[Wien] empty DOS for isolated atom or molecule

[Peter Blaha]

This is almost a textbook example on quantum mechanics for molecules and 
solids and the corresponding numerical approximations.

You want a "DOS" for an isolated molecule !!!  and of course an isolated 
molecule has only DISCRETE molecular orbital levels.
If we approximate an isolated molecule by a "bandstructure" approach and 
periodically repeat the molecule at large distances, the proper "DOS" is 
a sum of some delta-functions (at the MO-levels). Their k-dispersion is 
zero.
Since you calculate the DOS on a discrete energy mesh (every 2 mRy) it 
is very unlikely that you "hit" the energy of the delta function (eg. 
the eigenvalue is at -0.51234567, but you calculate the DOS at -0.514, 
-0.512, -0.510,... and therefore it is all zero. (Check the integrated 
DOS in case.outputt, you will see jumps by 2 electrons from time to time).

Why do you "see" a DOS at high energies ??? These are antibonding 
states, which are very delocalized an space. Thus there is some overlap 
with the next molecule (15 bohr away), leading to some k-dispersion and 
suddenly you see some (artificial) DOS.

Just check case.output1 for the energy levels of your "DOS".

Yurko Natanzon schrieb:
> Dear Wien2k users and developers,
> I'm experiencing troubles in calculating DOS for isolated atom or
> molecule in a box. For example, for H2 molecule in a 15x15x15 a.u. box
> I get empty DOS for E<Ef and some occupied states at E-Ef > 10 eV. I
> tried to change the default input in case.int, but with no success.
> 
> I'd be glad if you guide me what I'm doing wrong. The case.int is the following:
>  -1.50 0.002 1.500 0.003   EMIN, DE, EMAX, Gauss-broadening(>;de)
> 
>     1                      NUMBER OF DOS-CASES specified below
> 
>     0    1   total         atom, case=column in qtl-header, label
> 
> The calculation was initialized with init_lapw -sp -numk 100 -rkmax 2.5 -vxc 13
> 
> with kind regards,
> Yurko
> 
>