Using cluster categories to show that seed of acyclic cluster algebra is determined by the cluster

43 Views Asked by At

In corollary 2 of the article "From triangulated categories to cluster algebras II" by Caldero and Keller, it is stated that a seed $(u,Q)$ (where $u$ is a cluster and $Q$ a quiver) of an acyclic cluster algebra $\mathcal{C}$ is determined by $u$. (i.e., two different clusters can't be associated with the same quiver in the exchange graph.) The authors say that this is an immediate consequence of the correspondence (Theorem 4 of the same article) between cluster variables/clusters in the cluster algebra and indecomposable objects/cluster-tilting objects of the cluster category.

I assume this may follow from the fact that cluster-tilted algebras $A_1=\text{End}_{\mathcal{C}}(T_1)$, $A_2=\text{End}_{\mathcal{C}}(T_1)$ derived from two different cluster-tilting objects $T_1,T_2$ in the cluster category, will have two different associated quivers $Q_{A_1}$, $Q_{A_2}$. Is this statement true? and if so, how can it be shown? If it is not, then why is the above corollary an immediate consequence of the theorem?

1

There are 1 best solutions below

0
On BEST ANSWER

This is not the correct implication: for example, taking $T_2$ to be the shift of $T_1$ will make $A_1$ and $A_2$ isomorphic, even though $T_1$ and $T_2$ will be different.

Instead, the reasoning is as follows: let $(u,Q)$ be a seed, with $u = (u_1, \ldots, u_n)$ a cluster. Then by Theorem 4, for each $i=1,\ldots, n$, there is a unique object $M_i$ without self-extension in the cluster category (up to isomorphism) such that $X_{M_i} = u_i$. The quiver $Q$ is then the Gabriel quiver of the endomorphism algebra of $M_1 \oplus \cdots \oplus M_n$. Thus $Q$ is determined by $u$.