convergence problem

[zhu]

Dear Sir,
I am a newbie. I am doing the geometry optimization for metal Lu, but the electronic relaxation does not converge (as shown in the INCAR file below). The POTCAR file contains 25 electrons. I’ve tried some different algorithms such as ALGO=Normal Fast and VeryFast, still the same problem. However, when it comes to the POTCAR including 9 electrons with the same algorithm, there’s no such problem at all. What should I do to make it work when POTCAR with 25 electrons is used?
Thank you in advance.

Cheers!

Sincerely,
Zhu
09-07-14
INCAR file
ISTART = 0
ICHARG = 2
ISPIN = 2
PREC = Accurate
NBANDS = 30
EDIFF = 1E-6
NELM = 100
ALGO = Fast VeryFast Normal
SYMPREC=1E-6
IMIX=4
AMIN=0.1
AMIX=0.2
AMIX_MAG=0.8
BMIX=0.0001
BMIX_MAG=0.0001
IBRION = 1
POTIM=0.1
NSW = 200
ISIF = 3
ENCUT = 333
ISMEAR = 1
SIGMA = 0.2

[admin]

the kind of POTCAR you use can have an effect on the convergence behaviour especially for rare earth metals: convergence usually is significantly better if you use the potenial which best corresponds to the electronic ground state configuration of your metal: In your case, this seems to be the 5p6 5d1 6s2 configuration (with the f electrons localized in the core, not itinerant)
-- your INCAR file looks ok
-- please in any case check (especially for a small unit cell) if your KPOINT s are converged

[zhu]

Thank you for your help. My object is to calculate some compounds containing Lu such as Lu2O3 . In that case f electrons are likely to influence the character of the system. Will the same problem exit again? If so, how can I do to solve it?