Show Posts

This section allows you to view all posts made by this member. Note that you can only see posts made in areas you currently have access to.


Messages - Anders Blom

Pages: 1 2 [3] 4 5 ... 361
31
Zero bias really just means no external bias. If the two electrodes are different, they will have different Fermi level and thus an internal bias still exists. So there is no special treatment of this case, the calculations proceed as for any device, except that the crucial in which the two Fermi levels are brought into alignment (no bias!) results in a shift of the bands in one electrode, and a current will flow from occupied to unoccupied bands (but backwards, since current is opposite electron flow). That is why left-right and right-left currents might differ in this case.

32
I suppose it will print it when it's finished. Note that the IVCharacteristics study object is designed to run a lot of different calculations distributed across many (very many!) nodes and cores, so it will wait to collect all results until the end, as some tasks might finish before others.

33
If they not made of the same material, they should not be equal... Each electrode should have the length needed to make it periodic.

34
General Questions and Answers / Re: Device design
« on: January 26, 2024, 20:42 »
It's really not possible to answer better than the general info unless you share what the system looks like, in detail. You can email the Python file to me if you don't want to share it publicly.

35
Correct, but HSE with LCAO is still reasonable for NEGF, we have tested it with 1000+ atoms. Needs a lot of memory though, that might be an even larger benefit of ADMM which, for now, we cannot obtain in NEGF,

36
General Questions and Answers / Re: Device design
« on: January 22, 2024, 22:17 »
It's kind of hard to do by yourself, it's really best to let the Builder do it. In theory the method is simple though, you need to identify a part of the left and right sides that are periodic, and then extend these in both directions. You can practice or test with a much simpler configuration, or see https://docs.quantumatk.com/tutorials/atk_transport_calculations/atk_transport_calculations.html for some guidance. But getting the exact script right by hand requires a bit of experience, for sure.

37
General Questions and Answers / Re: Phonon spectrum error
« on: January 22, 2024, 21:58 »
Looking at the structure, I would recommend centering the atoms in the Z direction, so that the 2D sheet is in the middle of the cell, instead of straddling the periodic boundary, since the structure is actually not supposed to be periodic in the Z direction (although it always will be in practice, but with a large vacuum gap). Perhaps the algorithm doesn't know what to do when there are bonds and symmetries extended across a direction in which you have 1 repetitions for the dynamical matrix, as you correctly should have for this calculation.

So, in the Builder, just shift all atoms by say fractional 0.3 in the z direction, then wrap and center, and try that geometry instead.

Also, remember to check if the structure is optimized first. It seems somewhat unlikely that the lattice constant will be EXACTLY 4 Angstrom...

38
General Questions and Answers / Re: IV-curve question
« on: January 19, 2024, 02:01 »
It will do it inside if needed, but if you leave the update line it will not do it twice.

39
The toolbar can be detached and made floating. Probably it's on a second screen or hiding behind another window on your desktop, but you can always restore it via the menu View>Restore Builder layout.

Edit: Oh, you say that doesn't work, that is indeed puzzling...

Try to open $HOME/.quantumatk/main.ini and remove the line DockState in the Builder section (make a backup of this file first).

40
A few things to check and comment on here.

First, do you see the same thing with LDA or GGA, i.e. is it specifically a problem of MGGA? If it is, it makes it a bit more interesting to investigate.

However, apart from perhaps a somewhat instructive tutorial example, the calculation doesn't really have any value. The transmission spectrum of any bulk material should really just be counting the states at each energy; in 1D this is trivial and gives you an integer staircase, while in 2D and 3D it's a bit more complex due to k-point sampling, but conceptually still the same. To make an interesting calculation, one should create an extended device with some scattering centers. If also in this case the transmission is non-zero in the band gap, I agree there is a big problem, but we have never seen that.

Now, I am not trying to minimize the potential issue, we should probably look into it, but it's also not a serious problem if this is isolated to the bulk transmission, as that is generally not something you need to compute; we support it mainly for legacy reasons. Or, to some extent for educational purposes, so again, it should of course calculate it correctly. Thus, if you can check it with LDA also, it will help us a bit.

41
General Questions and Answers / Re: Phonon spectrum error
« on: January 19, 2024, 01:36 »
Given that this is a rather large calculation, do you mind sharing the HDF5 file, it might save time. Assuming, that is, that the DynamicalMatrix object was actually saved in the file, which it might not be if the error occurred while saving it.

Also, which version used?

42
General Questions and Answers / Re: Help in python code
« on: January 19, 2024, 00:57 »
Sorry for the delay.

In general, I would avoid using DensityOfStates, as it's by now just a legacy object. PDOS really has all the same functionality, but is better supported and a LOT faster. And, you can use broadening in PDOS, but it's a bit hidden. Just set set spectrum_method=GaussianBroadening(broadening=0.03*eV).

So, for both DOS1 and DOS2, using ProjectOnShellsBySites in PDOS should give you the data you need, and then you can filter it down for plotting.

43
General Questions and Answers / Re: Simulation of COHP and pCOHP
« on: January 19, 2024, 00:21 »
You can't do that via the GUI for now, but in the script  you can create more complex projections. In your case, open the Python script in an editor and change the value of the "projection" keyword in ProjectedDensityOfStates to

ProjectionGenerator(atoms=ElementsProjection, l_quantum_numbers=True, m_quantum_numbers=True, spin=UpDownProjection)

44
It's not necessarily out of memory. Can you check the other log files for errors?
In particular the last one, mtp_training_defectconfiguration_0_bundle_2.log

45
General Questions and Answers / Re: Device design
« on: January 18, 2024, 23:39 »
The fundamental point to realize is that a device has the left and right leads repeated semi-infinitely to the left and right. So if you are ok with that in your design, it's not a problem to build it.

Pages: 1 2 [3] 4 5 ... 361