Dear Cyrille,<br><br>I recalculated the dimer with different values of nbnd. Now some interesting changes appeared:<br><br>Value of nbnd | Magnetization<br><br>10 4.00<br>12 6.88<br>14 6.49<br>16 6.49<br>
18 6.49<br>20 6.49<br>22 6.88<br>24 6.88<br>26 6.49<br>28 6.49<br><br>Also we dimer is now no longer at the equilibrium position; taking for example for nbnd=20:<br><br>Forces acting on atoms (Ry/au):<br>
atom 1 type 1 force = 0.02342880 0.00000000 0.00000000<br> atom 2 type 1 force = -0.02342880 0.00000000 0.00000000<br><br>For the record, the smearing contribution on these calculations goes like this : smearing contrib. (-TS) = -0.00101846 Ry<br>
which were made using degauss = 0.005, with 'methfessel-paxton' smearing type.<br><br>So looks like there are two things to improve: reduce the smearing contribution and also use a high nbnd. Peharps those high values of magnetization at nbnd=22, 24 could be due to smearing?<br>
<br>I'll test it and report back on the list.<br><br>Any other suggestions, just say it, I'll give it a try.<br><br>Thanks,<br><br>Giovani Faccin<br><br><br><br><br><br><div class="gmail_quote">2011/2/14 BARRETEAU Cyrille <span dir="ltr"><<a href="mailto:cyrille.barreteau@cea.fr">cyrille.barreteau@cea.fr</a>></span><br>
<blockquote class="gmail_quote" style="margin: 0pt 0pt 0pt 0.8ex; border-left: 1px solid rgb(204, 204, 204); padding-left: 1ex;">Dear Giovani<br>
<br>
I personnaly did calculations on Fe clusters some time ago. I always found "almost integer" magnetic moments.<br>
However I once encoutered some difficulties with the default value of "nbnd". I had to increase it. Try to set nbnd=20 for example...<br>
<br>
Magnetic clusters can be very tricky.. It also very common to find several magnetic solutions..<br>
<br>
good luck<br>
<br>
cyrille<br>
<br>
==================================================================<br>
Cyrille Barreteau phone : +33 (0)1 69 08 29 51<br>
CEA Saclay fax : +33 (0)1 69 08 84 46<br>
IRAMIS, SPCSI, Batiment 462 email <a href="mailto:cyrille.barreteau@cea.fr">cyrille.barreteau@cea.fr</a><br>
91191 Gif sur Yvette Cedex<br>
FRANCE<br>
~~~~~~~~~~~~~~~~~~~~~~~~<br>
Web: <a href="http://iramis.cea.fr/Pisp/cyrille.barreteau/" target="_blank">http://iramis.cea.fr/Pisp/cyrille.barreteau/</a><br>
==================================================================<br>
<br>
________________________________<br>
<br>
De: <a href="mailto:pw_forum-bounces@pwscf.org">pw_forum-bounces@pwscf.org</a> de la part de Giovani Faccin<br>
Date: lun. 14/02/2011 05:50<br>
À: PWSCF Forum<br>
Objet : Re: [Pw_forum] RE : A question on atomic magnetic moments<br>
<br>
<br>
Dear Cyrille,<br>
<br>
First of all, thanks for the reply on my post.<br>
<br>
Following your suggestion, I've run a few calculations with lower degauss values. In particular, the ones reported on the message for Duy Le were made with a much smaller value ( 0.005 ) than the original 0.02 from the first message. Besides that, I've tried degauss values as low as 0.002 without any significant improvement on the final magnetization.<br>
<br>
I'm still wondering how do people find 6.00 so easily for the magnetization. I'm trying with 2 different DFT codes and no way I'm reaching those results, even though bulk/dimer equilibrium distances and force constants appear to be ok.<br>
<br>
Any other ideas, just tell me, I'll try it and report back on the list.<br>
<br>
Thanks again,<br>
<br>
Giovani<br>
<br>
<br>
<br>
2011/2/13 BARRETEAU Cyrille <<a href="mailto:cyrille.barreteau@cea.fr">cyrille.barreteau@cea.fr</a>><br>
<br>
<br>
Dear Giovani<br>
Try to decrease the degauss value of your smearing.. and see what happens.<br>
<br>
cyrille<br>
<br>
<br>
==================================================================<br>
Cyrille Barreteau phone : +33 (0)1 69 08 29 51<br>
CEA Saclay fax : +33 (0)1 69 08 84 46<br>
IRAMIS, SPCSI, Batiment 462 email <a href="mailto:cyrille.barreteau@cea.fr">cyrille.barreteau@cea.fr</a><br>
91191 Gif sur Yvette Cedex<br>
FRANCE<br>
~~~~~~~~~~~~~~~~~~~~~~~~<br>
Web: <a href="http://iramis.cea.fr/Pisp/cyrille.barreteau/" target="_blank">http://iramis.cea.fr/Pisp/cyrille.barreteau/</a><br>
==================================================================<br>
<br>
________________________________<br>
<br>
De: <a href="mailto:pw_forum-bounces@pwscf.org">pw_forum-bounces@pwscf.org</a> de la part de Giovani Faccin<br>
Date: dim. 13/02/2011 07:16<br>
À: PWSCF Forum<br>
Objet : Re: [Pw_forum] A question on atomic magnetic moments<br>
<br>
<br>
Dear Paolo,<br>
<br>
Thanks for the reply.<br>
<br>
Just to make sure that the simulation cell size is not an issue (so that the system is really finite), I've tested different possibilities, including some quite big cells. Still, after a certain size the value of the magnetization is converging to 6.82 instead of the expected 6.<br>
<br>
So something else is causing this. Unfortunately I've got no clue as to what could it be.<br>
<br>
Should it be of any help, this is my input file:<br>
<br>
<br>
&CONTROL<br>
calculation = 'scf' ,<br>
restart_mode = 'from_scratch' ,<br>
wf_collect = .true. ,<br>
outdir = './output' ,<br>
pseudo_dir = '../pseudo/' ,<br>
prefix = 'fe' ,<br>
etot_conv_thr = 1.0D-9 ,<br>
forc_conv_thr = 1.0D-6 ,<br>
tstress = .true. ,<br>
tprnfor = .true. ,<br>
/<br>
&SYSTEM<br>
ibrav = 1,<br>
celldm(1) = 43,<br>
nat = 2,<br>
ntyp = 1,<br>
ecutwfc = 100 ,<br>
ecutrho = 300 ,<br>
occupations = 'smearing' ,<br>
degauss = 0.02 ,<br>
smearing = 'methfessel-paxton' ,<br>
nspin = 2 ,<br>
starting_magnetization(1) = 1.0,<br>
/<br>
&ELECTRONS<br>
conv_thr = 1.0e-9 ,<br>
mixing_beta = 0.7 ,<br>
diagonalization = 'david' ,<br>
/<br>
ATOMIC_SPECIES<br>
Fe 58.69000 Fe.pbe-nd-rrkjus.UPF<br>
ATOMIC_POSITIONS angstrom<br>
Fe 0.000000000 0.000000000 0.000000000<br>
Fe 2.047510 0.000000000 0.000000000<br>
K_POINTS automatic<br>
1 1 1 1 1 1<br>
<br>
Any suggestions are highly welcome.<br>
<br>
Giovani<br>
<br>
<br>
<br>
2011/2/12 Paolo Giannozzi <<a href="mailto:giannozz@democritos.it">giannozz@democritos.it</a>><br>
<br>
<br>
<br>
On Feb 11, 2011, at 17:09 , Giovani Faccin wrote:<br>
<br>
> My question: shouldn't those numbers be integers?<br>
<br>
<br>
only in finite systems, if you neglect spin-orbit.<br>
<br>
P.<br>
---<br>
Paolo Giannozzi, Dept of Chemistry&Physics&Environment,<br>
Univ. Udine, via delle Scienze 208, 33100 Udine, Italy<br>
Phone +39-0432-558216, fax +39-0432-558222<br>
<br>
<br>
<br>
<br>
_______________________________________________<br>
Pw_forum mailing list<br>
<a href="mailto:Pw_forum@pwscf.org">Pw_forum@pwscf.org</a><br>
<a href="http://www.democritos.it/mailman/listinfo/pw_forum" target="_blank">http://www.democritos.it/mailman/listinfo/pw_forum</a><br>
<br>
<br>
<br>
<br>
<br>
--<br>
Giovani<br>
<br>
<br>
_______________________________________________<br>
Pw_forum mailing list<br>
<a href="mailto:Pw_forum@pwscf.org">Pw_forum@pwscf.org</a><br>
<a href="http://www.democritos.it/mailman/listinfo/pw_forum" target="_blank">http://www.democritos.it/mailman/listinfo/pw_forum</a><br>
<font color="#888888"><br>
<br>
<br>
<br>
<br>
<br>
--<br>
Giovani<br>
<br>
</font><br>_______________________________________________<br>
Pw_forum mailing list<br>
<a href="mailto:Pw_forum@pwscf.org">Pw_forum@pwscf.org</a><br>
<a href="http://www.democritos.it/mailman/listinfo/pw_forum" target="_blank">http://www.democritos.it/mailman/listinfo/pw_forum</a><br>
<br></blockquote></div><br><br clear="all"><br>-- <br>Giovani<br>