[CPMD-list] Cell Size
Juerg Hutter
hutter at pci.unizh.ch
Tue Mar 13 08:35:10 CET 2007
Hi
> Can anybody tell me the following things.
> 1. how to choose cell size for different clusters like Si64, Si256, Si512. I
> used CPMD compatible *Troullier*-*Martins* *pseudopotential*.
I would start with a cell that adds about 5 Angstrom to the diameter of
your cluster. Use the POISSON SOLVER TUCKERMAN together with
SYMMETRY 0. If you see problems (e.g. very slow or no convergence, strange
bond length etc) or you are at the limit of the CPU time available,
you might want to increase or decrease the cell to find the optimal
values for your needs.
> 2. Choosing energy cutoff
General rule: each atom (i.e. pseudopotential) needs a certain cutoff.
You can test for this cutoff by doing a small molecule
calculation and changing the cutoff while monitoring a
property of interest.
The calculation of your final system will need the
highest cutoff of all different atom types involved.
For a pure silicon system I would suggest a cutoff between 15 and 25
Rydberg. If you use a TM PP and the Kleinmann-Bylander scheme you
should have LMAX=D LOC=D
> 3 . Which optimization technique is efficient for Geometry Optimization.
This worked fine for me
LBFGS
CONVERGENCE ORBITALS
1.0e-5
CONVERGENCE ADAPT
0.02
CONVERGENCE ENERGY
0.05
RANDOMIZE COORDINATES
0.00001
PRINT LSCALE ON
PRINT FORCES ON
best regards
Juerg Hutter
>
> If any mistakes in my input file please suggest me.
>
>
>
> Here I'm attaching Input file for Si64
>
>
> &INFO
> Date : 09FEB2007,13:30
> WAVE FUNCTION Optimization
> &END
> &CPMD
> BFGS
> CENTER MOLECULE ON
> oPTIMIZE WAVEFUNCTION
> OPTIMIZE GEOMETRY
> ORTHOGONALIZATION LOWDIN
> RESTART WAVEFUNCTION COORDINATES LATEST
> rESTART ALL
> kOHN-SHAM ENERGIES
> 192
> CONVERGENCE ORBITALS
> 1.0D-7
> MAXSTEP
> 1000
> SPLINE POINTS
> 1000
> &END
>
> &DFT
> NEWCODE
> FUNCTIONAL LDA
> &END
>
> &SYSTEM
> ANGSTROM
> SYMMETRY
> 0
> CELL
> 28.000 1.0 1.0 0 0 0
> CUTOFF
> 17.56
> pOISSON SOLVER TUCKERMAN
> &END
>
> &ATOMS
> *SI.OUT1 KLEIMAN-BYLANDER
> LMAX=P LOC=S
> 64
> 0.0000 0.0000 0.0000
> 2.7150 0.0000 2.7150
> 2.7150 2.7150 0.0000
> 0.0000 2.7150 2.7150
> 1.3575 1.3575 4.0725
> 1.3575 4.0725 1.3575
> 4.0725 1.3575 1.3575
> 4.0725 4.0725 4.0725
> 5.4300 0.0000 0.0000
> 8.1450 0.0000 2.7150
> 8.1450 2.7150 0.0000
> 5.4300 2.7150 2.7150
> 6.7875 1.3575 4.0725
> 6.7875 4.0725 1.3575
> 9.5025 1.3575 1.3575
> 9.5025 4.0725 4.0725
> 0.0000 5.4300 0.0000
> 2.7150 5.4300 2.7150
> 2.7150 8.1450 0.0000
> 0.0000 8.1450 2.7150
> 1.3575 6.7875 4.0725
> 1.3575 9.5025 1.3575
> 4.0725 6.7875 1.3575
> 4.0725 9.5025 4.0725
> 0.0000 0.0000 5.4300
> 2.7150 0.0000 8.1450
> 2.7150 2.7150 5.4300
> 0.0000 2.7150 8.1450
> 1.3575 1.3575 9.5025
> 1.3575 4.0725 6.7875
> 4.0725 1.3575 6.7875
> 4.0725 4.0725 9.5025
> 0.0000 5.4300 5.4300
> 2.7150 5.4300 8.1450
> 2.7150 8.1450 5.4300
> 0.0000 8.1450 8.1450
> 1.3575 6.7875 9.5025
> 1.3575 9.5025 6.7875
> 4.0725 6.7875 6.7875
> 4.0725 9.5025 9.5025
> 5.4300 0.0000 5.4300
> 8.1450 0.0000 8.1450
> 8.1450 2.7150 5.4300
> 5.4300 2.7150 8.1450
> 6.7875 1.3575 9.5025
> 6.7875 4.0725 6.7875
> 9.5025 1.3575 6.7875
> 9.5025 4.0725 9.5025
> 5.4300 5.4300 0.0000
> 8.1450 5.4300 2.7150
> 8.1450 8.1450 0.0000
> 5.4300 8.1450 2.7150
> 6.7875 6.7875 4.0725
> 6.7875 9.5025 1.3575
> 9.5025 6.7875 1.3575
> 9.5025 9.5025 4.0725
> 5.4300 5.4300 5.4300
> 8.1450 5.4300 8.1450
> 8.1450 8.1450 5.4300
> 5.4300 8.1450 8.1450
> 6.7875 6.7875 9.5025
> 6.7875 9.5025 6.7875
> 9.5025 6.7875 6.7875
> 9.5025 9.5025 9.5025
> &END
>
>
> Thanks in Advance
>
> --
> SHASHIDHAR . G
> Project Intern,
> L 2 Cluster, Ist Floor,
> Centre for Development of Advanced Computing(CDAC)-Pune,
> Pune University Campus,
> Ganesh Khind,
> Pune-411 007 . INDIA.
> Phone- (020) 25694083/ 25704228
>
More information about the CPMD-list
mailing list