[Pw_forum] Can ph.x use 'cg' diagonalization?

GAO Zhe flux_ray12 at 163.com
Sat Jan 7 02:35:22 CET 2012


Dear QE developer and users:
I am using ph.x to calculate phonon dispersion. But I met a problem during my computation.
My structure needs 12 dynamic matrics files. But after 3 *.dyn file were calculated, the error occurred during band structure calculation:
Band Structure Calculation
Davidson diagonalizaion with overlap
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
  from cdiaghg:  error #            126
  diagonalization  (ZHEGV*)  failed
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
Therefore, I changed to diagonalization='cg' in scf calculation and recalculate it, the problem remained. It seemed no matter which diag method I used in scf calculation, only Davidson algorithm is used in ph.x.
Is it possible to use conjugate gradient in ph.x ?
Thanks a lot.


--
GAO Zhe
CMC Lab, Materials Science & Engineering Department,
Seoul National University, South Korea
 
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://www.democritos.it/pipermail/pw_forum/attachments/20120107/3238828b/attachment.htm 


More information about the Pw_forum mailing list