[CPMD-list] optimization of quartz

[Martijn Zwijnenburg]

Hi,

I'm trying to optimize a unit-cell of quartz (see input-file below) and 
everything seems to work fine. Only minor point is that it seems to 
require a massive ammount of geometry steps before the 
calculation converges (>300). So I was wondering if there are 
alternatives to STEEPEST DESCENT CELL for cell optimization?

Furthermore, I was wondering what are "decent" convergence 
criteria for cell optimization and what is their relation to the stress 
tensor shown in the output? Maybe, mine are to "harsh".

Cheers,

Martijn

P.S. The speed of an individual geometry step is not a problem 
(roughly 80 geometry steps a day).

! Optimization of alpha-quartz unit cell
!
&CPMD
  OPTIMIZE GEOMETRY
  BFGS
  STEEPEST DESCENT CELL
  HESSIAN UNIT
  LANCZOS DIAGONALISATION 
  LANCZOS PARAMETERS
  1 6 10 1.D-8
  STRUCTURE BONDS ANGLES DIHEDRALS
  SPLINE RANGE
  5.00
  SPLINE POINTS
  2500
  CONVERGENCE
  1E-6 5E-6 1. 
  RESTART WAVEFUNCTION COORDINATES CELL HESSIAN LATEST
&END
&SYSTEM
  SYMMETRY
  14
  CELL
  4.91000 1.0 1.10002 0 0 -0.5
  ANGSTROM
  CUTOFF
  80.00
  STATES
  36
  POINT GROUP
  AUTO
  KPOINTS MONKHORST-PACK
  2 2 2
&END
&ATOMS
*Si_SG_BLYP KLEINMAN-BYLANDER
  LMAX=P
 3
 2.0440  1.1800  0.0000
 2.2080 -1.2750  3.6010
 0.0000  2.5490  1.8010

*O_SG_BLYP KLEINMAN-BYLANDER
  LMAX=P
 6 
 0.6390  1.6670  0.6290
 2.4890  2.1750  4.2310
 1.1240  3.5230  2.4300
 1.7630 -0.2800  4.7730
 3.1280  1.0680  1.1710
 3.6140 -0.7880  2.9720
&END
&DFT
 FUNCTIONAL BLYP
&END

-------------------------------------------------------------------------
Martijn Zwijnenburg
Lab. of Applied Organic Chemistry and Catalysis
Delft University of Technology
Julianalaan 136
2628 BL Delft
The Netherlands
Tel: 0031-(0)152782691
Fax: 0031-(0)152784700
e-mail: M.A.Zwijnenburg at tnw.tudelft.nl
web page: http://come.to/tock