The ABINIT Users Mailing List ( CLOSED )

["Anglade Pierre-Matthieu" <anglade@gmail.com>]

Hello fet,

From your input we guess that you are trying to get amorphous
materials from a really out of equilibrium structure. I've discussed
with P. Boulanger about your file and here are a few comments:

1) getwfk -1 is not usefull here since you expect all your
configuration to lead to very different amorphous state. At best this
makes your calculations longer.

2) Hopefully we interpret your starting acell, for your four datasets,
correctly. Otherwise we really wonder what they are intended to. We
think that you expect quite different relaxations under different
pressures.

3) your ecut looks small for a cell with oxygen. By contrast your k
point mesh looks terribly converged. Even for the smallest cell and
not mentioning the 3 next where it is going to be much worst than a
waste of computation time. You may want to find an equilibrium between
k-mesh convergence and ecut convergence.

4) It looks like your starting configuration are much too far from
equilibrium for CG to converge. You may better try to start with
something like "ionmov 7", get semi relaxed atmic position then
continue with ionmov 3.

Regards

PMA


On 5/2/07, fet@iisb.fraunhofer.de <fet@iisb.fraunhofer.de> wrote:
> Hi,
>
> Here is a copy of my input file for the calculation on amorphous
> silicon. The forces just don't converge any help is appreciated. I keep
> getting the following warning
>
> fconv : WARNING -
>   ntime=   25 was not enough Broyd/MD steps to converge gradients:
>   max grad (force/stress) = 1.6705E+02 > tolmxf= 5.0000E-04 ha/bohr
> (free atoms)
>
>
> Thanks
>
> #ABINIT input file to calculate electronic structure of bulk
> # amorphous Silicon dioxide
> #
>
> #prtden   1             #print the electron density
> #prtdos   2             #print the LDOS
> #prteig   1             #print the eigenenergies
> enunit   1             #output energies in electron volts
> getwfk  -1
>
> ndtset   4             #two data sets: one crystalline one amorphous
> udtset   1 4
> #-----------------------------------------------------------------------
> #first data set defining crystalline SiO2 with FCC lattice structure and
> 6 atom basis
>
> #define cell dimensions
> acell?:    10.217 10.217 10.217 acell?+ 5.0 5.0 5.0
> rprim    0.5 0.5 0.0   #define FCC lattice for SiO2
>          0.0 0.5 0.5
>          0.5 0.0 0.5
>
> #chkprim  0            #allow the usage of a non-primitive cell:
>
> #define the atoms
> natom    6            #6 atoms per unit cell
> ntypat   2            #two types of atoms
> znucl    14 8         # atoms with atomic numbers 14 for silicon and 8
> for oxygen
> typat    2*1 4*2      #2 silicon atoms and 4 oxygen atoms
> xred
>          3*0.0
>          3*0.25
>          3*0.125
>          1/4 1/8 1/4
>          1/8 1/4 1/4
>          3/8 3/8 3/8
> #---------------------------------------------------------------------------
>
> #Exchange-correlation functional
> ixc     11           #GGA perdew-burke-ernzerhof functional
>
>
> #Definition of the planewave basis set
> ecut1?    14.0
>
>
> #Definition of the SCF procedure
> nstep   20          # Maximal number of SCF cycles
> toldff  5.0d-5
>
> #diemac  12.0       # Although this is not mandatory, it is worth to
> iprcell  45         # precondition the SCF cycle. The model dielectric
>                     # function used as the standard preconditioner
>                     # is described in the "dielng" input variable section.
>                     # Here, we follow the prescription for bulk silicon.
>
>
> #Definition of the k-point grid
> kptopt    1         # Option for the automatic generation of k points,
> taking
>                     # into account the symmetry
> ngkpt     8 8 8
> nshiftk   4
> shiftk    0.5 0.0 0.0
>           0.0 0.5 0.0
>           0.0 0.0 0.5
>           0.5 0.5 0.5
>
> #Optimization of the lattice parameters
> ionmov  3
> ntime   25
> tolmxf  5.0d-4
>
> #dilatmx 1.05
> #ecutsm  0.5
>
>
> Anglade Pierre-Matthieu schrieb:
> > > Hi,
> > >
> > > For us to be able to help you efficiently you should send us your
> > > input and output file. Many configuration problems can prevent
> > > convergency of a calculation
> > > (http://www.abinit.org/about/?text=../Infos/netiquette).
> > >
> > > If you have made any error in your input it is natural that your
> > > calculation do not converge :
> > > not enough bands, atoms misplaced, bad estimation of your static
> > > macroscopic permittivity (diemac) ...
> > >
> > > If all the above is correct, you may try to use "iprcel".
> > >
> > > regards
> > >
> > > PMA
> > >
> > >
> > > On 4/27/07, Chol-Jun Yu <yucj@ghi.rwth-aachen.de> wrote:
> >> >> Hello,
> >> >>
> >> >> you can cited the following reference: Phys. Rev. Lett. 97,
> >> >> 066101(2006).
> >> >>
> >> >> Cheers.
> >> >>
> >> >> On Friday, 27. April 2007 13:01, fet@iisb.fraunhofer.de wrote:
> >>> >> > Hello Abinites,
> >>> >> >
> >>> >> > Has anybody tried to model amorphous materials and defects with a
> >> >> supercell
> >>> >> > before. I tried for a 10 layer supercell of Silicon dioxide ... it
> >> >> just
> >>> >> > does not seem to converge.
> >>> >> >
> >>> >> > Any help??
> >> >>
> >> >>
> >> >> Yu Chol Jun
> >> >> 
> -----------------------------------------------------------------------------
> >> >>
> >> >> Computational Materials Engineering(CME)
> >> >> Center for Computational Engineering Science(CCES)
> >> >> Institute of Mineral Engineering(GHI), RWTH Aachen
> >> >> Mauerstrasse 5, D-52064 Aachen, Germany
> >> >> Tel: ++49 241 80 94969
> >> >> Fax: ++49 241 80 92271
> >> >> e-mail: yucj@ghi.rwth-aachen.de
> >> >> 
> -----------------------------------------------------------------------------
> >> >>
> >> >>
> >> >>
> > >
> > >


--
Pierre-Matthieu Anglade