The ABINIT Users Mailing List ( CLOSED )

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

1) you may try ISCF=7 which sometimes is better than iscf=5. Anyway
both are generaly better than iscf 3.
2) if your total energy is not already nicely enough  converged at
step 5 (max resid~1e-4) you setting for iprcel is not optimal and you
might get better results with iprcel=49. If this is still not enough
you can change the calling condition of the preconditioner in
prcref.F90: for instance you may want to calculate the RPA dielectric
matrix three time instead of two at step 1, (iprcel-40) and
2*(iprcel-40)
3) if your cell is large enough it is useful to increase nline and
nnsclo. (check the differences if any on a single run)
4) you might benefit from an increase of tsmear at start.
5) if your computation stops after a broyden step. You can always
restart with the new positions of the atoms...


1) Your kpt mesh looks a bit weird. i.e. it is possible that you may
get better precision with less kpt using a few more point int the z
direction and a few less in  x and y. Did you looked at the evolution
of kptrlen with respect to this parameters? I guess that ngkpt 8 8 1
won't change at all your kptrlen while ngkpt 8 8 2 will makes an
interesting increase.


On 3/31/06, ha sa <ha_sadi@yahoo.com> wrote:
>
> Dear abinit users,
>  I'm doing slab calculations for a supercell of seven atoms of iridium and
> seven layers of vacuum, the problem is that the gradient(force/stress) is
> converging very slowly, I reached the ninth Broyden step and the gradient
> still of the order of 10^-3.
>    Is there any way that can be used to speed up the gradient convergence.
>
> Another question:
> is there an autosave feature in abinit? actually I face a problem that
> sometimes I run some jobs and they keep running for days then my computer
> stops responding or the run exits for some reason and I lost all the
> information, then when I want to rerun it I start from zero. Is there any
> way to save information every 60 minutes for example, from which the program
> can continue.
>
> Regarding the first question I attached my input file.
>
> Thanks in advance.
> Hanan
>
>
>
> #Definition of the unit cell
> acell  3*7.29
>  rprim 0.5 -0.5  0.0
>        0.5  0.5  0.0
>        0.0  0.0  7.0
> iprcel 45
>
> #Definition of occupation numbers
> occopt 4
> tsmear 0.02
>
> #Definition of the k-point grid
> kptopt 1
> ngkpt  12 12 1
> nshiftk 2
> shiftk  0.5 0.0 0.0
>         0.0 0.5 0.0
>
> #Definition of the atom types
> ntypat 1
> znucl 77
>
> #Definition of the atoms
> natom 7
> typat 7*1
> xcart
>      2*0 0
>      0 2*3.645
>      2*0 7.29
>      0 3.645 10.935
>      2*0 14.58
>      0 3.645 18.225
>      2*0 21.87
> #Exchange-correlation functional
> ixc 1           # LDA hgh
>
> #Definition of the planewave basis set
> ecut  30.0         # Maximal kinetic energy cut-off, in Hartree
>
> #The relaxation
> ionmov 3
> tolmxf 5.0d-5
> ntime 100
>
> #Definition of the SCF procedure
> iscf 3
> nstep 100          # Maximal number of SCF cycles
> toldff 5.0d-6
>
>
>
>  ________________________________
> Yahoo! Messenger with Voice. Make PC-to-Phone Calls to the US (and 30+
> countries) for 2¢/min or less.
>
>


--
Pierre-Matthieu Anglade