FQFT and cohomology
There are well-known geometric models for some cohomology theories. For instance
It is an old idea that analogous to how differential K-theory is modeled by parallel transport in vector bundles and hence by functors , elliptic cohomology should somehow be modeled by functors , where is a 1-category of cobordisms equipped with maps to .
The fundamental idea of this program is essentially to encode parallel transport along cobordisms with maps into a given space pretty much along the lines of functorial differential cohomology. One expects that this encodes actually the differential refinements of the corresponding cohomology theories, such as differential K-theory. However, currently in the program one divides out concordance which effectively divides out the differential information and keeps just the underlying topological information.
The fundamental result of this program so far is that there is a useful notion of 2d FQFT over such that its partition function is indeed topological modular form-valued, so that it is a candidate for a model of tmf.
The following is going to be an exposition of this partial result:
Outline of the constructions and statements
See also the entry
A recent survey of the program is
The program was initiated in
After the original sketch in terms of extended FQFT subsequent work concentrated on ordinary 1-categorical constructions, the goal being to make very clear, and transparent and rigorous the constructions involved and the claim that
This latest development on this is in
warning I am being told that this is by now outdated and to be replaced by an improved version, which however is apparently not available yet