Skew-symmetric cluster algebras of finite mutation type (Q437814)

Cluster algebras, developed by S. Fomin and A. Zelevinsky in 2000, are commutative algebras, whose generators and relations are constructed by a recursive process. The generators are collected into groups of \(n\) elements called clusters connected by local transition rules determined by a skew-symmetrizable integer matrix associated to each cluster, called the exchange matrix. The classification of cluster algebras of finite type was achieved by \textit{S. Fomin} and \textit{A. Zelevinsky} [Invent. Math. 154, No. 1, 63--121 (2003; Zbl 1054.17024)] who proved their correspondence with root systems of simply laced Dynkin diagrams. A wider class is formed by cluster algebras of finite mutation type, which have finitely many exchange matrices but are allowed to be of infinite type. The main result of this paper is the classification of all cluster algebras of finite mutation type with skew-symmetric exchange matrices. Besides cluster algebras of rank 2 and cluster algebras arising from triangulations of surfaces, there are exactly 11 exceptional skew-symmetric cluster algebras of finite mutation type. More precisely, nine of them are associated with root systems \(E_6, E_7, E_8, \widetilde{E}_6, \widetilde{E}_7, \widetilde{E}_8, E_6^{(1,1)}, E_7^{(1,1)}, E_8^{(1,1)}\). The remaining two, called \(X_6\) and \(X_7\), were found by \textit{H. Derksen} and \textit{T. Owen} [Electron. J. Comb. 15, No. 1, Research Paper R139, 15 p. (2008; Zbl 1180.05052)] and are respectively of rank 6 and 7. The authors reformulate the classification problem in terms of quivers by assigning to every exchange matrix an oriented weighted graph. Previous results of Derksen and Owen state that a quiver is mutation-finite if and only if no quiver in its mutation class contains an edge of weight greater than 2. The main idea of the authors is to show that all mutation-finite quivers, except a finite number of them, have the property of being decomposable into five different types of blocks. Results of \textit{S. Fomin, M. Shapiro} and \textit{D. Thurston} [Acta Math. 201, No. 1, 83--146 (2008; Zbl 1263.13023)] state that block-decomposable quivers are in one-to-one correspondence with quivers arising from triangulations of surfaces, which are known to be mutation-finite. Hence the classification problem is reduced to finding all mutation-finite non-decomposable quivers. The authors find several conditions on subquivers of a quiver implying its block-decomposability. This allows them to show that any quiver which is non-decomposable and minimal for the inclusion contains at most seven vertices. Such quivers can be entirely classified using a computer program and are shown to be mutation equivalent to \(E_6\) or \(X_6\). Then all non-decomposable mutation-finite quivers with less than 10 vertices are found using the same computer assisted procedure. They also get that no mutation-finite non-decomposable quivers of order 11 exists, which implies that the order of any non-decomposable mutation-finite quiver does not exceed 10, completing the proof. The authors also deduce that all minimal quivers with infinite mutation class contains at most ten vertices. This gives rise to a criterion to check in polynomial time if a quiver is mutation-finite. Namely, a quiver of order greater than 10 is mutation-finite if and only if all subquivers of order 10 are mutation-finite. Other results of the paper include a discussion on the growth rate of the associated cluster algebras and the study of quivers of order 3.