The ABINIT Users Mailing List ( CLOSED )

[Matthieu Verstraete <mjv500@york.ac.uk>]

Hello again -

> I think you mean the occupation number from previous run at some high
> electron temperature? The Fermi function will have to be modified for
> a change in the chemical potential (fermi level) which is a function
> of electron temperature. Is it taken into account in the code?

If the ground state is done with occopt 3 and a given temperature, fine - 
the states and potential should be good. However in the elphon code the FS 
integration is then done with a gaussian smearing around the (correct) 
E_F. What you could add is a FD smearing around E_F instead of the 
Gaussian - I think this would be more physical. The right way to go would 
be to return to T dependent equations, and I believe a f_nk(1-f_nk) comes 
out, which is just the FD smearing function up to a factor. To be checked, 
but this could indeed be read in from the GS run (in that case you impose 
the elphon smearing width to be the GS T, which is fine but not flexible 
if you want to change the width in elphon).

The trouble with the Gaussian is that it gives quite a different shape 
(exp(-e^2) decay instead of exp(-e)) and it's not easy to find an 
equivalent width of Gaussian to reproduce the FD effects.


Matthieu