One can naturally think of a cospan as the abstraction of a cobordism. For instance an interval object cospan models the standard topological interval regarded as a cobordism from pt to pt. The cospan cotrace on the interval glues the two ends of the interval together to produce a circle regarded as a cospan from to itself.
The concrete dual of a cospan, obtained by mapping it into some target object, is a span, which in the context of groupoidification and geometric function theory can be interpreted as a generalized linear map. On such a generalized linear map, there is a notion of trace, the span trace.
The cospan cotrace is the concept dual to that: the image of the cotrace of a cospan under mapping it into a target object is the span trace of the result of mapping the original cospan to that target object.
For
a cospan with identical left and right index object , its cospan cotrace is the composite of the result
of dualizing one leg of the cospan with the cospan
i.e. the pushout
regarded as a cospan from the initial object to
More generally, the trace of a multi-cospan over identical of its index objects is the composite with the multi-cospan
Let the ambient category be Top, let be the standard topological interval and let be a small interval, for some – to be thought here as a collar of the point .
Let
be the interval regarded as a collared cobordisms from the point to the point. Its cotrace, the pushout
is the result of gluing the ends of the interval to each other, i.e. the circle
Urs: This may require a bit more care
with the topology involved. I still need to check the reference below for more details.
See also
Let the ambient category be Cat, let be the standard interval object in Cat and let be the terminal category.
Let
be the standard interval object in Cat regarded in the standard way as a cospan from the point to the point.
Dualizing it to
corresponds to thinking of it as a “bent interval”
Accordingly, the co-span
can be thought of as
Gluing these two arcs together yields the cotrace, the pushout
which is the result of gluing the ends of the interval object to each other, which here is the skeleton of the fundamental category of the directed circle
namely the monoid of natural numbers, regarded as a one-object category:
If instead we start with the standard interval object in groupoids, with the nontrivial morphism from to being an isomorphism, then the co-trace in question is the skeleton of the fundamental groupoid of the ordinary topological circle
While the concept is obvious, it is apparently (?) not discussed yet in the (young) literature on the subject. On the blog the concept was mentioned in
Last revised on June 17, 2019 at 11:48:54. See the history of this page for a list of all contributions to it.