The ABINIT Users Mailing List ( CLOSED )

[mmikami@rc.m-kagaku.co.jp]

Hi,

I cannot graps how it failed, but my guess (question) is:
the simple, non-spin-polarized LDA computation with the default occopt(=1) 
will give a solution of a (narrow-gap) semiconductor for "FeGe" ?? (Since the 
FeSi has such a narrow gap (0.1 eV in LDA calculation), the simple LDA 
compuation of FeGe may lead to metallic solution. You might want to try 
literature survey before your own calculations.)

I guess that LDA with metallic occupancy (occopt = 3 or 4 or ... or 7 & set 
tsmear) may give a convergence somehow. But it should be noted that the 
converged solution may not be the REAL ground state for such a (possible 
strong-correlated) sysytem.
But it would be still helpful to understand what is going on in the 
approximation that you employ.

It may also be mentioned (as a general remark) that the internal 
coordinations ("xred") should be fully optimized before running optcel(>0) 
job. In my cases, I often use two datasets for such purpose; the first 
dataset with only ionmov = 2 or 3, then the second dataset with optcel(>0) & 
ionmov(=2 or3) & the xred's optimized in the 1st dataset.

Good luck,
Masayoshi