The ABINIT Users Mailing List ( CLOSED )

[matthieu verstraete <matthieu.jean.verstraete@gmail.com>]

Indeed,

If you have internal degrees of freedom you might try to relax them with a fixed cell before continuing with the stress. Does this converge properly?

Also have a look at which atoms have large forces left, and the pressure. It could be that tolrff 0.02 is not enough to get the stress converged enough (alternatively you can increase strfact). By checking the changes in atomic position and cell shape from one step to the next, you may see oscillation in one or both, indicating the algorithm can't find a nice minimum. Keeping the cell fixed may allow you to stabilize this.

And as always it still could be a pseudopotential problem... Have you used the 3 separately or on simpler structures? In particular the Ca (try elemental Ca and an oxide to be sure).

Matthieu

On Thu, Mar 5, 2009 at 8:12 AM, Anglade Pierre-Matthieu <anglade@gmail.com> wrote:

Hi,

When there is a lot of degrees of freedom the "ionmov 2" algorithm
looses some of its efficiency. It is not impossible that you get
better result with a viscous dumping or beeler molecular dynamics.

regards

PMA

On Wed, Mar 4, 2009 at 11:57 PM,  <drhendry@gmail.com> wrote:
> Hello,
>
> I'm trying to optimise the geometry of a 20 atom CaTiO3 orthorhombic bulk
> dielectric.  The process is taking considerably longer and does not seem to be
> converging compared with a 6 atom TiO2 rutile bulk dielectric.  The max force
> does not seem to go below ~ 1e-3, wheres TiO2 max force goes to machine
> precision.  I'm using Troullier-Martins psps and the input file is below.
>
> Could there be some way of improving this result, or is this to be expected?
>
> Thank you very much, in advance,
>
> David.
>
> #Definition of lattice parameters
> #********************************
>   acell  1.1034564219E+01  1.5140354875E+01  1.0791197046E+01
>
> #Definition of atomic positions and types
> #****************************************
>   natom       20
>   ntypat      3
>   znucl       20 22 8
>   typat       2 2 2 2 1 1 1 1 3 3 3 3 3 3 3 3 3 3 3 3
>   xcart
>  1.74102557844419E-21  6.94219419473125E-21  6.66433845029123E-21
>  5.51728210951351E+00  2.01064184754249E-07  5.39559852296522E+00
>  2.72273479496933E-07  7.57017743773540E+00  6.66433845029123E-21
>  5.51728238178699E+00  7.57017763879958E+00  5.39559852296522E+00
>  5.52568087138869E+00  3.78516729878331E+00  1.46347634448159E-02
>  -8.39821732821760E-03  1.13551877777517E+01  5.41023328641004E+00
>  -5.52568032684173E+00  1.13551875766875E+01 -1.46347634448159E-02
>  1.10429629809022E+01  3.78516749984750E+00  5.38096375952040E+00
>  -7.62652580040250E-03  3.78509505890158E+00 -2.23307892067039E-01
>  5.52490917986087E+00  1.13552600176334E+01  5.17229063089818E+00
>  7.62707034736165E-03  1.13552598165692E+01  2.23307892067039E-01
>  5.50965558371311E+00  3.78509525996576E+00  5.61890641503226E+00
>  2.78700708595054E+00  1.43408602047991E-01  2.66783642053104E+00
>  2.73027502356297E+00 -1.43408400983806E-01  8.06343494349627E+00
>  -2.78700845918014E+00  7.71358803235092E+00 -2.66784679121249E+00
>  8.30429111324061E+00  7.42676704418406E+00  2.72775173175273E+00
>  -2.78700708595053E+00 -1.43408602047991E-01 -2.66783642053104E+00
>  8.30428919546405E+00  1.43408803112175E-01  2.72776210243418E+00
>  2.78700900372710E+00  7.42676684311987E+00  2.66784679121249E+00
>  2.73027365033337E+00  7.71358823341510E+00  8.06344531417771E+00
>
> #Echange-correlation functional
> #******************************
>   #ixc         1
>
> #Plane wave basis and k-point grid
> #*********************************
>   ecut        30
>   ngkpt       7 5 7
>   nshiftk     1
>   shiftk      0.5 0.5 0.5
>   nband       48
>
> #Relax ions and unit cell
> #************************
>    #chkprim    0
>    #tsmear     0.01
>    kptopt     2
>    iscf       7
>    npulayit   7
>    iprcel     45
>    tolrff     0.02
>    #tolvrs     1.0d-18
>    #tolmxf     1.0d-8
>    nstep      200
>    optcell    2
>    ionmov     2
>    ntime      200
>    dilatmx    1.1
>    ecutsm     0.5
>    prtwf      0
>    prtden     0
>

--
Pierre-Matthieu Anglade

-- 
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Dr. Matthieu Verstraete

European Theoretical Spectroscopy Facility (ETSF) 
Dpto. Fisica de Materiales, 
U. del Pais Vasco,            
Centro Joxe Mari Korta, Av. de Tolosa, 72,   Phone: +34-943018393
E-20018 Donostia-San Sebastian, Spain        Fax  : +34-943018390

Mail : matthieu.jean.verstraete@gmail.com
http://www-users.york.ac.uk/~mjv500