[Wien] Cohesive energies in WIEN2k

[Stefaan Cottenier]

> (1) First it states clearly not to use LSTART but LAPW to compute
> energies. 

In case.outputst, a total energy for the free atom by lstart is given
(fully relativistic), which will be different from the total energy of
the atom in a large empty box in a usual scf-cycle (scalar
relativistic). No problem to use lstart during initialization, but do
nut use the energy by lstart but the selfconsistent energy, that's the
message.

>   (2) It states that we have use identical RMT (which I set to say
> 2.6 in StructGen) and RKMAX. My question how to I make RKMAX 
> identical to RMT? Which case file contains RKMAX

Nono, not rkmax identical to rmt, but rkmax and rmt in the solid state
calculation identical to rkmax and rmt in the free atom calculation
(large empty box). rkmax is of course found at the top of case.in1 (it
is a VERY important parameter, and if you don't know where you can find
it or what it means, you'd better read the usersguide very carefully
once more, or you are likely to run into big problems).

>   (3) The k-point stuff is confusing as hell. In the initialization
> process using w2web, 
>   it will ask for number of k points. If I enter 100 k points say,
> and run x kgen, it tells 
>   me that 19 k points generated. What happened to the 100 k points. I
> do not get
>    which one is which.

The input to kgen is the number of k-points in the full Brillouin zone;
What you get returned is the number in the irreducible part of the
Brillouin zone. The relation between both is roughly
"irreducible=total/number-of-symmetry-operations" (but highly affected
by edge effects for small total number of k-points).

>   (4) Finally, why should the single atom at the origin be placed in
> an FCC  lattice 
>   and not a Simple Cubic lattice say ? In other words why FCC and not
> some 
>   other lattice.

I admit I do not get that point either, although I suspect it to be
simple. I would think that for a given lattice constant the interatomic
separation is largest in simple cubic?

Stefaan


Disclaimer: http://www.kuleuven.be/cwis/email_disclaimer.htm