QuantumATK Forum

QuantumATK => General Questions and Answers => Topic started by: Fraenzine on October 16, 2015, 13:04

Title: Integrating DOS
Post by: Fraenzine on October 16, 2015, 13:04

Hello,

Is there a way to integrate the density of states, to see how many electrons are at each atom and orbital?

Thanks!

Title: Re: Integrating DOS
Post by: Anders Blom on October 16, 2015, 15:20

I believe your are looking for the Mulliken populations (http://www.quantumwise.com/documents/manuals/latest/ReferenceManual/index.html/ref.mullikenpopulation.html)

Title: Re: Integrating DOS
Post by: Fraenzine on October 19, 2015, 15:09

No, actually I wasn't, I simply wanted the integral over my density of states.

However, Mulliken populations would be also a very good idea and a thing that I'd like to do. From the link that you send me, I did not really understand how to apply it.
If I have all my output files from vasp, structure, charge density (CHGCAR) and density of states (DOSCAR) files, how can I now apply the Mulliken population analysis on it?

Thanks :)

Title: Re: Integrating DOS
Post by: Anders Blom on October 20, 2015, 09:54

The DOS is just a number, as function of energy, so you can use any technique to integrate it up to a certain energy, or between two energies. You can use for instance http://docs.scipy.org/doc/scipy-0.16.0/reference/generated/scipy.integrate.simps.html#scipy.integrate.simps where y=dos.evaluate(spin=Spin.Sum) and x=dos.energies(). Or a subset of energies - integrating the full range, with the Fermi function, just gives you the number of electrons in the system, and there are easier ways to calculate that!

Just note that integrating the DOS will tell you nothing about how the electrons are distributed per atoms. This is true both for DOS obtained from ATK, or VASP, or other sources. It is not possible to perform a Mulliken analysis on the VASP results because it's primarily an atomic orbital concept, with no direct equivalent in a plane-wave basis.

Title: Re: Integrating DOS
Post by: Umberto Martinez on October 20, 2015, 13:41

Quote

No, actually I wasn't, I simply wanted the integral over my density of states.

the DOSCAR file you have already includes the integrated DOS data.
check vasp documentation for details.

what we can also do is to include this info in the DOS plot, something that is not possible in the current version of VNL.

Title: Re: Integrating DOS
Post by: kstokbro on October 20, 2015, 14:44

You may also use the Bader charge plugin in VNL to get the bader charges on each atom. This is the standard approach to get a charge on each atom.

Title: Re: Integrating DOS
Post by: Fraenzine on October 20, 2015, 19:56

Thanks for your replies! :)

Regarding the Bader analysis tool. I had seen it before and wanted to use it. However, it is not working (when I click onto it it appears light grey).
I tried to open it by selecting first the CHGCAR or DOSCAR file and also the AECCAR0 -2 files, which were I created by the LAECHG=.TRUE. tag in the INCAR.

I checked the AddOn Manager, apparently it is installed though.

Title: Re: Integrating DOS
Post by: Anders Blom on October 31, 2015, 12:38

It is not sufficient to select just the files, you must select the corresponding objects on the LabFloor. Specifically, for VASP output, you should select 4 objects: the BulkConfiguration, the ElectronDensity, the core electron density (from AECCAR0) and the non-pseudoized valance electron density (from AECCAR2). When you do (hopefully) the Bader plugin will magically be enabled :)

Title: Re: Integrating DOS
Post by: Fraenzine on November 10, 2015, 13:56

Thank you very much! It worked indeed :)
I haven't thought about enabling the configuration (CONTCAR)...