[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: trouble with feff8



Dear Michel,

The trouble with feff8 you encountered is common for feff8, so
I sent my reply to you also to all feffusers.
First, the problem always will occur if someone tries to calculate
LDOS below interstitial potential (V_int=-21.5 for Fe) with
ZERO broadening. In that case the "photoelectron" wave vector k
is purely imaginary, Re(k)=0 and subroutines for bessel functions
are set to do calculations only for Re(k)>0. A simple fix in this
case is to add small NONZERO broadening: e.g.
LDOS -25  -20  0.01
seems to fix problem in your input file.

Also feff8 even with small broadening sometimes gives negative LDOS within
about 0.5 eV from the V_int. It is purely numerical error that arizes
for |k|<<1. The calculation are done so that some finite number is 
obtained by divizion of two numbers close to 0. (i.e. A=0/0). 
The algorithm needs to be changed for |k|<<1 region. We will fix that
later.

Let me know if addition of small broadening helps your calculations.

Best wishes
Alex Ankudinov

On Thu, 14 Nov 2002, Michel Jaouen wrote:

> 	Hi Alex,
> 
> 	I have found a strange (for me!) problrm with feff82. I want 
> to calculate the ldos for Fe with a high precision to compare to the 
> results I obtain with Wien2k. So I run the code per piece : it gives 
> a new Fermi level at about -11eV so that to obtain the bottom of the 
> s-dos I must calculate the ldos between -25 and -20 eV. In that case 
> I obtain a strange error (see the file enclosed). I must say that it 
> works in the energy range -20 up to 20 eV, but it seems to fail under 
> -20 eV. Is there a limitation in the code? I join the feff.inp file I 
> use for running the code on our cluster (16 nodes) using the parallel 
> version. Any solution?
> 
> Best regards,
> 
> Michel
>