[Octopus-notify] [Octopus] #37: Unconsistent and radom SCF convergency

Octopus octopus-notify at tddft.org
Mon Jul 3 01:19:22 WEST 2006


#37: Unconsistent and radom SCF convergency
---------------------------+------------------------------------------------
  Reporter:  xavier        |       Owner:  anonymous           
      Type:  defect        |      Status:  new                 
  Priority:  critical      |   Milestone:                      
 Component:  octopus-core  |     Version:  trunk               
Resolution:                |    Keywords:  ifort, gfortran, scf
---------------------------+------------------------------------------------
Comment (by athimm):

 My guess would be unitialized variables, that's the only way you can get
 non-deterministic (iff the input is otherwise equal).

 Since gfortran is not zeroing memory it should be even more affected (?).

-- 
Ticket URL: <http://www.tddft.org/trac/octopus/ticket/37>
Octopus <http://tddft.org/programs/octopus/>
A TDDFT solver



More information about the Octopus-notify mailing list