FAST calculation
Posted: Mon Nov 29, 2010 4:07 pm
Dear Folks,
I am novice to periodic structure calculations and i have the following doubt.
I have a large system (72 atom surface and 14 atoms for molecules) and trying to geometry optimize it. I use PAW_PBE pseudopotenial and the following INCAR (see below) and 1X1X1 Kpoint centered on gamma.
The calculation take long time for convergence and sometime stops in the middle (don't worry about this., because the difference in energy between the two consequent ionic steps is too high and causes this problem).
I am looking for your suggestions that could help me to reduce the computational time. (For example doing calculations using a double zeta basis set when we use basis sets followed by triple zeta quality for the selected structures),
Is there any strategies that you can suggest me to do a similar rough calculation using periodic boundary conditions before using a standardized set to compare different orientation of molecules adsorbed on the surface in a very fast way.
I know reducing the Energy cutoff and pseudopotential can reduce the time. but what would you suggest instead of PAW_PBE pseudopotenial. Also what should i setup for IALGO,
IBRION tags etc.,
My INCAR is
System =
! 0=new, 1=wave (d), 2=same cutoff&latt (1&2 need WAVECAR file)
ISTART=1
! 0=wave (d), 1=CHGCAR
ICHARG=1
NPAR = 8
LPLANE = .TRUE.
ENCUT= 400.00 eV
ENAUG= 700.00 eV
LREAL = auto
PREC = Accurate
ALGO = Fast
NSIM =5
NGX = 99
NGY = 99
NGZ = 201
ENMAX=400
NBANDS = 600
EDIFF=1e-5 ! converge force better
ISMEAR=0
SIGMA=0.01
ISPIN=2
NUPDOWN=0
NSW=220 !!!! Do not relax for now !!!
IBRION = 2 ! 1: quasi-NR, 2: CG algorithm for ions; use 2 for EDIFFG=-0.05
LWAVE=.TRUE.
LCHARG=.TRUE.
Even a little suggestion could be really useful.
If you need more information please let me know.
Best regards,
K
I am novice to periodic structure calculations and i have the following doubt.
I have a large system (72 atom surface and 14 atoms for molecules) and trying to geometry optimize it. I use PAW_PBE pseudopotenial and the following INCAR (see below) and 1X1X1 Kpoint centered on gamma.
The calculation take long time for convergence and sometime stops in the middle (don't worry about this., because the difference in energy between the two consequent ionic steps is too high and causes this problem).
I am looking for your suggestions that could help me to reduce the computational time. (For example doing calculations using a double zeta basis set when we use basis sets followed by triple zeta quality for the selected structures),
Is there any strategies that you can suggest me to do a similar rough calculation using periodic boundary conditions before using a standardized set to compare different orientation of molecules adsorbed on the surface in a very fast way.
I know reducing the Energy cutoff and pseudopotential can reduce the time. but what would you suggest instead of PAW_PBE pseudopotenial. Also what should i setup for IALGO,
IBRION tags etc.,
My INCAR is
System =
! 0=new, 1=wave (d), 2=same cutoff&latt (1&2 need WAVECAR file)
ISTART=1
! 0=wave (d), 1=CHGCAR
ICHARG=1
NPAR = 8
LPLANE = .TRUE.
ENCUT= 400.00 eV
ENAUG= 700.00 eV
LREAL = auto
PREC = Accurate
ALGO = Fast
NSIM =5
NGX = 99
NGY = 99
NGZ = 201
ENMAX=400
NBANDS = 600
EDIFF=1e-5 ! converge force better
ISMEAR=0
SIGMA=0.01
ISPIN=2
NUPDOWN=0
NSW=220 !!!! Do not relax for now !!!
IBRION = 2 ! 1: quasi-NR, 2: CG algorithm for ions; use 2 for EDIFFG=-0.05
LWAVE=.TRUE.
LCHARG=.TRUE.
Even a little suggestion could be really useful.
If you need more information please let me know.
Best regards,
K