
The fourth run was done to avoid the strange nodes
that appeared in the second-zeta of the Li 2s orbital.

The fifth run (simlex.in.5, Basis_history.5, simplex.out.5) 
 was motivated to avoid the messsage of 
 "too small split norm" in the O orbitals.

The sixth and seventh run, in the same spirit, 
but to further improve the basis set.


The basis sets produced after optimizations numbers 5 and 6
could not improve the energy of the optimization number 4.
Even more, after a complete relaxation of the structure,
the relative differences in energies increased up to 55 meV.

The basis set produced after optimization number 7
improves the energy upto the level of basis set 4
(only a few meV of difference), however
a full optimization of the structure produces energies
that are much worser than the ones obtained with the basis set 4
(up to 30 meV of difference), and does not improve the geometry
of the ferroelectric phase.

The basis set stored in LiNbO3basis.fdf corresponds to 
optimization number 4.
