The ABINIT Users Mailing List ( CLOSED )

[Masayoshi Mikami <mmikami@rc.m-kagaku.co.jp>]

Dear "XD0553" (Sir/Madam),

I do not know what kind of structure you are treating,
but my guess is the followings:

1) You might have forgotten "angstrom"  in "acell" line ?
    acell 10.1214 10.4105 9.7661 angstrom

2) In your xred
        xred
        2.5E-01     7.6511E-01   5.0E-01  # Gd
        2.5E-01    3.08600E-01   -9.8764E-019 # Ta
        1.03100E-01  8.6200E-02   2.53900E-01 # O
  It seems that distance between Gd and Ta seems too close,
  so that the total energy can be way too positive...

  Try drawing your crystal structure with an appropriate tool...
  (I have checked your structure with "CrystalMaker")

 I suspect that "-9.8764E-019" (z of #Ta) might be wrong,
because experimental XRD analysis will not produce such
a small value...

Regards,
Masayoshi

On 2007/05/14, at 18:02, xd0533@163.com wrote:

> dear forum,
>      The etotal I get is  positive and it always doesn't converge.  
> The pseudopotentials I use is FHI(GGA). I don't know the reason.The  
> input file is as follows:
>  ndtset 7
> #Definition of the unit cell and ecut,
> #for which one will have to make a convergence study
>  acell 10.1214 10.4105 9.7661
>  angdeg 90 96.69 90
>  ecut: 20    ecut+ 5
>
> #First dataset :
> spgroup 13
>  natom 12
>  natrd  3
>  xred
> 2.5E-01     7.6511E-01   5.0E-01
> 2.5E-01    3.08600E-01   -9.8764E-019
> 1.03100E-01  8.6200E-02   2.53900E-01
> #Definition of the atom types
> ntypat 3
> znucl 64 73 8
> #Definition of the atoms
> typat 1 1 2 2 3 3 3 3 3 3 3 3
> #Definition of the k-point grid
> kptopt  1
> nshiftk 1
> shiftk  0.5 0.5 0.5  # These shifts will be the same for all grids
> ngkpt  4 4 4                   # Only one k point is needed for  
> isolated system,
>                            # Only one k point is needed  
> for                # ixc  13
> #Definition of the SCF procedure
> nstep 200          # Maximal number of SCF cycles
> toldfe 1.0d-5
> diemac 4.0        # Although this is not mandatory, it is worth to
>                   # precondition the SCF cycle. The model dielectric
> diemix 1.0        # function used as the standard preconditioner
> I hope your reply.
>                                                             best  
> wishes
>    xd