[Pw_forum] About stress calculation.

[Stefano de Gironcoli]

stress calculation converge very slowly with cutoff.
before taking seriously differences of small numbers you should check 
carefully convergence.

stefano

Dariusz Chrobak wrote:

> Dear users,
>
> I have a problem with stress calculation for Ni2MnGa (L21 structure). 
> I am using pseudopotential from PWscf site.
> The tetragonal distorsions  c/a=0.995 and c/a=1.005 
> (epsilon_xx=epsilon_yy=e, epsilon_zz=-2e where c/a=(1-2e)/(1+e)) were 
> aplied to the regular cell with a0=10.9718 a.u..
>
> The stress calculation performed for these cases gave me the folowing 
> results:
>
> 1. c/a=0.995
>          total   stress  (ryd/bohr**3)                  (kbar)     
> P=    0.00
>   0.00000328   0.00000000   0.00000000          0.48      0.00      0.00
>   0.00000000   0.00000328   0.00000000          0.00      0.48      0.00
>   0.00000000   0.00000000  -0.00000656          0.00      0.00     -0.97
> 2. c/a=1
>          total   stress  (ryd/bohr**3)                  (kbar)     
> P=    0.00
>  -0.00000003   0.00000000   0.00000000          0.00      0.00      0.00
>   0.00000000  -0.00000003   0.00000000          0.00      0.00      0.00
>   0.00000000   0.00000000  -0.00000003          0.00      0.00      0.00
> 3. ca=1.005
>          total   stress  (ryd/bohr**3)                  (kbar)     
> P=   -0.22
>   0.00000093   0.00000000   0.00000000          0.14      0.00      0.00
>   0.00000000   0.00000093   0.00000000          0.00      0.14      0.00
>   0.00000000   0.00000000  -0.00000634          0.00      0.00     -0.93
>
> So, my problems are:
> 1. The elastic constant c'=(c11-c2)/2 (sigma_xx=sigma_yy=2c'e) 
> calctuated from stresses are different.
> 2. Sigma_ij should change sign if c/a changes from  0.995 to 1.005.
>
> Please, help me find what I am doig wrong.
>
> Best regards
> Dariusz Chrobak
>
>  
>