NELECT problem

Message

kongfei77 · #1 Post by **kongfei77** » Tue Apr 04, 2006 2:36 pm

I am using the method in VASP manual Section 8.6.2 in page 103 for accurate bulk relaxations with internal parameters (one).

They recommended 3 steps there. When I switched to 3rd step, (ISMEAR = -5), no relaxation (setting FFF in the POSCAR files), during run, this always appear:

Number of Electrons is NELECT = 44.9605375625845
DAV: 1 -0.325940257424 E჎ -0.32594 E჎ -0.68111E-02 280 0.158E჌BRMIX: very serious problems

I tried playing with other parameters, still can't get this error off although there is convergence. Any suggestions?

Thanks
kongfei77

kongfei77 · #2 Post by **kongfei77** » Wed Apr 05, 2006 8:42 am

Several postings regarding this issue, like for example, Warning: DENTET, can't reach specified precision advised the user to adjust NEDOS or increase k points.

Other options include odd numbers for kpoints and gamma method. None of these seems to work and I still receive the warning: dentet, can't reach specified precision IF I insist on using ISMEAR = -5.

However, if spin polarized calculation is included, this problem disappears or if I switch to ISMEAR = 1/2.

Any reasons?

Thanks,
kongfei77

#3 Post by **admin** » Tue Apr 11, 2006 1:13 pm

have you chosen enough k-points to cover the Brillouin zone completely with tetrahedra?
ISMEAR 1 of 2 uses a completely different algorithm (Monkhurst-Pack) for BZ integration.