The ABINIT Users Mailing List ( CLOSED )

[wlu7910@126.com]

Dear Anglade Pierre-Matthieu,
   Thank you very much for you rapidly response.
   About 1)I read some papers (eg. APL 91,031906, 2007). They can get the 
stress for a different strain. In Abinit output file, I find that there are 
stress tensors that diagonalized,but which value is the stress of the cell? 
Did your answer previously mean that the stress of the cell is equal 
(-pressure). Forgive me that my English understanding is not satisfying.
  About 3) Do you mean that the pressure should be equal for primitice cell 
and conventional cell if the calculation is convergent?  Can you tell me how 
to set the shiftk for conventional cell?  When I test the convergence through 
increasing "kptrlen", I should omit the "ngkpt",right?  if shiftk should be 
omitted also by using "kptrlen".
  Thank you very much again for your time.

Best Wishes,
   WLu

Hi,

1)About your first question:
The usual relation between the stress tensor and the pressure are fulfilled 
by Abinit output.

2)I think it is possible to compute the strain with the conventional cell by 
applying rotation matrix to your strain or to your cell. 

3) The relationship between your two computed stress is likely the big word 
of Abinit calculations: CONVERGENCE. You are witnessing effects of 
convergence. Your ecut seems perfectly converged (I have small study of 
Abinit pseudo for Al here: 
http://anglade.googlepages.com/anglade-abinit-hands-on.pdf). So likely your 
k-point mesh is not converged enough. You mention that the shift you use are 
the same. Yet the cells are different so the sampled k-points are 
different... By the way I think that this convergence test you perform is 
very well suited for the kind of calculation you want to do. You just have to 
increase "kptrlen" up to finding the same values for pressure in your two 
cells. 

Regards

PMA


On Nov 22, 2007 6:26 AM, <wlu7910@126.com> wrote:

Dear all,
  I am a newer to ABINT. I want to ask if the pressure given in the *.out 
file is equal to (-stress)?
  Another question is that I want to calculate the shear stress for a series 
of strains along the direction eg. (111)<1 -1 0>. But I don't know how to add 
this kind strain to the primitive vector (for example fcc). what I know now 
is to redefine the structure of the original fcc structure (cleave plane 
(111) with surface vector [ 0.5 -0.5 0] and [0.5 0.5 -1], and the third 
direction is perpendictular with (111)), then add small strain to the third 
direction along [0.5 -0.5 0] direction. Now I meet a problem, If I can use 
the conventional cell to get the stress of the distorted structure in ABINIT? 
  I use the tutorial structure of Al (still keep fcc structure without 
distortion)with lattice constant 7.8 bohr (do not optimization)to test the 
stress in primitive cell and conventional cell, respectively. The result of 
these pressures are [Pressure= - 5.6996E+00 GPa] and [Pressure= -6.9105E+00 
GPa] (ecut=20 H and shiftk is same for the two tests). So what's the 
relationship of the pressure calcuated in primitive cell and conventional 
cell?

  Thank you very much for your help! 

Best Regards,
  WLu