question about simulated annealing

Message

wenxuan · #1 Post by **wenxuan** » Thu Aug 20, 2009 8:59 am

dear friends

I want to carry out simulated annealing for geometry optimization of an oxide. I want to heating up to 1500K from 1030k in 10000 steps. My INCAR file was set as follows:

ENCUT = 450 eV
EDIFFG = 0
EDIFF = 1E-7

NSW = 10000
IBRION = 0
POTIM = 2.0
SMASS = 2
TEBEG = 1030
TEEND = 1500

NPACO = 200
APACO = 20.0

IALGO = 8
LREAL = .TRUE.

NBLOCK = 1
KBLOCK = 200

ISTART = 1
ICHARG = 1

the following is my OSZICAR:
1 T= 1018. E= -.52866716E+02 F= -.53656031E+02 E0= -.53650722E+02 EK= 0.78932E+00 SP= 0.00E+00 SK= 0.27E-07
2 T= 992. E= -.52866644E+02 F= -.53635723E+02 E0= -.53630124E+02 EK= 0.76910E+00 SP= -.18E-04 SK= 0.72E-06
3 T= 961. E= -.52865170E+02 F= -.53610487E+02 E0= -.53604542E+02 EK= 0.74541E+00 SP= -.94E-04 SK= 0.53E-05
4 T= 928. E= -.52865522E+02 F= -.53584687E+02 E0= -.53578383E+02 EK= 0.71942E+00 SP= -.27E-03 SK= 0.21E-04
5 T= 893. E= -.52866168E+02 F= -.53558237E+02 E0= -.53551610E+02 EK= 0.69261E+00 SP= -.60E-03 SK= 0.61E-04

i have two question:
The parameters in my INCAR are right or wrong?
Why the temperature "T" in OSZICAR decrease rather than increase?

aron · #2 Post by **aron** » Thu Aug 20, 2009 3:08 pm

The simulation is not coupled to the heat bath at every step, so it may drift down at the beginning if it was not equilibrated to 1030 K (e.g. starting from a 0K structure). The coupling frequencey is determined by your SMASS paramater: http://cms.mpi.univie.ac.at/vasp/guide/node118.html

aron · #3 Post by **aron** » Thu Aug 20, 2009 3:12 pm

One more thing: EDIFF = 1E-7 is incredibly high convergence for an MD calculation. Even for a small system, it would take a very long time to reach 10000 steps.

wenxuan · #4 Post by **wenxuan** » Tue Aug 25, 2009 12:44 am

Thanks very much.
Could you please suggest me what changes I should make in the INCAR file so that I can achieve my desired annealing? How long does it take to heat the system from 1030 to 1500?