1
$\begingroup$

Let $A$ be a Nakayama algebra, that is an Artin algebra such that any indecomposable module has a unique composition series. The easiest examples of such algebras are $K[x]/(x^n)$.

Question 1: In case $Ext^1(X,Y) \neq 0$ for two indecomposable $A$-modules $X,Y$, can one write down an explicit nice non-split short exact sequence $0 \rightarrow Y \rightarrow Z \rightarrow X \rightarrow 0$, where $Z$ is constructed in some canonical way out of $X$ and $Y$?

Question 2: Can such an $Z$ as above have at most 2 indecomposble summands?

$\endgroup$

1 Answer 1

2
$\begingroup$

I'll answer Question 2 first. The answer is yes.

If $Z$ has more than two indecomposable summands, then $\text{soc}(Z)$ has at least three summands, but $X$ has simple socle since it's uniserial, so $Y$, the kernel of $Z\to X$, has at least two summands. But $Y$ is also uniserial, and so has simple socle.

For Question 1:

If $Z$ is indecomposable (uniserial), then it's easy to see what the exact sequence must be, and when it exists, so suppose $Z=Z_1\oplus Z_2$ is a direct sum of two indecomposables.

An indecomposable is determined by the list of its composition factors from head to socle, which can be represented by a word $w$ in a set indexing the isomorphism classes of simples. Denote the module corresponding to a word $w$ by $[w]$, if there is such a module.

Since $Y$ has simple socle, its map to one of the $Z_i$, say $Z_1$, must be injective, but not surjective since the exact sequence doesn't split. But then its map to $Z_2$ must be surjective, or else $X$, the cokernel of $X\to Z$, has nonsimple head.

So if $X=[w]$ and $Y=[w']$ then the extensions all look like the following, with the obvious maps.

If $[ww']$ is a module, then we can take $Z=[ww']$

If $w=uv$ and $w'=vt$, where $[uvt]$ is a module, then $Z$ could be $[uvt]\oplus[v]$.

There could be more than one choice that works.

$\endgroup$
1
  • $\begingroup$ Thanks. This looks easier than expected and seems to work for general algebras with X and Y uniserial. Note that it is stated as an open problem to decide when Z is in this case also uniserial in the open problem section (number (3)) in the book of Auslander,Reiten and Smalo. $\endgroup$
    – Mare
    Oct 24, 2019 at 18:06

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.