The ABINIT Users Mailing List ( CLOSED )

[Marc Torrent <marc.torrent@cea.fr>]

Which version of Abinit do you use ?: 4.6.5 recommended for PAW runs...
What do you mean by "convergence" ?, on ecut, on kpts, during SCF cycle ?
Convergence during SCF cycle : change mixing settings (iscf 17, decrease 
diemix...)
Convergence on ecut: keep pawecutdg constant... you shouldnt need 
ecut>20 for Fe for dEtotal<1mHa.
How many empty bands do you have ? Try occopt=7... it could change the 
convergence...

You use prtdos=1 : not that the printed density is not yet the "true" 
one in PAW... this is a feature to be implemented (we have to had 
spherical terms of the density...).

Marc Torrent
CEA Bruyeres-le-Chatel - France


taokun76@hotmail.com a écrit :
> Dear abinit user,
>
> When I calculated the magnetism of the bcc Fe with the PAW potential downloaded from 
> the website, I got a correct result. However, when a slab calculation of bcc Fe with 
> five Fe layers and 3 vacuum layers was performed (PAW potential), it had some 
> difficulties to convergence,even if "ecut" was increased to 60Ha.
>
> Attached below is my input file. Did anyone has similiar experiences? Any 
> comments would be appreciated, thanks in advance!
>
>
> # Fe normal bcc structure for test of a ferromagnetic calculation (PAW method)
>
> nband 96
>
> ecut 60
> pawecutdg 120
>
> #spin related quantities
> spinat 0.0 0.0 4.0
>        0.0 0.0 4.0
>        0.0 0.0 4.0
>        0.0 0.0 4.0
>        0.0 0.0 4.0
> nsppol 2
>
> #print the results
> prtdos 1
> prtden 1
>
> #relaxation
> ionmov 3
> tolmxf 5.0d-4
> ntime 100
>
> iscf 14
> occopt 3
> tsmear 0.01
>
> kptopt 1
> ngkpt 4 4 1
> nshiftk 1
> shiftk 0.5  0.0  0.0
>        0.0  0.5  0.0
>
> acell 3*5.38
> rprim 1.0  0.0  0.0
>       0.0  1.0  0.0
>       0.0  0.0  4.0
>
> natom 5
> ntypat 1
> typat 5*1
> znucl 26
> xcart  3*0
>        3*2.69
>        2*0     5.38
>        2*2.69  8.07
>        2*0     10.76
>
> nstep 100
> toldfe 1.0d-6