Messages

16

General Questions and Answers / Re: Analyzing spin-resolved currents across different versions of QuantumATK

« on: September 3, 2025, 03:10 »

Frank, I missed this while on vacation. 2022 files should be fully compatible. It should be very easy to solve with a few lines of Python code, but would need a few more details, like if this is an IVCharacteristics object, or noncollinear calculations, etc. We can handle it over emails.

17

General Questions and Answers / Re: Kerker Preconditioner for LCAO DFT calculations in X-2025.06

« on: September 3, 2025, 03:07 »

Don't use the Kerker preconditioner, I have never seen it give any benefits.

18

General Questions and Answers / Re: Non collinear Spin Relaxation

« on: September 3, 2025, 03:05 »

Pls share input and output, else there is no way to tell if you made some simple mistake. But it is an old tutorial, so perhaps something needs updating, we can check.

19

General Questions and Answers / Re: DFT Phonon Transmission

« on: September 3, 2025, 03:03 »

If you go to 1x1 then at least turn on the Wigner-Seitz scheme, else you are making a Gamma point phonon approximation. I would probably check if some of the machine-learned forcefields like MACE or M3GNet give at least decent phonon dispersions, and if so use them, for a much, much, much shorter calculation time, which hopefully is at least equally qualitative as a very approximate DFT simulation.

20

General Questions and Answers / Re: Issue to apply uniaxial stress on built heterostructure properly

« on: September 3, 2025, 03:00 »

It's somewhat straightforward for a bulk crystal, but how the strain should be applied in a heterostructure is much more complex. I don't think it is enough to just change the axes A B or C, you need a more careful analysis, not least because of the lattice mismatch in the interface.

21

General Questions and Answers / Re: RAMAN Spectrum on monolayer MoS2

« on: September 3, 2025, 02:57 »

Sorry, but it's an impossible question to answer short of myself running all calculations to check that every parameter is using correct sensible values (enough k-points, maybe needs more?), if the structure is correctly optimized (why not optimize the cell size?), etc, etc.

22

General Questions and Answers / Re: Use cases of ChargedPointDefect vs. ChargedPointDefectConfiguration

« on: September 3, 2025, 02:55 »

Yes, that is the main difference. A new tutorial on this topic is in the works.

23

General Questions and Answers / Re: HartreeDifferencePotential caculation

« on: September 3, 2025, 02:54 »

Maybe check the electron temperature,  the default was 300 K when the tutorial was written, now it's 1000 K, maybe that causes convergence problems. We will also check ourselves if the calculation settings are still correct, the tutorial is 9 years old and some adjustments may be needed for newer versions of the code.

24

General Questions and Answers / Re: Is it possibility do collinear spin transport with Fixed spin Ni (electrode)?

« on: September 3, 2025, 02:49 »

As mentioned before, there is no difference between up and down spin unless you have a magnetic axis in the system, which defines up/down relative some reference direction. And in the collinear case there is no such direction, and since Gold is non-magnetic there is not even a parallel or antiparallel spin setup, so your up and down for Ni will remain degenerate and cannot be distinguished. In fact, I don't think you would see a different even in a noncollinear simulation since the magnetic field is not actually part of the simulation, and I honestly don't see physically how it would change anything in the experiment either, if the field is turned off before the measurement?

25

General Questions and Answers / Re: Can I use an external 802.51 Wi-Fi adapter with the same network for Server?

« on: September 3, 2025, 02:36 »

Sorry, I don't understand the question, please try to rephrase what you actually need.

The basics are simple: you need a network connection from the machine running the software to the machine hosting the license. The license can be tied to the Ethernet or Wifi adapter, no problem with either, and in fact does not have to match the network you connect over. The MAC address is only used to ensure the license is valid for the specific machine.

26

Links to Resources and Publications / Re: Paper from 2024 ("Device from bulk" inactive in GUI)

« on: August 19, 2025, 00:16 »

For some crystals it is simply not possible to cleave them in a way that creates an AB plane perpendicular to C. For trigonal cases, it is possible and quite obvious in the hexagonal representation, but mainly only for 0001 surfaces. In short, it will be difficult to use a rhombohedral crystal as electrode...

27

General Questions and Answers / Re: License limit

« on: August 13, 2025, 23:33 »

You can add the environment variable FLEXLM_DIAGNOSTICS=3 to a run, this will show licenses being checking out and in. That way you can track the license usage for some simple calculations, and you will see the logic.

28

General Questions and Answers / Re: License limit

« on: August 13, 2025, 23:31 »

There is no need for an estimate, you can tell exactly in a simple way. The kind of calculation you mention should use exactly 1 license of the qatk_quantum type, because only 1 calculation is running, and if you run it mpiexec -n it will additionally check out n-1 qatk_dp licenses. That's it.

29

Links to Resources and Publications / Re: Paper from 2024 ("Device from bulk" inactive in GUI)

« on: August 13, 2025, 16:48 »

The trigonal crystal cannot be cleaved to make an orthogonal cell, except in certain directions. The electrodes must have one direction (C) that is perpendicular to the two others (AB).

30

General Questions and Answers / Re: License limit

« on: August 13, 2025, 16:46 »

The licenses are more related to what's in the Python script than the settings, at least in terms of the main qatk_quantum licenses. The -n keyword for MPI basically only determines how many DP licenses will be used for parallelization, but if you have a complex task in the script like a ChargedPointDefect calculation, which runs several DFT calculations, these can be run simultaneously, or sequentially, depending on how many nodes you allocate, but also how many processes you assign per task.

So in a sense it's very simple, but the complexity comes in the interplay between what the script contains and the hardware you allocate, and cannot be stated generically, except for the fact that every DFT calculation needs one qatk_quantum license (similar for forcefield etc) and can be parallelized by qatk_dp licenses.