[CPMD-list] =?GB2312?B?SG93IHRvIHBpY2sgYSBwcm9wZXIgQ09OVkVSR0VOQ0UgT1JCSVRBTCCjvw==?=

[Axel Kohlmeyer]

On Sun, 13 May 2007, liu liuxiandong wrote:

XL> Dear CPMDers,
XL> 
XL> When I run a OPTIMIZE GEOMETRY, I find that the CONVERGENCE GEOMETRY has
XL> been reached and the programme terminates normally before the ORBITAL
XL> converges to 1.0D-6.
XL> My system is a solid-water interface system containing 76 atoms in total.
XL> Is the convergence of orbital 5D-6 is "tight" enough for such systems ?
XL> And how to pick this tolerance properly ?

the usual rule of the thumb is that the convergence for the
wavefunction (orbitals) should be about two orders of magnitude
tighter than the geometry convergence to get reliable results.

to get confirmation on the correctness of your result you 
could restart without adaptive convergence and set a tight
wavefunction convergence parameter (with adaptive wavefunction
convergence, you can do this regardless...).

cheers,
 axel.

XL> 
XL> Thanks for your suggestions.
XL> 
XL> Xiandong
XL> 
XL> 
XL> My input is as follows,
XL> 
XL> 
XL>  RESTART WAVEFUNCTIONS COORDINATES HESSIAN
XL>   OPTIMIZE GEOMETRY XYZ
XL>   HESSIAN UNIT
XL> 
XL>   CONVERGENCE ADAPT
XL>   0.02
XL>   CONVERGENCE INITIAL
XL>   5.0D-6
XL>   CONVERGENCE ORBITALS
XL>   1.0D-6
XL>   CONVERGENCE GEOMETRY
XL>   2.0D-4
XL> 
XL>     LBFGS NREM
XL>       20
XL> 
XL> 
XL> &END
XL> 
XL> &DFT
XL>   FUNCTIONAL BLYP
XL> 
XL> &END
XL> 
XL> &SYSTEM
XL> 
XL>   SYMMETRY
XL>    14
XL>   ANGSTROM
XL>   CELL ABSOLUTE DEGREE
XL>    21.7780 10.3365 11.5168 90.0000 90.0000 90.0000
XL>   CUTOFF
XL>    100.0
XL> &END
XL> 
XL> &ATOMS
XL> *O_MT_BLYP.psp KLEINMAN-BYLANDER
XL>  LMAX=P
XL>  38
XL> 
XL> *H_MT_BLYP.psp KLEINMAN-BYLANDER
XL>  LMAX=S
XL>  20
XL> 
XL> *Si_MT_BLYP.psp KLEINMAN-BYLANDER
XL>  LMAX=D
XL>  8
XL> 
XL> *Al_MT_GIA_BLYP.psp KLEINMAN-BYLANDER
XL>  LMAX=D LOC=0
XL>  8
XL> 
XL> &END
XL> 

-- 
=======================================================================
Axel Kohlmeyer   akohlmey at cmm.chem.upenn.edu   http://www.cmm.upenn.edu
   Center for Molecular Modeling   --   University of Pennsylvania
Department of Chemistry, 231 S.34th Street, Philadelphia, PA 19104-6323
tel: 1-215-898-1582,  fax: 1-215-573-6233,  office-tel: 1-215-898-5425
=======================================================================
If you make something idiot-proof, the universe creates a better idiot.