The notion of cohomology – in its generalization beyond generalized (Eilenberg-Steenrod) cohomology to nonabelian cohomology – is related by abstract duality to homotopy. In the (infinity,1)-topos Top this is traditionally called Eckmann–Hilton duality.

Therefore it does make sense to speak of general cohomology as co-homotopy.

Indeed, cohomology is a concept dual to homology only in the very restrictive simple case of chain homology and cohomology. And this is really just because in this simple case it so happens that homotopy of chain complexes is their homology. This is part of the statement of the Dold-Kan correspondence: the nerve operation on chain complexes N:Ch +SimpAbN : Ch_+ \to SimpAb identifies chain homology groups with simplicial homotopy groups.

So on general grounds the word “cohomotopy” is actually better suited than “cohomology” for a concept of such fundamental importance.

But the use of “cohomology” is deeply rooted in tradition and will hardly be changed just because the nnLab says so. Even though it should.

Revised on July 19, 2009 20:27:51 by Toby Bartels (