The ABINIT Users Mailing List ( CLOSED )

[Xavier Gonze <gonze@pcpm.ucl.ac.be>]

Dear Hingyi Zhao,

Well, Pierre-Matthieu's answer reflects a good practice, but ABINIT
will accept you to use tolmxf combined with toldfe or tolvrs or  
tolwfr ...
Note also, the very recent new input variable (from Matthieu  
Verstraete),
tolrff , that might even be better than toldff when combined with  
tolmxf.

Best regards,
Xavier


On 16 Apr 2007, at 20:19, Anglade Pierre-Matthieu wrote:

> The answer is NO :
> tolmxf tells when the optimisation of your positions will stop.
> Toldff tells Abinit when a SCF calculation is considered converged.
> Basically you computation is performed the following way :
> SCF optimisation 1 -> stopped at toldff
> atoms movement 1
> SCF optiimisation 2 -> stopped at toldff
> atoms movement 2
> ...
> SCF optimisation N -> stopped at toldff
> The forces are below tolmxf -> stop the computation.
>
>
> On 4/16/07, Hongyi Zhao <hongyi.zhao@gmail.com> wrote:
> > Hi all,
> >
> > Now, I do a structure relaxation of H_2 molecular in order to find  
> > its
> > optimum interatomic distance, the following its my input file:
> >
> > -----------
> > ionmov  3
> > ntime   10
> > tolmxf  5.0d-4
> > xcart  -0.7   0.0 0.0
> >         0.7   0.0 0.0
> >
> >         toldff  5.0d-5
> >
> >         nband    1
> >         acell 10 10 10
> >         ntypat 1
> >         znucl 1
> >
> >         natom 2
> >         typat 1 1
> >         ecut 10.0
> >
> >
> >         nkpt 1
> >         nstep 10
> >         diemac 1.0
> >         diemix 0.5
> > -----------
> >
> > My question is: can i only use the parameter *tolmxf* without the
> > parameter *toldff* to do this thing?
> >
> > Thanks in advance.
> >
> > ---
> > Hongyi Zhao
> > GnuPG DSA: 0xD108493B
>
>
> --
> Pierre-Matthieu Anglade