stress tensor: stress.F90

Total energy, geometry optimization, DFT+U, spin....

Moderator: bguster

Locked
johnbrehm
Posts: 12
Joined: Mon Sep 25, 2017 9:30 pm

stress tensor: stress.F90

Post by johnbrehm » Thu Sep 13, 2018 4:57 pm

I came across a post from 2009 concerning

Hi,
from the comments of the code in stress.F90, where the stress is calculated:

!! NOTES
!! * Concerning the stress tensor:
!! See O. H. Nielsen and R. M. Martin, PRB 32, 3792 (1985).
!! Note that first term in equation (2) should have minus sign
!! (for kinetic energy contribution to stress tensor).
!! Normalizations in this code differ somewhat from those employed
!! by Nielsen and Martin.
!! For the stress tensor contribution from the nonlocal Kleinman-Bylander
!! separable pseudopotential, see D. M. Bylander, L. Kleinman, and
!! S. Lee, PRB 42, 1394 (1990).
!! Again normalization conventions differ somewhat.

!!Also, if you do your run with prtvol 10, all the contributions to the stress tensor will be output to the log file (not the main output file, just the log file).

My questions:
1. When I examine the current code, stress.F90, I notice the same comments.
what is the status? has it been fixed?
If not: If the first equation on some equation (2) should have a minus sign -- does it have it in the abinit code?
if not: do I need to subtract of this kinetic energy contribution twice to get the correct stress?

2. I used prtvol 10. I still only get 6 tensor components (in each of .log and .out): 11, 22, 33, 32, 31, 21.
Should I not be getting 3 more: 23 13 13?
Or are assumptions being made?

Locked