The ABINIT Users Mailing List ( CLOSED )

["matthieu verstraete" <matthieu.jean.verstraete@gmail.com>]

Hello 张�s

no simple answer yet, but several things are strange about your relaxation:

1) the energy goes up after the 6th Broyden step, and you end with a configuration that is oval, and has a higher energy than the initial state
2) the scf energy in the next run is even higher!

The explanation of maxfor is as follows:

The forces themselves are small but the value you get in the last column of the ETOT lines comes from the stresses. In particular the one along y becomes "less small". The stresses are not taken into account in the calculation of maxfor in the relaxation, because you are keeping the unit cell fixed (abinit shows the value which has to be optimized, not the other terms, eg if you keep the cell or some atoms fixed).

In the SCF, all forces and stresses (with a conversion factor strfact of 100) are included in the calculation of maxfor : if your maximum stress is 1e-4 then you'll get a maxfor of 1e-2.

This still does not explain why the energy goes up: the occupations are reset to 2 2 2 0 0 at each kpoint, but should be relaxing back (you are not reading the wf back in as a starting point?). During the Broyden, one can imagine that the algorithm overshot and arrived in a local minimum, favored by the cell which is perhaps too small. What worries me more is that the energy in the scf is not the same as at the end of the relaxation.

How is the pseudopotential? Have you tested it on diamond or graphene, or C_2?

This may also be a real bug, and xlf is notoriously painful for revealing it's own set of bugs.

Matthieu

On Thu, Sep 18, 2008 at 6:42 AM, 张�s <zhangting1980323@gmail.com> wrote:

Dear abinit users:

      I'm optimizing the geometry structure of CNT. For calculating the phonon perturbation, I need the fully relaxed structure. But the already optimized structrue(with tolmxf =1.0d-7) is calculated to be with about 3e-2 Ha/Bohr's max force in the next SCF jdset in phonon calculations. Why these two forces do not consist with each other?  Inclosed in my input and out files of optimization and SCF tasks. What's wrong within my calculation settings? Thanks!


Regards

                                                                               
                                                                                         Zhang Ting
                                                                                         Peking Univ.
                                                                                         Sep. 18th, 2008

--
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Dr. Matthieu Verstraete

European Theoretical Spectroscopy Facility (ETSF) 
Dpto. Fisica de Materiales, 
U. del Pais Vasco,            
Centro Joxe Mari Korta, Av. de Tolosa, 72, Phone: +34-943018393
E-20018 Donostia-San Sebastian, Spain Fax  : +34-943018390

Mail : matthieu.jean.verstraete@gmail.com
http://www-users.york.ac.uk/~mjv500