Exact flatness of energy bands implies some reasons behind. Here we present one of them, "molecular orbital (MO) representation", which seems to be applied for various classes of tight binding models. Mathematically if the rank of the hamiltonian as a linear operator is less than the number of atomic sites, the kernel of the linear operator has a finite dimension. This is the zero mode flat band. The MO rep. presents nice physical reasons for it. Original proposal by YH with Isao Maruyama in 2011 in EPL and arXiv is counting dimensions of non-orthogonal projections but the hopping of the MO's is allowed as we pointed out (it should be). It's a fun to guess what kinds of the MO representation is possible for a known flat band system. Try ! Also several physical reason why the flat band crosses/touches to dispersive bands in many cases are discussed. Our new paper has appeared in EPL, "Molecular-orbital representation of generic flat-band models", by T. Mizoguchi and Y. Hatsugai , also arXiv.




