]

Dear friends,
     When I did one test about the transport properties of Li-H2-Li system by the both methods ATK-DFT and Device Extended Huckel using the newest version ATK11.2 released on 3 March, some errors occurred:
     1.   It seems that the method Device Extend Huckel can not do the optimization of one two-probe system.
     2.   Some analysises can not be obtained by the Device Extend Huckel method, such as the ElectronDensity, Forces, LocalDeviceDensityOfStates, Effective potential, and stress (also for ATK-DFT)
     3.   For one spin non-polarized transport calculation, some parameters are only alternative, up or down, such as for the calculation effective potential, Eigenstate, TransmissionEigenstate, TransmissionPathways,
     4.   About the TransmissionPathways researches,  for one spin non-polarized transport calculation, Four situations should be inverstigated, such as Left+Up, Left+Down, Right+Up, and Right+Down.
     5.   For the ATK-DFT calculation, I have a trouble to decide the Poisson Slover parameters, for 1D Graphene nanoribbon transport calculation, which one is better for such system, Multi-grid, FFT2D, or FFT?
     Thanks.

[Anders Blom]

To be fair, none of these are really errors

1. In future editions of ATK we will add this capability via pair potentials, but as such the Huckel method can not compute forces.
2. This is to be expected. The EHT method does not support computation of as many quantities as DFT does.
3. You need to specify all quantum numbers of the object you want to compute. Even if the calculation is not spin-polarized, spin exists. However, I'm not sure I understand exactly what the issue is...?
4. Same as 3...
5. FFT2D should be you choice in basically all situations for two-probes

[John]

Thank you, Anders Blom.
     For these properties ,such as the effective potential, Eigenstate, TransmissionEigenstate, and TransmissionPathways of one spin non-polarized transport calculation, I think it could be better if the sum results (up + down)can be also obtained. Like that there are three options when calculating the ElectronDensity, Sum, Up, or Down.  For one Spin polarized, the Up and Down results can be analyzed, respectively. While for one spin non-polarized calculation, Whether is it better to only analyze the sum results (Up+down)?
    Another question, It seems unable to automatically display the bond length when the mouse points at it.
    

[Anders Blom]

Good feature request for the bond length, we'll definitely make a note of it!

For an unpolarized calculation, all results for spin up and down are identically the same, so summing just means multiplying by 2.

[Anders Blom]

Btw, there is a way already to see the bond length, although not quite as easily as just holding the mouse over it.

If you select the two atoms connected by the bond of interest in the Viewer, then go to the menu and choose Panels>Z-matrix, you will see the bond length.

[John]

Thank you, Anders Blom.
    Whether are there any bugs for the newest version ATK11.2 for Windows 32-bit. The spin parameters of the Transmission Pathways calculation have only two choices, "Up or Down", like Figure1, but more parameter "Total" can be found in the the VNL tutorial(Page 24). Maybe,other properties calculations , should also have the "Total" parameter, like Eigenstate and TransmissionEigenstate calculations.
     Thank you.  

[zh]

For a spin-polarizied system, the availability of only "spin up" and "spin down" for the transmission spectrum calculation is reasonable.  By analyzing the results related separately to the spin up and the spin down may be more meaningful than the analysis of the total spectrum.  The total spectrum can be given by the sum of the calculated transmission spectra of "spin up" and "spin down".

[Anders Blom]

A few small but rather important changes were made between the beta-version of 11.2 and the final release, regarding how to treat spin for non-polarized calculations. "Total" spin for these objects is not available in 11.2.0, but it was in the beta-version, from where the screenshot in the tutorial unfortunately was taken. Sorry for the confusion!