I'm scratching my head trying to understand what an extended TQFT associates to $(n-2)$-hypersurfaces.
Here's some intuition that I've developed. For an $(n-1)$-hypersurface chopped into $(n-2)$-hypersurfaces with the initial and final set to null $(n-2)$-hypersurfaces, $Z$ associates a vector space. Hence whatever objects are associated to $(n-2)$-hypersurfaces must: 1. Form a monoidal category. 2. Have a notion of "trace" that evaluates the vector space.
The only object from mathematics that I can think of that has the second property is a cell complex (it's cohomology is a vector space).
However, the nlab article suggests that the sought object is an linear additive category. I don't quite understand this. Here's a few questions that I have:
- So is it correct that different types of $(n-2)$-hypersurfaces are assigned different categories?
- It must be that linear additive categories themselves form a monoidal category, but I fail to see how.
- What is the unit linear additive category in the monoidal category of linear additive categories?
- Say I have a morphism from a unit to the unit in the monoidal category of linear additive categories, which corresponds to a $(n-1)$-hypersurface. How do I trace it to get the vector space?