About the stress data in OUTCAR file

[TING QIN]

Hello I am carrying out elastic moduli calculation for bcc fcc and sc structures for tungsten. I use strain tensor whose components are also called true strains to generate new POSCAR files from the relaxed POSCAR and get the stress data from OUTCAR. However, I found that the value of C44 is almost two times of the experimental value. We know that the off-diagonal components of the engineering strain matrix are two times of the off-diagonal components of the strain tensor. So if I use engineering strains instead of true strains to calculate the elastic moduli, then the elastic moduli data agree well with the data published and experimental data. So does vasp print the stresses as (sigma_xy+ sigma_yx) or (sigma_xy+ sigma_yx)/2 in the OUTCAR files?Thanks very much.

[admin]

In OUTCAR, simply the xy, yz and zx components of the force stress tensor are written.
When using stress tensor data, please be aware that the calculations have to be done with the highest possible precision (PREC=Accurate, ENCUT should be set explicitely in addition to AT LEAST 1.3* max(ENMAX) of the PPs. Otherwise, the calculated data are not reliable. Please check if you have done so

[shurab]

Is it in VASP 5.1 with IBRION=6 and ISIF>=3?