Model for FeSi, a strongly correlated insulator

Revised to reflect the differences between the second order perturbation theory method developed by Scheitzer and Czycholl, and followed by us, and the iterated perturbation theory (IPT)of Georges and Kotliar, et al. Because of our domain of interest, this does not materially change our conclusions, but it is important to realize that the method used here fails in the limit of large $U$, where IPT has had good results.