ATK 2010.8

[jdgayles16]

I am having trouble with the new ATK I cant send anything to the job manager, or viewer, the window opens and it is just blank. also the scripter looks different than in the tutorials, I have attached a file of what my scripter looks like. Im runing vista on 3gb of memoey, and i unistalled the old version of vnl.

Jacob

[Anders Blom]

Yes, it's a bit different than in the tutorials, still; we're working on it. However, the one about transport in graphene is updated, so you can check that for some basics.

Take a step back and think what you really want to do. Probably you have some geometry that you have built; drop that on the scripter. Then, you double-click "New calculator" to define the Kohn-Sham method, then add some analysis, e.g. TransmissionSpectrum. Right now it looks like you just randomly clicked a bit

[jdgayles16]

Nice!! I like it  .


New question,
How do i set up a parallel calculation for the new atk? I couldn't find documentation

[jdgayles16]

I guess the parallel version is not out yet, can we know when?

[Anders Blom]

ATK 10.8.b1 is fully parallelized. You run it in parallel just as 2008.10, by making the script in VNL (or by hand), and then running it under "mpiexec". No changes here, thus, except now you can actually also run it in parallel (MPI) on Windows.

[jdgayles16]

Nice!! thanks I think it working ok now!! 

[jdgayles16]

Question,
Could someone expand on the DensityMAtrix constraint and the EquivalentBulk parameters? Also I am a little confused on what it means by electrode constraint length, should I choose the length of my electrode or is it something else entirely?

Thanks
Jacob

[Anders Blom]

The short story is that you should use the default Off constraint in ATK 10.8. We have fixed it so that it gives a correct voltage drop (more details will be available soon, in the "upgrade guide"). The DM constraint is still available, but only intended for tricky situations when convergence is hard to obtain.

Thus the only thing you really need to be concerned about for the EquivalentBulk class is if you want to run this part of the calculation with e.g. a lower tolerance you can modify locally the IterationControlParameters. Otherwise just go with defaults flat out!