QuantumATK Forum
QuantumATK => General Questions and Answers => Topic started by: Jahanzaib on July 11, 2023, 03:47
-
Dear Expert
I am facing alot of issues in geometry optimizations - I just optimized FeSn cluster on my stanene ribbons - I have run my calculation on cluster - but it took more then 4 days still not optimized. I have attached graph so please guide me.
- I have attached the inputfile (output file is too big- not attached) of my system have following geometry optimization parameters
max_forces=0.1*eV/Ang,
max_stress=0.2*GPa,
max_step_length=0.005*Ang,
- In addition, I tried different 3 other options where I changed force, stress and step size
1) max_forces=0.01*eV/Ang, max_stress=0.02*GPa, max_step_length=0.2*Ang,
2) max_forces=0.01*eV/Ang, max_stress=0.02*GPa, max_step_length=0.0005*Ang,
3) max_forces=0.03*eV/Ang, max_stress=0.05*GPa, max_step_length=0.001*Ang,
Can you some one look into my file and guide me what I am doing wrong or why the geometry is not optimized?
Thank you
-
Hi,
Although I am not sure if this is the right way for magnetic materials but here is the following:
1. Try first optimizing using unpolarized spin GGA.PBE then use optimized geometry to optimize for spin polarized, SGGA.PBE.
2. Increase your k points. (1x1x1 is very low)
3. Reduce damping_factor to 0.01 and increase max_steps=300
4. Keep max_step_length as default.
-
Thank you for your response.
My cell is very large in periodic direction having 165 atoms so I don't think I need large k points? I haven't adjust damping factor - I will adjust them and let's c.
Do you think with very large unit cell, i need larger k points? In principle, if a unit cell is large, then fewer k points are enough.
Thank you
-
Hi
I would still recommend increase K points to at least to 3x3. Maybe it will help.