Messages

16

General Questions and Answers / Re: Extended Huckel Overlap Element Integration

« on: November 18, 2014, 22:57 »

The overlap matrix element is calculated using a combination of radial FFT and spherical bessel functions. There exist no alternative methods in ATK, because the correct implementation is 'exact' with the default parameters.

There are two parameters that you can set for this (reciprocal_energy_cutoff and number_of_reciprocal_points)
http://www.quantumwise.com/documents/manuals/ATK-13.8/ReferenceManual/ref.numericalaccuracyparameters.html
However I strongly recommend not playing with these parameters, as they will not make the calculation faster or more precise.
We are even consider removing them since playing with them could only lead to wrong calculations.

17

General Questions and Answers / Re: Are LDA and SZP appropriate for two probe calculations??

« on: June 10, 2014, 10:48 »

Hi ATK-VNL team,

I HAVE 2 QUESTIONS:
(1)In two probe calculations, we generally use localized basis sets such as Single, or double, plus polarization. But according to  Stange et al. J. Chem. Phys. 128, 114714 (2008), the convergence of transport properties using localized could be problematic. I want to know the reliability of our calculated results???

Now I have not read the paper, but if it converges or not - depends on many other things. However if it converges as it does in the most cases, you should be okay

(2)Also we use LDA to calculate HOMO-LUMO gap in two probe calculations (for eg. Wang et al. Phys. Lett. A 374, 4876 (2010)). But according to T. Rangel el al. Phys. Rev. B 84, 045426, this approximation strongly underestimate the HOMO-LUMO gap. So again the use of LDA is reliable for our calculation?

Yes LDA underestimates the band gap, so the true question is how it affects your calculations. If you are modelling a transport through a cupper nanowire, it is properly not a problem. If you are trying to model the exact diode characteristic of molecular junction, you should be aware that breakdown points etc may be shifted in voltage.

18

General Questions and Answers / Re: How to initial decide the molecule position in a two probe device

« on: March 12, 2014, 09:58 »

It also comes down on what you want to model.

If you put a molecule near surface, it will affect both the molecule and the surface in a experiment and in a simulation. This will likely be a small effect, but it will be there.

Then becomes if you want to into account this effect or not. If you want to take this effect into account, you will have to follow the procedure suggested by zh.

If not, you will have to a scan calculating the total energy as a function of the seperation between the surface and the molecule, and find the minimum of this curve.

19

General Questions and Answers / Re: MD simulations-reg

« on: March 4, 2014, 08:03 »

I really dont understand your point.

MD is not an optimization as it does not move the configuration to a minimum

20

General Questions and Answers / Re: No of atoms and simulation-reg

« on: March 3, 2014, 23:19 »

The number of atoms matters for the calculation size, but so does other things, try to play around with the tutorial and see how the performance is of your machine.

21

General Questions and Answers / Re: Program at Backlog-reg

« on: March 3, 2014, 23:18 »

It depends lot on how you run it.

22

General Questions and Answers / Re: MD simulations-reg

« on: March 3, 2014, 08:55 »

MD is not really a method for optimizing a structure. It can be used as method for simulating an annealing process which could perhaps be a experimental local minimum.

23

Installation and License Questions / Re: Segmentation fault

« on: January 28, 2014, 14:06 »

We have managed to find a temporary work around for this,
so if you want to run ATK or VNL, you will have to set this following environment variable before executing:

LD_PRELOAD="libstdc++.so.6" atkpython my_script.py

or

LD_PRELOAD="libstdc++.so.6" vnl

24

General Questions and Answers / Re: III-V nanowire

« on: January 26, 2014, 18:48 »

It is very easy to build nanowire in VNL once you have the "building blocks" to build it with:
http://quantumwise.com/documents/tutorials/latest/NanowireDevice/NanowireDevice.pdf

However the biggest challenge is that, is that for In_0.53Ga_0.47As, it is not well known what the crystal looks like,
and how the elements are arranged.

25

General Questions and Answers / Re: Checking for Correct Optimized Structure

« on: January 25, 2014, 17:20 »

Dear Mr.Nordland,

I attached the file but the size was larger than the limit in the forum.So I couldn't attach.

Kindly give me an alternative mail ID to send the files sir.

Thank you.

Just attached the python scripts to this post.

26

General Questions and Answers / Re: Checking for Correct Optimized Structure

« on: January 24, 2014, 16:06 »

The system shouldn't not hang after a long calculation.

-However I can't approve your systems by looking at screenshot, I will the coordinates obtained by relaxing with both methods.

27

General Questions and Answers / Re: error message

« on: January 24, 2014, 16:04 »

You can also try to post the file here or send it to me - and I can see what I can do about it.

28

General Questions and Answers / Re: bandstucture

« on: January 19, 2014, 15:55 »

Also - the bandstructure of a molecule is a strange thing. Perhaps you should calculate the MolecularEnergySpectrum

29

General Questions and Answers / Re: error message

« on: January 18, 2014, 21:12 »

Virtual Nanolab has no support for the old VNL Sample format (or any other file named .vnl)

30

General Questions and Answers / Re: Electron localization function

« on: January 18, 2014, 21:11 »

It is first implemented in much later version.