2
$\begingroup$

Question: Does every short exact sequence of injective bounded cochain complexes, $0\rightarrow I^\bullet\rightarrow J^\bullet\rightarrow K^\bullet\rightarrow 0$, split?

I am interested in a discrete setting. Let $\Pi$ be a finite poset, and fix a field $k$. The objects I am interested in are equivalently:

  • a finitely dimensional $k$-representation of the poset $\Pi$,
  • a finitely dimensional sheaf on $\Pi$ with Alexandrov topology with $k$ coefficients,
  • a finitely dimensional module over the bound quiver algebra over $k$ corresponding to $\Pi$.

For any of those we can define injective objects, and bounded cochain complexes. An injective cochain complexes is a cochain complexes $I^{\bullet}$ such that each $I^{d}$ is injective.

Given a SES in the question, $0\rightarrow I^d \rightarrow J^d \rightarrow K^d \rightarrow 0$ splits for every $d$. But can this split be done consistently over all degrees so taht it is compatible with the cochain maps?

$\endgroup$
3
  • 2
    $\begingroup$ No, if that was true the category of finite dimensional representations of your poset would be semi simple. Notice that you can realize any sort exact sequence of representations as $H^0$ of a short exact sequence of such complexes. $\endgroup$ Commented Jun 5 at 16:56
  • $\begingroup$ Is it easy to construct / find an example of such SES that does not split? $\endgroup$ Commented Jun 6 at 12:00
  • $\begingroup$ Yes, it’s rather elementary. Think of 0 < 1. $\endgroup$ Commented Jun 6 at 12:39

0

You must log in to answer this question.