2
$\begingroup$

Let $Q$ be the quiver:

$$\begin{array}{ccccc} & & \\ 1 & \rightrightarrows & 2 & \rightrightarrows &3 & \end{array}$$

I want to find all simple modules over the complex path algebra of $Q$.

In class we defined the category $Q\textbf{-Mod}$ of modules $M$ over a quiver $Q$. Where $Q\textbf{-Mod} \ni M$ consists of vector spaces $M_i$ for every vertex $i \in Q_0$ and $k$-linear maps $M(\alpha) \colon M(i) \to M(j)$ for every $\alpha \in Q_1$ with morphisms between any two such modules $M,N$: $\phi \colon M\to N$ is a collection $\{\phi_i\}_{i\in Q_0}, \phi_i \colon M(i) \to N(i)$ such that the obvious square commutes. We also proved that we have a bijection between $Q\textbf{-Mod}$ and $K[Q]\textbf{-Mod}$, the category of modules over the path algebra of $Q$ over $K$.

Since we have this bijection (which I believe actually gives us equivalence of categories?) we can work with $Q\textbf{-Mod}$ instead of $K[Q]\textbf{-Mod}$, right?

Then $M\in Q\textbf{-Mod}$ is said to be simple if its only submodules are $M$ and $0$. So if the dimension of the direct sum $\bigoplus_{i\in Q_0} M(i)$ is equal $1$ then $M$ must be simple so I think that

$$\begin{array}{ccccc} & & \\ \mathbb{C} & \rightrightarrows & 0 & \rightrightarrows &0 & \end{array}$$

$$\begin{array}{ccccc} & & \\ 0 & \rightrightarrows & \mathbb{C} & \rightrightarrows &0 & \end{array}$$

$$\begin{array}{ccccc} & & \\ 0 & \rightrightarrows & 0 & \rightrightarrows & \mathbb{C} & \end{array}$$

are all simple. But are there any other such simple modules? Or rather is this even correct? It looks too simple. To be honest I find this very hard and confusing. Is it better to work directly with modules over $\mathbb{C}[Q]$ instead?

$\endgroup$
1
$\begingroup$

If your quiver $Q$ has no oriented cycles (and only in this case), then it is true that the simple representations of $Q$ are exactly the one-dimensional representations of $Q$ (so there is exactly one for every vertex of $Q$).

One way to see this is to use Nakayama's lemma: Let $R$ be a unital (not necessarily commutative) ring, and let $J$ be its Jacobson radical. If $M$ is a finitely-generated right $R$-module, then $M\cdot J$ is a proper submodule of $M$.

Now, let $M$ be a simple $\mathbb{C}[Q]$-module (this implies that $M$ is finitely-generated). The Jacobson radical $J$ of $\mathbb{C}[Q]$ is the two-sided ideal generated by the arrows of $Q$. Since $M\cdot J$ is a proper submodule of $M$ (by Nakayama's lemma) and $M$ is simple, we must have $M\cdot J = 0$. In other words, the action of all the arrows of $Q$ on $M$ is the zero action.

This means that in the $Q$-representation corresponding to $M$, all the linear maps attached to arrows of $Q$ are zero maps. In this case, the only way for $M$ to be simple is to be one-dimensional.


Some comments:

  • Since we have this bijection (which I believe actually gives us equivalence of categories?)

Indeed, it is an equivalence of categories.

  • Is it better to work directly with modules over $\mathbb{C}[Q]$ instead?

It depends on the situation. The best is to know both categories.

$\endgroup$
0
$\begingroup$

By Gabriel's theorem, the indecompsable module are classified by positive roots. For your case A_3, they are like a consecutive string of dimension 1 vector spaces. As a result, the only simple modules are 1 dimension at 1 vertex and 0 otherwise. You assertion is correct. The three are the only simple module for A_3.

$\endgroup$
2
  • $\begingroup$ But is my example really $A_3$? My reason for asking is that when trying to research quivers a bit my example looks more like a "extented " Kronecker quiver or doesn't the extra arrows matter when searching for simple modules? Sorry if this is naive. $\endgroup$ – Martin Garb Dec 7 '16 at 9:43
  • $\begingroup$ Indeed, the OP's quiver is not of type $A_3$. $\endgroup$ – Hugh Thomas Dec 24 '16 at 2:12

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.