Re: Convergence of SCF Loop in FEFF8 for TiO2

["John J. Rehr" <jjr@leonardo.phys.washington.edu>]

Dear Users,

 I'm sending again - previous msg had incorrect copy of Moreno's list:

 Sergio Moreno reported a lack of convergence of the SCF loop
for TiO2 with the following list of POTENTIALS:

POTENTIALS
*    ipot   Z  element            l_scmt  l_fms   stoichiometry
        0   22   Ti                2       2       0.01
        1   22   Ti                2       2       2
        2    8   O                 1       1       4

 I was able to reproduce the problem on our SGI with his input file.
It seems that with this input, the SCF loop is almost converging, but
somehow going into a limit cycle.

 Then I changed the POTENTIALS as follows, and it converged in 11 iterations.
 The -1 for l_scmt means the FEFF default, which is usually recommended.
For this system it is 2 ( i.e., s,p,d basis) both for Ti and O.
Looks like the potentials for the O in TiO2 want/need a little d character
to converge.

POTENTIALS
*    ipot   Z  element            l_scmt  l_fms   stoichiometry
        0   22   Ti                -1       2       0.01
        1   22   Ti                -1       2       2
        2    8   O                 -1       1       4

 I also changed RPATH to 0.1 - you don't need or should use
the path expansion with FMS calculations, so set RPATH less than
distance to first shell.

*RPATH     4.6
RPATH     0.1

  J. Rehr

----- End of forwarded message from John J. Rehr -----