[Pw_forum] SCF converges to different total magnetization values.

[Paolo Giannozzi]

On Fri, 2012-10-19 at 11:58 +0200, Yury Vishnevskiy wrote:

> Why is this happening? 

hard to say without outputs or reproducible data

> (how is decides that the system is metallic?) 

like this:
  !
  ! ... Check: if there is an odd number of electrons, the crystal is a
metal
  !
  IF ( lscf .AND. ABS( NINT( nelec / 2.D0 ) - nelec / 2.D0 ) > eps8 &
            .AND. .NOT. lgauss .AND. .NOT. ltetra .AND. .NOT.
tfixed_occ ) &
      CALL infomsg( 'setup', 'the system is metallic, specify
occupations' )

In this peculiar case, since magnetization is kept fixed, likely
this check is overzealous

P.
-- 
Paolo Giannozzi, IOM-Democritos and University of Udine, Italy