[Pw_forum] SCF convergency

[Jun-jun Yang]

Eyvaz and list users,

Thank you for your timely response.

I changed some of the parameters, and now the calculation converges within 
50 iterations (conv_thr=1.0e-6). But I noticed that the total energy is a 
little higher than before (-5807 Ry
vs. -5865 Ry). It seems that it's most likely a local minimum. Should I 
decrease the convergence
threshold to 1e-7 or 1e-8? Probably, I need to start the calculation from 
another point to avoid
the local minimum. How can we do this in PWScf?

In my calculation, as you suggested, I set start_magnetization to 0.9 for 
Fe, but I really don't
know what number I should give for Cr. I just set 0.9 for Cr, 0.0 for O. Are 
there any guideline
to set these values?

Regards,


Yang, Jun-jun



>From: Eyvaz Isaev <eyvaz_isaev at yahoo.com>
>Reply-To: pw_forum at pwscf.org
>To: pw_forum at pwscf.org
>Subject: Re: [Pw_forum] SCF convergency
>Date: Sun, 30 Oct 2005 08:35:32 -0800 (PST)
>
>Hi,
>
>Hopefully you know that celldm(1)=15.8321 is in atomic
>units, but not Angstrom.
>
>You asked to do spin-polarized calculations (nspin=2),
>but put initial magnetic splitting  to zero.
>So, you have to change
>starting_magnetization(1) = 0.9 and add more 2 same
>lines for Cr and O atoms (your first atomic type is
>Fe). Presumably, you should not specify the bands
>number, so, remove nbnd=330. Be careful with this
>number, it means that you have around 600-660
>electrons in your crystal.  If you are sure that your
>nbnd covers bands above the Fermi level,then it is OK.
>
>
>Ecutrho seems to be low, increase up to 200Ry.
>Try mixing_beta=0.1.
>degauss=0.5? It is very large. Try degauss=0.03, it is
>in Ry.
>To me conv_thr=1.d-5 seems too large. Default conv_thr
>(1d-6) looks more acceptable.
>
>Finally, if something went OK, be ready for LDA+U
>calculations. Fe-Cr-O looks like a strongly correlated
>system.
>
>Best regards,
>Eyvaz.
>
> > &system
> >     ibrav = 1, celldm(1) = 15.8321, nat = 56, ntyp =
> > 3
> >     nbnd = 330
> >     nspin = 2, degauss = 0.5
> >     occupations = 'smearing'
> >     smearing = 'gaussian'
> >     starting_magnetization(1) = 0.0
> >     ecutwfc = 25.0
> >     ecutrho = 140.0
> > /
> > &electrons
> >     diagonalization = 'cg'
> >     mixing_mode = 'plain'
> >     mixing_beta = 0.40
> >     conv_thr = 1.0d-5
> > /
> >
> > ATOMIC_SPECIES
> > Fe  55.847	Fe.pbe-sp-van.UPF
> > Cr  51.996	Cr.pbe-sp-van.UPF
> > O   15.999	O.pbe-van_ak.UPF
> >
> > ATOMIC_POSITIONS angstrom
> > Fe    0.00000	 0.00000    0.00000
> > Cr    5.23625	 5.23625    5.23625
> > O    3.23391	 3.23391    3.23391
> > Fe    0.00000	 4.18900    4.18900
> > Fe    4.18900	 0.00000    4.18900
> > [...]
> >
> > K_POINTS automatic
> > 1 1 1 0 0 0
> >
> >
> > I have sent this mail three times with a graph
> > attached. The first two posts
> > had been rejected
> > because size of the mail was a little larger than
> > the limit. The third one
> > should be below 40KB,
> > but I haven't seen its appearing on the list yet.
> > So, I have to post it
> > again without a graph. I
> > apologize for that you may receive two mails with
> > similar content.
> >
> >
>_________________________________________________________________
> > On the road to retirement? Check out MSN Life Events
> > for advice on how to
> > get there!
> >
>http://lifeevents.msn.com/category.aspx?cid=Retirement
> >
> > _______________________________________________
> > Pw_forum mailing list
> > Pw_forum at pwscf.org
> > http://www.democritos.it/mailman/listinfo/pw_forum
> >
>
>
>
>
>
>__________________________________
>Yahoo! Mail - PC Magazine Editors' Choice 2005
>http://mail.yahoo.com
>_______________________________________________
>Pw_forum mailing list
>Pw_forum at pwscf.org
>http://www.democritos.it/mailman/listinfo/pw_forum

_________________________________________________________________
Don’t just search. Find. Check out the new MSN Search! 
http://search.msn.click-url.com/go/onm00200636ave/direct/01/