My Community

Hello all,



I am new to VASP
While doing volume optimization runs ,in the first step i did it only for 4-5 ionic steps.My INCAR file is given below



``
$System = Bi
NSW = 5
IBRION = 2 # ionic relaxation routine -1:no 0:MD 1:Newton(fast) 2:conj.gradient(safest)
ISIF = 3 # relax ions only
EDIFFG =-0.01
EDIFF =0.001
PREC = ACCU
ENCUT = 250
ISMEAR = 1 # 0=Gaussian smear.->insulators, 1=methfessel,sigma~0.2=default, -5=tetrah+Bloechl
SIGMA = 0.1
NBANDS = 44
LCHARGE=.TRUE
ICHARGE= 0

AT the end of a run,I am getting a CONTCAR file which is different from that of the initial POSCAR

My POSCAR is

Bi
1.000000
8.406531 0.000000 0.000000
0.000000 5.524531 0.000000
0.000000 0.000000 13.449531
6
Direct
0.364807 0.143680 0.353006
0.094422 0.856320 0.353006
0.905578 0.143680 0.353006
0.635193 0.856320 0.353006
0.635193 0.381166 0.646994
0.364807 0.618834 0.940982
~
My CONTCAR at the end of the run is

Bi
1.00000000000000
8.9388116459053215 -0.3175655038736505 0.0303374996553128
-0.2086949702793479 5.8013382724968094 -0.0161740147684171
0.0485361557125065 -0.0393763016359897 13.2308574575097655
6
Direct
0.3662336300554883 0.1520376945803200 0.3531530426511323
0.1193190415111828 0.8109312307885993 0.3532856422602222
0.8807730485593732 0.1892973420339173 0.3528872266937521
0.6338733533723985 0.8477736314077964 0.3533465507699455
0.6339599444891295 0.3820419411762139 0.6471735621987516
0.3658409820124277 0.6179181600131534 0.9401539754261962

0.00000000E+00 0.00000000E+00 0.00000000E+00
0.00000000E+00 0.00000000E+00 0.00000000E+00
0.00000000E+00 0.00000000E+00 0.00000000E+00
0.00000000E+00 0.00000000E+00 0.00000000E+00
0.00000000E+00 0.00000000E+00 0.00000000E+00
0.00000000E+00 0.00000000E+00 0.00000000E+00
~
The unit cell vectors in cartesian coordinate do not match
is it OK or i am doing something wrong

Thanking you in advance
[blockquote]Subhadip Das
junior research fellow
National Chemical Laboratory
Pune
Maharashtra
INDIA

The ISIF = 3 option allows for relaxation of both volume and shape. The basis vectors can therefore change if your equilibrium geometry isn't stable and/or if the stress tensor is inaccurate.

So, the first option is to contrain the relaxation by using another ISIF option, if you don't want the basis vectors to change (see the manual for an appropriate value). However, note that if your geometry isn't elastically (or dynamically) stable you should think about if your supplied geometry really is what you want to investigate.

For the accuracy of the stress tensor (which is used to relax the cell), you must in general consider the presence of Pulay stresses:

http://cms.mpi.univie.ac.at/vasp/vasp/V ... tress.html



Cheers,
/Dan

<span class='smallblacktext'>[ Edited Thu Jul 22 2010, 10:20PM ]</span>

Then how will I ensure that my basis set is converged with respect to changes in volume if I am doing volume optimization runs?

You perform convergence tests with respect to ENCUT ( and the number of k-points if you haven't done that already) and investigate how the equilibrium structure will look like. An increase of ENCUT with at least 20-30% w.r.t. to the default value is usually required. Please also use PREC =Accurate for the cell relaxation.

An alternative to doing the volume relation directly is also to do many fixed volume calculations. For further details, please see

http://cms.mpi.univie.ac.at/vasp/vasp/A ... s_two.html


Cheers,
/Dan



<span class='smallblacktext'>[ Edited Fri Jul 23 2010, 10:55AM ]</span>