Let $T$ be a triangulated category and $A$ and abelian full subcategory of $T$. Consider the Yoneda extension groups $Ext_{A}^n$. For any two objects $X$ and $Y$ of $A$ here is a map $$ Ext^n_A(X, Y) \longrightarrow Hom_T(X, Y[n]). $$

I am looking for a reference for this theorem: assume that $A$ is admissible (meaning that every short exact sequence in $A$ comes from a distinguished triangle in $T$) and stable under extension. Then the above map is an isomorphism for $n=1$ and injective for $n=2$.

To be honest, I am not completely sure that this is true in this generality. I have seen people using this when $A$ is the heart of a t-structure on $T$ (which is an example of full admissible abelian category), but I couldn't find any reference in BBD.

  • $\begingroup$ Maybe this is part of admissible and I'm misreading it, but one needs to ask for some sort extension closure to have any hope of an isomorphism for $n=1$. For instance if E is an object with endomorphism ring a division algebra, then the full subcategory on sums of copies of E is semisimple abelian, but E could have self extensions. $\endgroup$ Jan 27, 2017 at 20:08
  • $\begingroup$ Dear Greg, you are of course right! I forgot to ask that $A$ is stable under extension. This is not part of the definition of admissible. I will edit it now. $\endgroup$
    – triang
    Jan 27, 2017 at 20:18

1 Answer 1


Here are two places in the literature where this is discussed, probably there are more:

On p.3 of

  • P. Deligne and A. Goncharov: Groupes fondamentaux motiviques de Tate mixte. Ann. scient. ENS 38 (2005), pp. 1-56.

The proof shows that $Hom$ is a $\delta$ functor and $Ext$ is the universal $\delta$-functor and then invokes a theorem of Buchsbaum.

There is also a proof in Section 1 of

  • M. Levine: Tate motives and the vanishing conjectures for algebraic K-theory. In: Algebraic K-theory and algebraic topology (Lake Louise, 1991), 167–188, NATO Adv. Sci. Inst. Ser. C Math. Phys. Sci., 407, Kluwer, 1993.

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.