One usually considers the category of complex linear group representations for a fixed group $G$. It is defined as the category whose objects are group morphisms $G \rightarrow GL(V)$ where $V$ is a complex vector space and whose morphisms are $G$-equivariant linear maps (equivalently, it is the category of functors $Vect_\mathbb{C}^G$).

My question is: can we consider a category of group representations where the group is allowed to change?

I would be tempted to define it as follows. Objects would be triples $(V, G, \pi: G \rightarrow GL(V))$ and morphisms $\big(V, G, \pi: G \rightarrow GL(V)\big) \xrightarrow{\Phi} \big(V', G', \pi': G' \rightarrow GL(V')\big)$ would be pairs $\Phi:=(\phi, \alpha)$ where $\phi$ is a linear map from $V$ to $V'$ and $\alpha$ is a group morphism from $G$ to $G'$ such that the following property is satisfied: $$\forall g \in G, \pi'(\alpha(g)) \circ \phi= \phi \circ \pi(g)$$

Does this definition make sense? If yes, why is this category never considered?

  • 1
    $\begingroup$ This is an example of the Grothendieck construction. $\endgroup$ – Zhen Lin Dec 21 '15 at 11:23
  • $\begingroup$ @ZhenLin, could you develop a little bit more (as I am not a mathematician and I am not familiar with Grothendieck's work)? Which construction are you talking about? Do you have a reference I could read about it? $\endgroup$ – FZalamea Dec 21 '15 at 11:41
  • 1
    $\begingroup$ It is only known as the Grothendieck construction, that is the problem. $\endgroup$ – Zhen Lin Dec 21 '15 at 11:44
  • $\begingroup$ The fact that the normally considered category is equivalent to functors $\operatorname{Vect}^G_\mathbb{C}$ - do you have any reference to a source where this is stated/proved? (And I'm sorry to post this as an answer, I wanted to comment, but I am new and not allowed yet :-( ) $\endgroup$ – P. Schulze Dec 13 '18 at 13:55

The Grothendieck construction (see e. g. wikipedia for start) is the following:

Let $C$ be any category and $F \colon C \to \def\Cat{\mathsf{Cat}}\Cat$ a functor. The Grothendieck construction for $F$ is the following category $\Gamma(F)$:

  • Objects of $\Gamma(F)$ are pairs $(A, x)$, where $A \in \def\Ob{\mathrm{Ob}}\Ob(C)$ and $x \in \Ob\bigl(F(A)\bigr)$.
  • $\def\Hom{\mathrm{Hom}}$Morphisms $f\colon (A,x) \to (B,y)$ are pairs $f = (f_0, f_1)$ where $f_0 \in \Hom_C(A,B)$ and $f_1 \in \Hom_{F(B)}(F(f_0)x, y)$.
  • Composition of $f \in \Hom{\Gamma(F)}((A,x), (B,y))$ and $g \in \Hom_{\Gamma(F)}((B,y), (D,z))$ is given by $$ (g_0, g_1) \circ (f_0, f_1) = (g_0 \circ_C f_0, g_1 \circ_{F(D)} F(g_0)f_1\bigr) $$

Your construction is the Grothendieck construction for the functor $F \colon \mathsf{Group}^{\rm op} \to \mathsf{Cat}$, $F(G) = [G, \def\V{\mathsf{Vect}_{\mathrm C}}\V]$, which sends a group to the category of its representations.

A morphism $f \colon (G, \pi) \to (G', \pi')$ in the Grothendieck construction is a pair $(\alpha, \tau)$, where $\alpha \colon G \to G'$ is a group homomorphism and $\tau \colon F(\alpha)\pi \to \pi'$ is a natural transformation of functors $G' \to \V$. Such a natural transformation is (as $G'$ only has one object), a $\V$-morphism, that is a linear map.


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.