Let $G$ be a group acting on a category $C$. That is we have a morphism of groups $G \to Aut(C)$.

We can now form a new category as follows:

  • Its objects are tuples consisting of an object $x$ of $C$, together with morphisms $\Phi_g : g \cdot x \to x$, satisfying the condition : $\Phi_g \circ (g \cdot \Phi_h) = \Phi_{gh}$
  • A morphism $(x,\Phi) \to (y,\Psi)$ is a morphism $F:x \to y$ such that $\Psi_g \circ (g \cdot F) = F \circ \Phi_g$

Does this new category (or at least its objects) have a name?


Yes. It's called the category of homotopy fixed points $C^G$ for the action of $G$ on $C$. (You need a unit condition on $\Phi_e$ but that's not a big deal.) See, for example, the sequence of blog posts starting here.

A typical example is that $G = \text{Gal}(L/K)$ is a Galois group and $C$ is the category of objects of some kind "over $L$" (e.g. $L$-vector spaces, $L$-algebras, $L$-schemes); then Galois descent, when it holds, asserts that $C^G$ is the corresponding category of objects "over $K$." (Most resources about Galois descent on the internet will go to great lengths to avoid saying this.)

Note that $\text{Aut}(C)$ is a 2-group, not a group, and this distinction really matters. It's possible for $\pi_0(\text{Aut}(C))$ to be trivial (meaning every equivalence $C \to C$ is equivalent to $\text{id}_C$) but for $\pi_1(\text{Aut}(C))$ to be interesting, in which case actions of $G$ on $C$ can still be interesting even though the individual functors associated to each $g \in G$ are equivalent to the identity. The correct definition of an action of a group on a category is therefore slightly more involved and this slightly modifies the definition of a homotopy fixed point.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.