QuantumATK Forum

QuantumATK => General Questions and Answers => Topic started by: yasheng on April 22, 2016, 17:21

Title: LDA+U with OpenMXBasisSet
Post by: yasheng on April 22, 2016, 17:21

Hi All,

If we want to do LDA+U calculations using OpenMXBasisSet in ATK, how do we adjust the script to include the hubbard U parameter in basis set?  (demonstration for NiO or CuO shall be fine).

In tutorial http://quantumwise.com/documents/tutorials/latest/NiO_LDA_U/index.html/chap.crystal.ldau.html
FHI basis sets are used.

Thank you.

Regards,
Yasheng

Title: Re: LDA+U with OpenMXBasisSet
Post by: Anders Blom on April 23, 2016, 06:44

Just select the OMX basis set. Nothing special to change besides that (except a slightly different U maybe, there is some small sensitivity to the basis set).

Title: Re: LDA+U with OpenMXBasisSet
Post by: yasheng on April 25, 2016, 10:57

Hi Anders,

If the U term is set by default with OMX basis set, how do we check the value of U? 
Thanks,

Title: Re: LDA+U with OpenMXBasisSet
Post by: Anders Blom on April 26, 2016, 01:07

No, that's not what I mean. I meant, use the Script Generator in VNL to specify the basis set and the U parameters, that's the easiest way.

Title: Re: LDA+U with OpenMXBasisSet
Post by: Jess Wellendorff on April 26, 2016, 08:50

You can set the +U parameters in the "Basis set/exchange correlation" tab in the Script Generator. Just select Onsite or Dual Hubbard U, then the +U options will appear in the basis set settings.

Title: Re: LDA+U with OpenMXBasisSet
Post by: ramkrishna on April 27, 2016, 22:35

Hi Jess,
I was also trying to add the U value within the omx basis set and  so I tried your suggestion. Now, when I open the file in editor I could not  find the U term in the baisis (even in show default format) like other FHI or HGH basis sets. Is it really adding U in that basis or not? quite confusing.

Title: Re: LDA+U with OpenMXBasisSet
Post by: Jess Wellendorff on May 3, 2016, 13:57

You have unfortunately found a bug in the software. DFT+U should indeed work with OMX potentials, but for some reason it currently does not. Will most likely be fixed for ATK 2016.

Title: Re: LDA+U with OpenMXBasisSet
Post by: yasheng on May 3, 2016, 15:25

Hi Jess,

I was confused about this about few weeks, now it is clear that it is a bug.  I look forward for the new version.

If DFT+U works with OMX and the script generator has the problem, can you please show DFT+U implementation with OMX basis sets in a script?   e.g. for copper oxide or nickel oxide.

Thank you for your help.


Regards,
Yasheng

Title: Re: LDA+U with OpenMXBasisSet
Post by: Jess Wellendorff on May 4, 2016, 10:00

It is unfortunately not just a Scripter bug. DFT+U with OMX potentials were not implemented in ATK either. That is a bug, because there is no reason this should not be possible, it just wasn't done yet. We'll have to wait for the ATK 2016 release...