gap with hybrid functional (wien mailing list)

abdel Mar.. Wed, 18 Sep 2013 08:52:06 -0700

dear Wien2k community,
i'm interested on the calculation of gap for 4f materials with hybrid 
functional.
(1) with onsite B3PW91
 (2) with full hybrid  B3PW91

for (1)  the value with grep GAP  case.scf  in agreement with  difference 
betwwen (the top) VB  
and (the bottom) of CB in Total DOS plot, also in agrement with expt. value ~ 
3eV

for (2) i have 2 values 
 big one with grep GAP in the scf file   ~ 3.4 eV
small from total DOS plot ~ 0.8 eV 

can any one tell me please where this difference comes from for the same full 
hybrid functional 
and between on site and full hybrid functional? 

Regards 
 
M.

Re: [Wien] gap with hybrid functional

tran Wed, 18 Sep 2013 09:24:39 -0700

Hi,

For DOS calculation with full hybrid functionals you need to
execute lapw2 and tetra with -hf:
x lapw2 -qtl -hf (-up/dn)
x tetra -hf (-up/dn)

Maybe this was the problem. This is not explained in the users's guide,
but we will add a paragraph about that soon.

F. Tran

Re: [Wien] gap with hybrid functional

西村 真一 Wed, 18 Sep 2013 16:13:32 -0700

Dear Dr. Tran and WIEN2k community,

I have a question related to this topic.

DOS plot with the "-hf" switch is truncated at a lower energy than Emax value 
in the case.int file.
The Emax in "-hf" mode is determined by the "nband" variable in case.inhf file?

Could you please tell me the detailed definition of the DOS plot generation 
with full hybrid functional calculation?

Thank you in advance

Shinichi Nishimura

Re: [Wien] gap with hybrid functional

tran Thu, 19 Sep 2013 00:39:00 -0700

For LDA/GGA calculations, emax in case.in1 determines how many states
are calculated and stored in case.vector (used for DOS plotting).
The corresponding energies are printed in case.energy.

For hybrid calculations (-hf), this is nband in case.inhf which determines
the number of states stored in case.vectorhf (used for DOS plotting).
The corresponding energies are in case.energyhf. nband means that nband
LDA/GGA orbitals in case.vector will be used as basis functions for the
Hartree-Fock Hamiltonian, therefore the number of states in case.vectorhf
can only be smaller (typically much smaller) than the number of states in
case.vector.

It is obviously useless to choose for emax in case.int a value which is
larger than the energy of the highest state in case.vector(hf).

For DOS plotting you need to use -hf:
x lapw2 -qtl -hf (-up/dn)
x tetra -hf (-up/dn)

F. Tran

For DOS plotting,