[Q-e-developers] ACE allows us to plot band structures of Hybridfunctionals ?

[西原慧径]

thank you, Stefano.
I hope someone implements this method!

Best regards,
Satomichi


2017/04/06 23:16 "stefano de gironcoli" <degironc at sissa.it>:

Dear Satomichi Nishihara,
  I think it would.
  I think the interpolated ACE projectors should be a good approximation of
the ones one could obtain with a very dense grid so it should be ok up to
the bands included in the summation. I don't know if anybody is trying to
implement this but it would be a nice feature to have because it would
allow to do nscf calculations with hybrids, at least approximately.

stefano


On 06/04/2017 15:19, nisihara225 at gmail.com wrote:

Dear Stefano,



you are correct. I misunderstood the formulation of ACE.

Because the projectors depend on a k-point, it is impossible to apply for
another k-point.

As you pointed out, interpolation of the projectors in BZ will yield
approximate band structure.

Will it be worth to try this approximate band structure ?



Best Regards,

Satomichi Nishihara





*差出人: *stefano de gironcoli <degironc at sissa.it>
*送信日時: *2017年4月5日 22:05
*宛先: *q-e-developers at qe-forge.org
*件名: *Re: [Q-e-developers] ACE allows us to plot band structures of
Hybridfunctionals ?



On 05/04/2017 13:43, nisihara225 at gmail.com wrote:

Dear QE’s developers,



I am studying ACE (Lin Lin, JCTC 2016, 12, 2242), now.

And I have one question: May ACE allow us to plot band structures of hybrid
functionals ?

NO, and maybe yes.
the ACE construction builds the projectors for the provided k-points.
This requires the knowledge of the full Vx phi for a number of orbitals
which you only know on the k-points you used in the scf cycle.
However one could Fourier interpolate/Wannier interpolate the ACE
projectors to a set of non-scf k-points to perform the band structure.
Beware: the ACE construction is good only for the bands you built the
projector for.

stefano


Once the projection vector of ACE (Eq(14) of JCTC2016) is constructed,

the exact-exchange operator is independent of all the wave functions in the
Fermi sea.

Then, I suggest saving the projection vector to storage after SCF
calculation and

reading it before NSCF calculation. One can set arbitrary k-points in NSCF.

This scheme is similar to non-local potential of PAW or density matrix of
LDA+U.

I would like to hear your opinions on this idea.



Best Regards,

Satomichi Nishihara






_______________________________________________

Q-e-developers mailing list

Q-e-developers at qe-forge.org

http://qe-forge.org/mailman/listinfo/q-e-developers
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.quantum-espresso.org/pipermail/developers/attachments/20170407/2209541a/attachment.html>