Skip to Content.
Sympa Menu

forum - Re: [abinit-forum] Order of band indices in electron-phonon matrix elements

forum@abinit.org

Subject: The ABINIT Users Mailing List ( CLOSED )

List archive

Re: [abinit-forum] Order of band indices in electron-phonon matrix elements


Chronological Thread 
  • From: matthieu verstraete <mjv500@york.ac.uk>
  • To: forum@abinit.org
  • Subject: Re: [abinit-forum] Order of band indices in electron-phonon matrix elements
  • Date: Tue, 27 May 2008 21:41:42 +0200 (CEST)

Dear Experts,
Dear disciple,

fact that the matrix elements are complex in general. As I go along the line
of the 2*nband*nband printed elements,

1) are the real and imaginary parts of the matrix elements printed in pairs
one after the other, or are all the nband*nband real parts printed first,
and then all imaginary parts, or somehow else?
yes: real(eig1_1,1) imag(eig1_1,1) real(eig1_2,1) imag(eig1_2,1)
...

eig1 is the 1st order eigenvalue (matrix) and _2,1 are the indices.


2) what band index is changed first (the inner loop): of the k+q
wavefunctions or of the k wavefunctions (for matrix elements <k+q|V|k>)?
You can confirm what I write above in the abinit subroutine that calculates the eig1. See in particular cgwf3.F90:


!The array eig1_k contains
<u_(iband,k+q)^(0)|H_(k+q,k)^(1)|u_(band,k)^(0)>
do iband=1,nband
eig1_k(2*iband-1 +(band-1)*2*nband)=scprod(1,iband)
eig1_k(2*iband +(band-1)*2*nband)=scprod(2,iband)
end do
<<

so the fast index is on k+q.

Matthieu



Archive powered by MHonArc 2.6.16.

Top of Page