[CPMD-list] Why convergence is so slow?

[Juerg Hutter]

Hi

the only improvement I can see right away would be
to use the LBFGS optimizer. This usually works
better for this type of systems.

regards

Juerg Hutter

----------------------------------------------------------
Juerg Hutter                   Phone : ++41 44 635 4491
Physical Chemistry Institute   FAX   : ++41 44 635 6838
University of Zurich           E-mail: hutter at pci.unizh.ch
Winterthurerstrasse 190
CH-8057 Zurich, Switzerland
----------------------------------------------------------


On Fri, 2 Dec 2005, Yang Zhenyu wrote:

> Dear CPMD users:
>
>      I try to calcluate a molecule adsorption on Ni surface with GH-pseudopotentials, but I find it is very hard to converge in the geometry optimization. So I post my input here, hoping anybody can give me some advices!
>
> -------------------------------
> &CPMD
>    OPTIMIZE GEOMETRY
>    TRAJECTORY XYZ
>    UNIT HESSIAN
>    BFGS
>    FREE ENERGY FUNCTIONAL
>    LANCZOS DIAGONALISATION
>    LANCZOS PARAMETERS
>      1   6 10   1.D-18
>    TROTTER FACTOR
>      0.001
>    BOGOLIUBOV CORRECTION OFF
>    GRAM-SCHMIDT ORTHOGONALISATION
>    CONVERGENCE
>      1.D-4  5.D-4
>    MAXSTEP
>      500
>    BROYDEN MIXING
>      0.3 200   0.01  0   8
>    ALEXANDER MIXING
>      1.1
>    ELECTRON TEMPERATURE
>      1000.
>    COMPRESS WRITE32
>    STRUCTURE BONDS
>    RHOOUT
>    WANNIER WFNOUT ALL
>    PRINT ON
> &END
>
> &SYSTEM
> POINT GROUP
> AUTO
> SYMMETRY
>  4
> ANGSTROM
> CELL
> 7.3818     1.0     5.0       0.00       0.00       -0.5
> CHARGE
> 0
> CUTOFF
> 60
> TESR
> 3
> &END
>
> &DFT
> NEWCODE
> FUNCTIONAL PBE
> &END
>
> &ATOMS
> *Ni-q18-pbe KLEINMAN-BYLANDER
> LMAX=P
> .....................................
>
> *N-q5-pbe KLEINMAN-BYLANDER
> LMAX=P
> .....................................
>
> *C-q4-pbe KLEINMAN-BYLANDER
> LMAX=P
> .....................................
>
> *H-q1-pbe KLEINMAN-BYLANDER
> .....................................
>   CONSTRAINTS
>   FIX ATOMS
>    8
>    1  2  3  4  5  6  7  8
>  END CONSTRAINTS
> &END
>
> ---------------------------------------------------
>
>
>
>
> ¡¡¡¡¡¡¡¡¡¡¡¡¡¡
> ¡¡Best Regards
>
> ¡¡Yang Zhenyu
>  State Key Laboratory of Nonlinear Mechanics (LNM)
>  Institute of Mechanics, Chinese Academy of Sciences
>  No.15 Beisihuanxi Road
>  Beijing 100080, P. R. China
>  FAX: 010-62579511
> ¡¡yangzy at lnm.imech.ac.cn
> ¡¡2005-12-02
>
> _______________________________________________
> CPMD-list mailing list
> CPMD-list at cpmd.org
> http://cpmd.org/mailman/listinfo/cpmd-list
>