Optimization and Total energy calculation with " did not converge"

[fangyongxinxi]

Dear, I have a problem when I calculate optmization and total energy calculation.

The input file is show in the attachment. In the scritp,  both optimize and total energy are calculated.
then I run the script like "atk14version input.py > out.log"

when the calculation finished,
I use:
"grep Force out.log" to check how is the optmization going,
and then I use
"grep did out.log", (here "did" means if the calculation did not converge, the word "did" will show in out.log), to check the converge.

As the attachment shows,
after six steps, the Force on each atom is less than 0.05 eV/A, and the optimization is stopped. And the the total E is calculated.

However, for each optimize step, and for Total E calculation, the SCF is not converged in 100 cycle.
so when I did "grep did out.log",  there is seven lines:
" Warning, the calculation did not converge to the requested tolerance !"

So,  my question is :
the optimizatied structure is ok ?
the number of the total energy is ok ?

If they are not ok, is there some calculation parameters should be reset to avoid the converge problem ?

Thanks

[Jess Wellendorff]

The optimized structure may or may not be ok if it is based on minimizing forces from a non-converged Kohn-Sham Hamiltonian. The only way you can know for sure if the structure is ok is to get the forces from a well converged calculation.

I suggest increasing the maximum number of allowed SCF steps and and/or changing the damping factor in the SCF loop.

The ATK manual tells you how to do this (and many other useful things):
http://www.quantumwise.com/documents/manuals/latest/ReferenceManual/index.html/ref.iterationcontrolparameters.html

[fangyongxinxi]

Thanks for your reply.
Next, I do as you said.

The optimized structure may or may not be ok if it is based on minimizing forces from a non-converged Kohn-Sham Hamiltonian. The only way you can know for sure if the structure is ok is to get the forces from a well converged calculation.

I suggest increasing the maximum number of allowed SCF steps and and/or changing the damping factor in the SCF loop.

The ATK manual tells you how to do this (and many other useful things):
http://www.quantumwise.com/documents/manuals/latest/ReferenceManual/index.html/ref.iterationcontrolparameters.html