[Pw_forum] How to orthonormalize evc

Fadjar Fathurrahman break_the_calculus at yahoo.com
Tue Oct 1 09:41:00 CEST 2013

Hi Paolo, 

Thank you for your suggestion.

Well, it seems that I got the result close to 1.d0/sqrt(2.d0) because I
tested it using gamma-point only (oxygen molecule). For other case (silicon fcc,
complex wfc) I indeed got value close to 1.d0.

Best regards,

Fadjar F.

Graduate Student
Department of Precision Science and Applied Physics
Graduate School of Engineering
Osaka University, Suita, Osaka 565-0871, Japan

 From: Paolo Giannozzi <paolo.giannozzi at uniud.it>
To: Fadjar Fathurrahman <break_the_calculus at yahoo.com>; PWSCF Forum <pw_forum at pwscf.org> 
Sent: Monday, September 30, 2013 9:33 PM
Subject: Re: [Pw_forum] How to orthonormalize evc

On Mon, 2013-09-30 at 02:02 -0700, Fadjar Fathurrahman wrote:

>   norm = sqrt( zdotc( npw, evc(1:npw,ibnd),1, evc(1:npw,ibnd),1 ) )

it is safe to use old-style syntax with old-style functions like zdotc.
For complex wavefunctions (i.e. not valid for the Gamma-only case):
   norm = sqrt( abs(zdotc( npw, evc(1,ibnd),1, evc(1,ibnd),1 )) )

> after init_wfc(ik) in c_bands.f90 resulting in value about
> 1.d0/sqrt(2.d0) (for norm conserving case). Isn't this 
> supposed to be close to one?

it is one.

Paolo Giannozzi, Dept. Chemistry&Physics&Environment, 
Univ. Udine, via delle Scienze 208, 33100 Udine, Italy
Phone +39-0432-558216, fax +39-0432-558222 
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.quantum-espresso.org/pipermail/users/attachments/20131001/24dfec2f/attachment.html>

More information about the users mailing list