Easy manual occupancy of Kohn-Sham levels with FERWE and FERDO

From Nano Group Budapest
Jump to: navigation, search

The vasp wiki tells that one can set manual uccupancy with FERWE and FERDO:

One has to converge a WAVECAR before proceeding further.

Let say your system contain 2042 electrons. Assume the ground state is a closed shell singlet. It looks enough to do the calculation without any spin polarization with ISPIN=1 since both the ground and the excited state must have the same spin singlet wavefunction. But in order to move an electron from the HOMO to the LUMO one has to preconverge the ground state with ISPIN=2.

To do an excitation, one has to promote an electron from spatial orbital #1021 to #1022. To do so, one has to

ISTART = 1

ISTART=2 also might work, but ISTART =1 is not a simple vasp continuation. ISTART=1 will transform the old WAVECAR into the new job's bases, or truncate the bands, or create new ones if the continuation job have more NBANDS setting than the original. Keep in mind NBANDS sighly varies with the number of MPI threads on which VASP is being executed, since NBANDS/THREADS must be an integer. VASP will automatically increase this value accordingly regardless your NBANDS is fixed.

ISMEAR = -2

ISPIN = 2

# 1020 1021 1022 1023 just a commend line

FERWE = 1020*1.0 1*0.0 1*1.0 1000*0.0 #The syntax is BANDS*OCC

FERDO = 1020*1.0 1*1.0 1*0.0 1000*0.0

The 1000*0.0 means zero electron for the next bands. Note this will do not override the NBANDS = 1150 setting is this will dont calculate 2022 bands, the bands after 1150 defined with FERWE and FERDO will be simply ignored.

The following SCF cycle alterations also required for HSE06 deltaSCF calculations:

ALGO = All TIME = 0.4 LDIAG = .FALSE. LSUBROT = .FALSE.

shf3

the shellpack have the option to copy the WAVECAR along the with the inputs. OTHER="diamSiV.WAVECAR.gz" Just uncomment the OTHER in your *.guide file!