MathOverflow is a question and answer site for professional mathematicians. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

A neutral Tannakian category over a field $k$ is a rigid $k$-linear abelian tensor category $\mathcal{C}$ whose unit $1$ satisfies $\mathrm{End}(1) \simeq k$, and is moreover equipped with an exact faithful tensor functor $\omega : \mathcal{C} \rightarrow \mathrm{Vect}_k$ into the category of finite dimensional $k$-vector spaces.

Question:- Why the condition $\mathrm{End}(1) \simeq k$ is necessary to get equivalence with the category of finite dimensional representations of some affine group scheme?

Thanks in advance!!

share|cite|improve this question
Because it's obviously satisfied in a category $Rep_k(G)$. – AFK Oct 9 '10 at 8:29
up vote 5 down vote accepted

If you say the other axioms correctly, then the condition on $\operatorname{End}(1)$ is redundant. Indeed, the word "tensor functor" implies that $\omega: 1 \mapsto k$, and the word "faithful" implies that $\operatorname{End}(1) \hookrightarrow \operatorname{End}(\omega(1))$. What you should include that you don't on your list is that $\omega$ be $k$-linear. You should also demand that $\mathcal C$ be a nontrivial category; then you cannot have $\operatorname{End}(1) = 0$, as $\operatorname{End}(1)$ acts on all other homsets via the $1$ action and in particular $\operatorname{id}\in \operatorname{End}(1)$ acts as the identity on all other homsets. With all of this, it follows that $\operatorname{End}(1) = k$.

Conversely, you can see the condition that $\operatorname{End}(1) = k$ as being a "nontriviality" condition. It is necessary only to assure that $\mathcal C \neq 0$. In particular, no group has zero representation theory, as every group has a trivial representation on $k$.

If you believed in "the empty group", then you would not need this restriction: the zero category is the category of representations of the zero ring, which is "the group ring of the empty group".

share|cite|improve this answer
Nice answer. After all I think $End(1)=k$ is a more natural condition than $C \neq 0$. The first property is (a priori) preserved under more operations as the second one. – Martin Brandenburg Jan 5 '11 at 17:58

Assuming you have such an equivalence of categories, the object $1$ is sent to the trivial representation of your affine $k$-group $G$. This is the representation that factors through the canonical homomorphism from $G$ to the trivial group, and its endomorphisms are the endomorphisms of the one dimensional $k$ vector space, i.e., the ring $k$. Since the condition $\operatorname{End}(1) = k$ is implied by the existence of such an equivalence, it is a necessary condition.

share|cite|improve this answer
Actually I was trying to find that where the condition $\mathrm{End}(1) = k$ is used in the proof of equivalence. – Yashica Oct 9 '10 at 12:41

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

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