Continuous cohomology of a discrete representation of a compact group #
Mathlib builds continuous cohomology as the homology of the homogeneous cochain complex, whose
terms are the invariants of the iterated coinduced representations C(G, C(G, …, X.V)). Over a
compact group G and for a representation whose underlying module is discrete, every one of those
function spaces is discrete in the compact-open topology, hence so is every term of the complex and
every subquotient of it. So continuousCohomology n X is a discrete topological module.
This matters because the topology is part of the statement. The low-degree explicit model presents
H¹ and H² as quotients of subgroups of G → M and G × G → M; those carry the pointwise
topology, whose quotient is not discrete for an infinite profinite G (with trivial ZMod 2
coefficients on an infinite product of copies of ZMod 2, no finite set of evaluations isolates
the zero character). A comparison between the explicit model and the canonical one must therefore
say in which category it holds, and the results here are what make the canonical side of that
comparison a discrete object.
This implements the "category of the comparison" milestone of Layer 3 of the human-authored
roadmap at TauCetiRoadmap/ProfiniteCohomology/README.md.
Every term of the coinduced resolution of a discrete representation of a compact group is
discrete: it is an iterated space of continuous maps out of the compact group G.
Every term of the homogeneous cochain complex of a discrete representation of a compact group is discrete.
The continuous cocycles of a discrete representation of a compact group are discrete.
The continuous cohomology of a discrete representation of a compact group is discrete.
This is the statement that lets a comparison with an explicit low-degree model be phrased as an
isomorphism in TopModuleCat k between discrete objects rather than as an additive isomorphism
after forgetting the topology.