We show that when sizable long-range Coulomb repulsion is included in
numerical studies of small clusters, different sizes and shapes should
be considered to obtain a ground state which represents correctly the
physics of the ground state (GS) in the thermodynamic limit. Also cut
ting off the interaction at small distances might lead to a spurious G
S. In particular the GS reported in the literature for the two-dimensi
onal three-band Hubbard model in the static limit at hole concentratio
n x = 0.25 is in fact a slightly excited state when an important (scre
ened or not) Coulomb repulsion at all distances is included.