Page 1 of 1

WARNING :stress and forces are not correct

Posted: Fri Nov 24, 2006 12:28 pm
by forwordom
Some problem :
1)After correct relaxation and static scf caculation ,i want to do band DOS and band structure ,but i have always get the following warning :
WARNING :stress and forces are not correct
why ? and if this warning will affect results ?
2)when relaxation, I have got message :ZBRENT: fatal internal in brackting system-shutdown; contact gK immediately
Could you tell me reason ?
3)and have got :ZBRENT: can't locate minimum, use default step . will this affect results ?

WARNING :stress and forces are not correct

Posted: Mon Nov 27, 2006 9:19 am
by admin
this has been discussed in the forum before,
1) if the k-mesh is changed, the read SCF charge density is not the self-consistent charge density for that new k-sampling set.
2) this means that the next geometry step could not be predicted,
please check
-- the relaxation history of your run (XDATCAR), maybe the first ionic step gave unresonable results, in that case decrease POTIM
-- check if your convergence parameter (EDIFFG) was chosen reasonably.
3) check if the system converged to a resonable solution (TOTEN smaller than for the geometries before that warning was written). It depends on the potential hypersurface of your system, one cannot say without looking at the numbers.