The ABINIT Users Mailing List ( CLOSED )

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

Hello, good morning ("Alo, Gunaydm" in Turkish),

I cannot catch your point without inputs/outputs
(Netiquette: http://www.abinit.org/community/?text=netiquette),
so my response may be vague...

It depends. If your subgroup has a loose (not rigid) structure,
the tolerance needed for convergence may be loosed than usual.
(e.g. systems with molecular fragments, etc., namely,
 with low phonon frequencies)
You might want to monitor the total energy.
The fluctuation of the energy is acceptable or not ?
If the system has such loose structure, the total energy
may fluctuate with such a small amplitude that it may be
more difficult to get it converged...

Good continuation,
Hosca kalm,
Masayoshi

P.S. In passing, after reading some recent posts,
I just thought that the following FAQ may be remembered:

- What can ABINIT do ?
http://www.abinit.org/about/?text=../Infos/FAQ#cando

- What cannot ABINIT do ?
http://www.abinit.org/about/?text=../Infos/FAQ#cannot

On 2005/02/21, at 16:57, e120231@metu.edu.tr wrote:

> Hello,
>
> I came across with a convergence problem due to tolmxf. I am working 
> on a CsCl (221) structure and I relaxed subgroups of this structure. I 
> was expecting that the subgroups should converge to CsCl structure and 
> they did. Abinit gives the expected results when the gradients 
> (tolmxf) are lowest (but not the lowest energy). However the value of 
> tolmxf is nearly 5.00E-4,ten times the default value. Is such a force 
> acceptable?
>
> Thank you
>
> Metallurgical and Materials Eng. Dept.
> METU
> Turkey