Thank you Professor Marks and Professor Blaha,
A mentioned before I ran these two systems
Graphene + M
but this time I ran them
First without "-min"
then with "-min"
Graphene + M
it took many cycles, but  it finally converged.
The same for
but it took many more cycles, more than 500, fortunately it took 1 min each 
This time the forces did nor oscillate.

De: Wien <> en nombre de Peter Blaha 

A structure optimization of a magnetic structure with several TM atoms
may take many iterations.

You should probably start out with a normal

runsp -fc 2 -orb -p  -i 80   (it may not converge in 40 cycles ?).

Then inspect the forces. How large are they ? If they are larger than 5
mRy/bohr, it is probably worth to optimize the structure and you restart
the runsp command using  -min

After a couple of cycles investigate   :ENE and :FR
both quantities should go down slowly, but expect jumps, etc.
As I said before, it may take even a couple of 100 cycles in a bad case.

Wien mailing list

Reply via email to