Messages

5161

General Questions and Answers / Re: a convergence problem

« on: April 22, 2009, 10:15 »

Ok, the evidence for problematic behavior is already evident from the equivalent bulk part. While the total energy converges (barely, and in very many steps), the density matrix actually does not converge!

Thus, your first goal is to stabilize the equivalent bulk run, or, as an alternative strategy, skip it completly and go directly to the two-probe calculation. For this, change line 376 in the input file to

Code

    initial_density_type = InitialDensityType.NeutralAtom

However, the real reason for the instability of the equivalent bulk run is most likely because the electrodes are too short. I would make them twice at long. At the same time, you probably do not need such a long central region, so you can keep the total size of the system fixed by removing 2 periods in the central part.

For safety, I would also increase the size of the XY unit cell to 25-30 Angstrom instead of the current 18. This doesn't really influence the stability, but it might leads to some inaccuracies in the results in the end.

Further steps towards better convergence would include reducing the mixing parameter (probably not necessary) and increasing the temperature to perhaps 1000 Kelvin.

But try to just increase the electrodes and the XY cell, and just rerun with current parameters as a first attempt. If that works, there's no reason to worry about the other things!

5162

General Questions and Answers / Re: a convergence problem

« on: April 22, 2009, 09:37 »

Could you attach the complete output log, please? Which version of ATK was used?

Also, are you running this script in parallel or just on a single machine?

5163

General Questions and Answers / Re: St9_badalloc error

« on: April 20, 2009, 20:21 »

Yes, it means out of memory (searching on this Forum would give you that answer too )

480 C atoms can fit into 2 Gb, but it depends a bit on the parameters. For instance, if you have a DoubleZetaPolarized basis set, it naturally requires more memory than SingleZeta. Also, a bit hidden, is the number of k-points.

However, in your case there seems to be a simpler way to reduce the memory used per machine. You have 6 PCs yet you run 12 MPI processes. That means each PC will receive 2 MPI jobs, and thus only 1 Gb is available for each process. If you instead run "mpiexec -n 6" each MPI process will have 2 Gb.

5164

General Questions and Answers / Re: How to calculate the Fermi surface?

« on: April 20, 2009, 12:30 »

This functionality is not implemented in ATK yet.

5165

Future Releases / Re: Usage of graphics card for future ATK technology

« on: April 20, 2009, 11:41 »

We do entertain some plans towards that direction, but we are not in a position to give any details at this point.

5166

Future Releases / Re: Would NanoLanguage support python 3.x?

« on: April 20, 2009, 11:38 »

I took the liberty of editing the subject to make it clearer that (as I understand it) you refer to NanoLanguage.

We are currently working on the next edition of NanoLanguage, which will contain some improvements and updates of the standard. It will, however, still be based on Python 2.x (specifically, 2.6). Support for 3.x might be added in the future, but it is currently not part of our plans.

5167

General Questions and Answers / Re: Question about geometry optimization for molecule

« on: April 19, 2009, 21:33 »

That's right, you don't need to change the charge of Bi from the default, but on the other hand as you noted yourself, changing it or not doesn't really have an impact on the calculation.

The default basis set parameters are common for all elements, but as we've learned in this case, they are not necessarily ideal for all elements. On the other hand, small changes of the basis set parameters often have a relatively small influence on the basis set and the results, and in those cases the exact value of the charge and the other parameters are not extremely important.

Actually, the LDA basis set for Mn is fine, it's only the GGA one which needs the modified charge parameters.

5168

General Questions and Answers / Re: a mpi calculation error, why?

« on: April 19, 2009, 21:08 »

Never seen this before...

Is the error reproducible, i.e. if you run it again, does it fail again in the same place?

Generally, speaking, this is not an ATK error, the error occurs strictly inside MPICH2. A first guess would be that the file system on node 8 is full or not properly mounted (perhaps a temporary network failure).

5169

General Questions and Answers / Re: About the energy resolution in Transmission Coefficients

« on: April 17, 2009, 18:01 »

You mean, is there any risk that you miss important features in the spectrum by using this resolution? No, that should be fine, if there are peaks etc which are so narrow, they are anyway not going to contribute to the current etc.

5170

General Questions and Answers / Re: Question about geometry optimization for molecule

« on: April 16, 2009, 23:17 »

I just wanted to offer a few additional comment about the second point, how to
a) know that there is a problem, and
b) how to solve

It is strongly recommended to always start out small, when you approach a new material. If you use house-hold elements like carbon this is less necessary, but even for gold, platinum etc, and especially spin-dependent elements like manganese and more rarely used transition elements, one should make a habit of first investigating the basic properties of the material. That means, look at simple dimers, the bulk band structure, how does the energy scale with the lattice constants, etc.

That is - precisely what you have done (I assume the Mn-Bi molecule dimer is not your real project? )

This way you learn which mesh cut-off that is required, if LDA is sufficient or GGA is needed, how many k-points, which basis set to use, etc, etc.

So, what to do if problems are encountered? Well, I must admit that tuning the basis set parameters is not the first thing you would try. It is, however, the place you probably end up if all other things fail In this particular case it also helped to plot the Mn basis functions (not a trivial exercise).

And then, once you do end up having to tweak the basis set, the only thing that typically helps is to tune the charge (slightly!); the basis set is not very sensitive to the other parameters, but it does respond nicely to the charge setting.

Now we have informed the community about that!

5171

General Questions and Answers / Re: how to deal with the install problem? Thank you very much!

« on: April 16, 2009, 16:30 »

My immediate suspicion is that you try to run "atk_exec" in the bin directory, rather than "atk", either directly or via some alias or link that points to the wrong place.

I don't see why "chcon" would be relevant in this case...

5172

Scripts, Tutorials and Applications / Re: Script for creating nanotubes

« on: April 16, 2009, 16:20 »

There is no real difference between screening layers and central region layers in this case, so just use the repetition factor the central region.

5173

General Questions and Answers / Re: Question on restarting calculation.

« on: April 16, 2009, 11:00 »

Tricky to say, but my best guess is that the calculation crashed in the equivalent bulk calculation, and never reached the two-probe part. Or the NetCDF file is corrupt, somehow (perhaps it crashed right at the saving of the file...).

5174

General Questions and Answers / Re: I can not find out the LUMO and HOMO!

« on: April 16, 2009, 09:03 »

I would agree that it's not entirely obvious in the MPSH case up-front, since there is no population information available.

However, unlike for molecular spectra, the MPSH energies are shifted such that E=0 is the (average) Fermi level of the system, and thus by definition the HOMO has the highest negative energy and the LUMO the lowest positive energy.

It doesn't really make sense to talk about populations of these levels, other than in the sense of applying the Fermi distribution, but that is precisely equivalent to the discussion just above

At finite bias things get a bit more complicated. It is however important to be aware of the fact the strictly speaking these are not "real" HOMO/LUMO levels. They are some kind of projected levels that correspond to the original molecular levels of the isolated molecule.

5175

Scripts, Tutorials and Applications / Re: Script for creating nanotubes

« on: April 16, 2009, 08:56 »

Well, screening layers are always necessary, if by that we just mean "layers in the central region". Perhaps what you meant, Serhan, is that if the system is perfect anyway, we can just skip the central region and attach the electrodes back-to-back. Geometrically that would be equivalent, and it would seem you could save a lot of time in the calculation this way (fewer atoms!).

However, the algorithms used in ATK are not designed for perfect systems, which are trivial; we don't even need to do the two-probe calculation, we can just inspect the band structure to figure out the transmission spectrum. Therefore, in the general formalism, we always need a central region. In fact, there is a crucial requirement that there is no basis set overlap between atoms in the left and right electrodes, or else the algorithm breaks down.