convergence problem with isolated Ti atom

[linpo]

The seemingly simple run did not converge in 300 electronic steps. I really did not expect the bad convergence. I was trying to find the total energy of an isolated Ti atom, but VASP refused to converge. The INCAR is

--------------------------------------------
SYSTEM = isolated Ti

ISTART = 0; INIWAV = 1; ICHARG = 2
IALGO = 48

PREC = High
ENCUT = 290.00
EDIFFG = -0.01; EDIFF = 1E-5

ISMEAR = 1; SIGMA = 0.2
IBRION = 2; ISIF = 2; NSW = 20

LWAVE = .False.; LCHARG = .False.
#LREAL = .FALSE.
#LPLANE = .TRUE.
NPAR = 1; NSIM = 4

NBANDS = 12
----------------------------------------------

And I used one k-point, Gamma centered. One Ti atom resides in a 10x11x12 cell. The PAW potential is from vasp resposire.

How to improve the convergence? Because it is an isolated atom, it should be easy to set up the job and test. I am curious to see different results.
<span class='smallblacktext'>[ Edited ]</span>

[admin]

please read the handbook chapter Examples:Atoms. You will at least have to set ISPIN=2 so that Hund's rule can be obeyed.
Also, it makes no sense to set the ionic relaxation tags (IBRION, EDIFFG,NSW,...) for a single atom in an otherwise empty box

[rpf]

Hello I have the same problem. I have tried a lot of different ways but It is impossible to get the correct ground state of Ti (and of Zn).

Though I apply the following recipe given by VASP website (as an example) it gives me a ground state with 3.4 magnetic moment. That is weird...

ISYM = 0 ! no symmetry
ISPIN = 2 ! allow for spin polarisation
VOSKOWN = 1 ! this is important, in particular for GGA
ISMEAR = 0 ! Gaussian smearing, otherwise negative occupancies
SIGMA = 0.1 ! intermid. smearing width
AMIX = 0.2 ! mixing set manually
BMIX = 0.0001
NELM = 20 ! 20 electronic steps
ICHARG = 1

Could you please solve this problem? It seems very simple but there is still no solution.

[rpf]

any solution ?