This thread is locked.Only browsing is available.
Top Page > Browsing
LCAO coefficients and overlaps from HS.scfout file are connected?
Date: 2018/03/21 03:38
Name: P

Hello,

I am trying to use the LCAO coefficients printed in the *.out file when the flag "MO.fileout on" is on, for postprocessing the wavefunction form openmx3.7. We know the LCAO basis is not orthogonal. My question is whether the overlaps can be extracted from the overlaps written in the HS.scfout.

From what I understand, from reading openmx documentation and their references, one LCAO basis element is a Bloch sum over one psudoatomic orbital (for example if I set up a fictitious system with a cubic unit cell and one hydrogen atom, then the LCAO basis element corresponding to the s-orbital of hydrogen is Sum[ e^{i k rj}*s-orbital_rj ] with rj ruining through all unit cell positions, and every new orbital I introduce to the calculation has a new similar LCAO basis element). In the HS.scfout the interface for developers say I can find overlaps. These overlaps are overlaps between psudoatomic orbitals? In that case I can use these to compute overlaps between LCAO basis element which should be just Bloch sums of psudoatomic orbital correct? Or perhaps I am misunderstanding what the printed data are, and what I am saying is just wrong.

Any clarification is much appreciated.

Best,
P
メンテ
Page: [1]

Re: LCAO coefficients and overlaps from HS.scfout file are connected? ( No.1 )
Date: 2018/03/21 05:37
Name: Chi-Cheng Lee  <cclee.physics@gmail.com>

Hello,

According to my understanding, the overlap matrix element stored in HS.scfout is already
the real-space overlap, which is the s-orbital_rj you mentioned. If you want to use it as the
basis functions for getting the Kohn-Sham eigenstates at k, you can perform the Bloch sums
|k s-orbital_rj> = Sum_rj e^{i k rj} * | rj s-orbital>, where the <0 s-orbital | rj s-orbital> is stored
in HS.scfout.

Cheers,
Chi-Cheng
メンテ
Re: LCAO coefficients and overlaps from HS.scfout file are connected? ( No.2 )
Date: 2018/03/21 05:38
Name: P

Good that clears it up thank you!
メンテ

Page: [1]