Hello Rohit,

As far as I know the recommended BZ integration scheme for metallic systems is the Methfessel-Paxton, e.g., ISMEAR=1.

Best regards,

Witold

Hi Rohit,

please give more details of your calculation. I generally see no problem with ISMEAR = -5 for a metal unless you need the forces, but conceivably the incorrect partial occupies may change the magnetic moments. You can try and set LORBIT = 11 and analyze in detail which moments change in which way. And, as Witold already commented, for metals ISMEAR>0 is recommended for relaxations.

How many magnetic atoms are there in your system? How high is the change per atom? Are all other settings equivalent for both calculations?

Cheers, Michael

Thanks Witold & Michael a lot for your responses !

I have used ISMEAR=1 for the relaxation purpose & for the static calculation used ISMEAR=-5. But when I got DENTET error with ISMEAR=-5 then changed ISMEAR=0 as suggested here http://www.error.wiki/The_old_and_the_new_charge_density_differ 

Yes all other settings are same for both the static calculations and I also used LORBIT=11 in both the cases. Please see attached file of magnetic moments from all magnetic atoms of my system which is thin film, I am doing supercell calculation.

For some cases, I got rid of DENTET error by increasing NEDOS and KPOINT with same ISMEAR=-5 but few are still giving same error.  I have not used ISMEAR=1 for static case, let me do that and then I will compare the moments from that also.

@Michael, Do you need more info, or any particular input or output files?

Hej!

Please attach the INCAR, OUTCAR and POSCAR of your system.

There are no large moments even for ISMEAR=-5, so I guess they form because of incorrect partial occupations. Do you expect a magnetic solution? Are you setting a magmom tag? Appatently your system shows nearly no or no local moments and the 8muB you are getting for ISMEAR=0 is all in the interstitial region. Is that something you expected?

Double check with ISMEAR=1, but I would trust 0 over -5 for this small local moments.

thanks for your valuable suggestions !

I got my problem solved by playing with KPOINTs and NEDOS tag along with ISMEAR=-5 I have increased #kpoints (20X20X1) and NEDOS=1000 and not getting the DENTET warning.

Initially I have tried using several odd numbered kpoint mesh and the last one was 19X19X1 and NEDOS=500. Does even and odd numbered kpoint mesh also matters?

ISMEAR=1 is yet to be done, that I will try with some well known example and see the effects.

thanks

Rohit