The ABINIT Users Mailing List ( CLOSED )

[verstraete@pcpm.ucl.ac.be]

The problem has already been raised several times in the forum (check the
archives). Mikami-san, could you put something on the FAQ, please?

The default CG line-minimization algorithm restarts if it finds a bad
curvature for the energy around its starting point. If it restarts 13
times (bad or slow convergence) the program stops. The 13 is arbitrary,
but indicates the problem is difficult (Xavier: perhaps an input variable 
to keep abinit from exiting because of this?).

Solution:
--------
Use iscf 3 (anderson mixing). It is slower but more robust, and never 
gives the "13 restart" error. I use it systematically without any 
difficulties or very noticeable slowdown w.r.t. iscf 5 (default).

cheers and happy gnu year.

Matthieu


On Mon, 5 Jan 2004, xianghjun wrote:

> I also have the same problem and wish somebody to help us.
> Best Regards
> xianghjun
> sefa dag wrote:
> 
> >
> >  Dear Sir/Madam,
> >
> >    I have a convergence problem. Just I am trying to do test 
> > calculations with  ABINIT program.
> > I want to look at the surface phonons on the C(100) surface. For this 
> > reason I am using  "t56.in" (from Tutorial)
> > input file  as template. I am using Troullier-Martins pseudopotential 
> > for C.
> > When I run abinit program, program stops and it gave this error message:
> >
> > scfcge : ERROR -
> >  Potential-based CG line minimization not converged after 13 restarts.
> >  Action : read the eventual warnings about lack of convergence.
> >  Some might be relevant. Otherwise, raise nband.
> >
> > leave_new : decision taken to exit ...
> >
> > I could not understand this. I increased number of bands. But it gave 
> > the same result. Also I am using enaugh number of k-points.
> > In the output file convergence
> > of energy seems okey
> > ...
> > ...
> > ...
> > ETOT 36  -71.813161986653     5.761E-09 1.618E-08 9.264E-02 9.874E-06 
> > 2.604E-02
> > ETOT 37  -71.813161982648     4.006E-09 6.845E-09 8.051E-02 5.358E-06 
> > 2.603E-02
> > ETOT 38  -71.813161980057     2.591E-09 2.972E-09 7.442E-02 3.124E-06 
> > 2.603E-02
> > ETOT 39  -71.813161978387     1.669E-09 1.319E-09 7.091E-02 1.917E-06 
> > 2.603E-02
> > ETOT 40  -71.813161977299     1.088E-09 5.976E-10 6.872E-02 1.220E-06 
> > 2.603E-02
> > ETOT 41  -71.813161976579     7.207E-10 2.753E-10 6.730E-02 7.978E-07 
> > 2.603E-02
> > ETOT 42  -71.813161976095     4.834E-10 1.281E-10 6.635E-02 5.321E-07 
> > 2.603E-02
> > ETOT 43  -71.813161975767     3.280E-10 5.974E-11 6.571E-02 3.599E-07 
> > 2.603E-02
> > ETOT 44  -71.813161975543     2.244E-10 3.026E-11 6.527E-02 2.459E-07 
> > 2.603E-02
> > ETOT 45  -71.813161975388     1.544E-10 1.506E-11 6.496E-02 1.692E-07 
> > 2.603E-02
> > ETOT 46  -71.813161975281     1.078E-10 7.314E-12 6.475E-02 1.170E-07 
> > 2.603E-02
> > ETOT 47  -71.813161975206     7.419E-11 3.475E-12 6.461E-02 8.123E-08 
> > 2.603E-02
> >
> > why it can not stop to do self-consistent calculation?
> >
> > In the input file there is tolvrs parameter. Tutorial says this must 
> > be between  e-18 and e-22.
> > i am taking:
> > tolvrs1   1.0d-18       # SCF stopping criterion
> > I also used toldfe parameter instead of this. It also gave the same 
> > error.
> >
> > What can I do about it? Could you please help me about it?
> >
> > Thank You
> >
> >
> 
> 
> 

-- 
===================================================================
ATTENTION!   PHONE NUMBER HAS CHANGED!!!    010 47 3359   !!!

Matthieu Verstraete               mailto:verstraete@pcpm.ucl.ac.be 
PCPM, Boltzmann, pl. Croix du Sud, 1            tel: 010/ 47 33 59
B-1348 Louvain-la-Neuve  Belgium                fax: 010/ 47 34 52