It is often stated in paper abstracts that

"One cannot build [[Wannier Functions]] for [[Topological Insulators]]", 

and when one reads the introduction, the statement changes to

"One cannot build **exponentially localized** [[Wannier Functions]] 
for [[Topological Insulators]]". 

Where does this distinction come from, and is it relevant for building WFs in practice?

The insulator picture

In a seminal paper [1], Prodan builds on top of Kohn’s idea of [[Nearsightedness]] in electronic matter.

SSH model

References