I have a reference request on following comment I found in nLab article on Karoubian categories & envelopes. It states:

The Karoubian envelope is also used in the construction of the category of pure motives, and in K-theory.

Almost every introduction to motives containing the basic constructions (eg Manin's "Correspondences, motifs and monoidal transformations" as 'standard' reference) includes the fundamental part where one passes from the category $(\mathsf{Sm}/k)$ of smooth curves over $k$ to its Karoubian closure.

On the other hand I'm not sure to which construction in K-theory where the Karoubian envelope is involved, the quoted sentence refers.

The first naïve observation is that for a ring $R$ the $K_0(R)$ in algebraic K-theory is obtained as a certain quotient group ("Grothendieck group") of the set of projective $R$-modules. This set of projective $R$-modules can be reinterpreted as the Karoubian envelope of the set of free $R$-modules.

Question: Is this the only explicit usage of Karoubian envelope in constructions in K-theory or are there more general cases? I'm quite not sure if the remark above only refers to this 'baby' case with $K_0(R)$, or there are more general constructions in K-theory where Karoubian envelopes are involved.

For example another nLab article on the more general development of K-theory nowhere explains where Karoubian envelope is explicitly used as a technical tool for certain constructions. Therefore I would like to know if there are some recommendable papers on development of K-theory where such constructions involving Karoubian envelopes are discussed.


2 Answers 2


In Schlichting's paper Negative K-theory of derived categories. Math. Z. 253, 97–134 (2006), a definition of negative $K$-theory of triangulated categories is given. These are abelian group valued functors $\mathbf{K}_{i}$ with $i\leq 0$.

These groups agree with the known negative $K$-theories, for example in the Thomason-Trobaugh paper mentioned by Prof. Grayson. However they use idempotents in an essential way.

In particular, if $E$ is an exact category then the bounded derived category $D^b(E)$ is triangulated and $\mathbf{K}_{0}(E)$ is defined to be the usual $K_0$ of $\widetilde{E}$, the idempotent completion of $E$.

If $E$ is idempotent complete, then the unbounded derived category $D(E)$ exists, and the $\mathbf{K}_{-1}(E)$ turns out to be the quotient of the abelian monoid (under direct sum) of isomorphism classes of idempotents in $D(E)$ by the submonoid of split idempotents. In particular $K_{-1}(E)=0$ if and only if $D(E)$ is idempotent complete.

More generally, if $A$ is an idempotent complete triangulated category then $\mathbf{K}_{-1}(A)$ is zero if and only if the Verdier quotient $B/A$ is idempotent complete for all full triangulated embeddings $A\to B$ with $B$ idempotent complete.

Incidentally, Schlichting conjectured that whenever $A$ is a small abelian category, then $\mathbf{K}_{i}(A)=0$ for all $i<0$.

This conjecture was generalised to stable $\infty$-categories by Antieau, Gepner, and Heller in K-theoretic obstructions to bounded t-structures. Invent. math. 216, 241–300 (2019), wherein they conjecture that $\mathbf{K}_{i}(A)=0$ for $i<0$ whenever $A$ is a stable $\infty$-category with a bounded $t$-structure.

Neeman gave a very simple and elegant counter-example to both these conjectures in a recent preprint https://arxiv.org/abs/2006.16536.


Thomason discusses the construction in (A.9.1) in his paper with Trobaugh, Higher algebraic K-theory of schemes and of derived categories, The Grothendieck Festschrift, Vol. III, Progr. Math., vol. 88, Birkhäuser Boston, Boston, MA, 1990, pp. 247–435. I don't know how he uses it in the body of the paper.

I use it in an essential way to derive a result for all exact categories in Corollary 6.5 of my paper Algebraic $K$-theory via binary complexes, J. Amer. Math. Soc. 25 (2012), 1149–1167.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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