Messages

4996

Scripts, Tutorials and Applications / Re: transmission spectrum

« on: July 30, 2009, 10:00 »

If you have the log file from the calculation, it should contain the transmission spectrum data as two columns (energy E, T(E)) somewhere towards the end.

If you don't have the log file any more (or, if you encounter the problem that there are too few decimals in the spectrum in the log file...), you can extract it from the VNL file using the method described here: http://quantumwise.com/forum/index.php?topic=3.0. Then, you can plot it in any software you prefer.

You may want to modify the script to load the VNL file internally (instead of prompting for it, interactively), and the same for the sample name (you know this from the Result Browser). Then, when you run the script, just pipe the output to a file, and you will have the spectrum in this file, ready for plotting in e.g. GNUPlot.

Hope you are able to figure this out, otherwise let us know if we need to post more detailed instructions.

4997

General Questions and Answers / Re: From crystalmaker to VNL

« on: July 30, 2009, 07:32 »

Can you attach the file, then we can have a look? It should be quite easy to write a small script to convert it (provided it's a text file).

4998

General Questions and Answers / Re: SymmetryReducedTransmission.py

« on: July 29, 2009, 13:17 »

These files are provided in the file "transmission_scripts.zip" on the Tutorials page, just next to where to download the PDF file from (right-most column of the table). Please use the Monkhorst-Pack script from the ZIP file, as it's substantially updated compared to the older one in the Examples.

4999

General Questions and Answers / Re: Python Files in VNL

« on: July 29, 2009, 12:59 »

The concepts of surface layers are a bit different from the respective perspectives of the Python configuration and the Atomic Manipulator, so this is nothing to worry too much about.

If the geometry becomes distorted, it might however be an indication that it's not entirely correctly set up. In particular, pay special attention to the "alignment atoms". One way to check is to instead drop the Python file on the Nanosope. If it also looks weird there, then you can be quite sure there is some error in the setup.

5000

General Questions and Answers / Re: Python Files in VNL

« on: July 29, 2009, 10:00 »

If you have the geometry already defined in the Python script, there isn't really any need to save it as a VNL file, you can just drag the geometry directly to the NanoLanguage Scripter, as ZH mentions. Having the geometry in a VNL file can however be convenient if you want to reuse it several times, or import it into another script, without having to drag along the original Python code.

Thus, if you're anyway playing with the Python code, you can just add a few lines of code to generate a VNL file:

Code

f = VNLFile("/home/user/my_geometry.vnl")
f.addToSample(my_configuration,"My nice geometry")

(obviously you have to change the filename and variable name, and the label, to match your specific script and conditions).

5001

Scripts, Tutorials and Applications / Re: calculting DOS and LDOS

« on: July 28, 2009, 09:11 »

Yes, you can certainly use the checkpoint file for this purpose!

If you are writing scripts by hand, have a look at the manual entry for restoreSelfConsistentCalculation().

In VNL, drop the original geometry on the NanoLanguage Scripter (to make it understand which system you are using). Then go to the "Self-Consistent Calculation" tab and tick "Restore calculation from checkpoint file", enter the NC filename, and UNCHECK "Only use initial density".

Then, set up the analysis as you would otherwise.

5002

Scripts, Tutorials and Applications / Re: transmission spectrum

« on: July 28, 2009, 09:03 »

To zoom out in the Result Browser plots, right-click the plot and choose "Zoom Out" from the context menu. The context menu also allows you to zoom in, but it's easier to just make a box with the mouse to zoom in.

The best strategy to get the plot you want is to zoom out very far (using the context menu), and then zoom in on the specific area of interest.

5003

General Questions and Answers / Re: DOS for bulk

« on: July 27, 2009, 14:47 »

For jiangning198511:

I have re-engineered the 1D DOS module to handle spin. Try the attached script instead of the original "dos1d.py". You use it exactly like the original; the script will internally determine if the calculation was spin-polarized or not. The two spin components will be plotted red and blue, but it's simple to modifiy the script to change this, or any other plotting preferences.

Note that Nordland's method, which is really simple and straightforward, works quite well too, although you need to fiddle with the script each time you want to change the spin component, of course. But it's still easier, perhaps, if you want separate plots of the up and down spins.

5004

General Questions and Answers / Re: how can I get the results from the .nc file?

« on: July 27, 2009, 14:38 »

That really depends. Do you have 1 nc file or many (i.e. one for each bias?)

I suggest you post the script you run, so that we can see how it was handling the NC and VNL files. You can mask/remove the geometry if you wish, it's irrelevant.

5005

Links to Resources and Publications / Re: Papers from 2009

« on: July 27, 2009, 11:19 »

5 new publications using ATK have been included in the reference list. The topics are:

• Transport properties of C20 fullerene molecules (An et al., Ludong Univ.)
• Negative differential resistance in fused thiophene trimer (Sen and Chakrabarti, Univ. of Calcutta)
• Polyacene and biphenyl systems with different terminal and intramolecular connections (Liu et al., Nanjing Univ.)
• Electronic structure and quantum transport of Pt-SrTiO3-Pt heterostructures, with special focus on effects of dopants (Wang et al., Tohoku/Tokyo Univ.)
• Switching characteristics of an optical molecular switch based on the 15,16-dinitrile dihydropyrene/cyclophanediene (DDP/CPD) molecule with two single-walled carbon nanotube (SWCNT) electrodes (Zhao et al., Universities of Jinan, Shandong, and Jining)

5006

Scripts, Tutorials and Applications / Re: How to Import Siesta into ATK

« on: July 17, 2009, 10:53 »

ATK and SIESTA are two completely different codes, although they are both based on the same method for carrying out the calculations. Therefore there is no immediate way to interface the two programs.

You could import and export geometric structures and use them in both programs, but for this you would have to either use a common file format (like xyz) or write a small script that converts a NanoLanguage structure to SIESTA input and vice versa.

5007

General Questions and Answers / Re: Voltage symmetry

« on: July 17, 2009, 09:47 »

Indeed, it would be sufficient to specify only the "bias" (the difference), and not the individual electrode voltages. But if multiple electrodes would be involved, one would need a common reference, or specify the voltages individually. The design is made to allow for this extension in the future

5008

General Questions and Answers / Re: A simple gui for mpi code

« on: July 14, 2009, 22:22 »

GNUPlot hardly seems the best program to use for this. Perhaps "rasmol" or something similar might be better.

As Nordland notes, beyond that answers on this Forum will be limited to issues related to ATK.

5009

Scripts, Tutorials and Applications / Re: Exporting electron density etc to XYZ file for plotting in other software

« on: July 14, 2009, 22:19 »

There are a couple of ways to approach this.

You can handle the many files within the script itself, using the "glob" module. This is a bit more complex, however, so I'd recommend using simple bash looping.

To do this, first we need to modify the code slightly, so that we can pass the name of the NC file on the command line:

Code

import sys
scf=restoreSelfConsistenCalculation(sys.argv[2])

Then, we execute the script as

for i in *.nc ; do atk exportED.py $i > electronDensity$i.dat ; done

You need to add paths etc. This has the slight disadvantage that the files will be called "electronDensity1.nc.dat". If you can live with that, all ok. If you prefer the filename "electronDensity1.dat", we just need to change it a bit:

for i in *.nc ; do j=`echo $i | sed 's/.nc//g'` ; atk exportED.py $i > $j.dat ; done

5010

General Questions and Answers / Re: About structural optimization of atomic cluster

« on: July 13, 2009, 12:21 »

The "standard" tricks for better convergence would be to lower the diagonal mixing parameter, perhaps to 0.05 or lower. Sometimes it helps to have more history steps (12-20), sometimes fewer. I would probably increase the mesh cut-off, provided you have enough memory for it.

Hope these simple suggestions help.