Quiver with Potential: Structure & Mutation

• Quiver with potential is a pair (Q, W) that combines a finite directed graph with a cyclic superpotential to encode complex algebraic and geometric structures.
• Mutation operations reverse arrows and introduce composite arrows, ensuring non-degeneracy and maintaining essential Jacobian algebra properties.
• Applications span representation theory, Donaldson–Thomas invariants, cluster categories, and noncommutative geometry in mathematical physics.

A quiver with potential is a pair $(Q, W)$ where $Q$ is a finite directed graph (a quiver) and $W$ is a formal linear combination of cyclic paths (cycles) in $Q$, considered up to cyclic equivalence. Such data encode both the combinatorial structure of the quiver and additional "superpotential" information, which is crucial in representation theory, noncommutative algebraic geometry, and the theory of Calabi–Yau (CY) categories. The algebraic structure defined by a quiver with potential—specifically, its Jacobian algebra—governs mutation theory, cluster categories, Donaldson–Thomas invariants, and the categorical and enumerative structures that underlie much of modern representation theory and mathematical physics.

1. Formal Definition and Fundamental Properties

A quiver $Q$ consists of a finite vertex set $Q_0$ and a finite set of arrows $Q_1$. The path algebra $kQ$ over a field $k$ has a basis given by oriented paths in $Q$. A potential $Q$0 is a (possibly infinite) $Q$1-linear combination of oriented cycles, taken in the quotient $Q$2 to enforce cyclic equivalence: only the cycles, up to cyclic rotation, matter.

The Jacobian algebra is defined as

$Q$3

where $Q$4 is the cyclic derivative of $Q$5 with respect to the arrow $Q$6; it sums all ways of removing an occurrence of $Q$7 within each cycle, and concatenating the remaining path.

A potential is called reduced if it contains no two-cycles (i.e., no cyclic paths of length two). The Splitting Theorem (Derksen–Weyman–Zelevinsky) states that every quiver with potential is right-equivalent to a direct sum of a reduced part and a trivial two-cycle part; the latter can be discarded without altering the Jacobian algebra (Völcsey et al., 2010).

2. Mutation and Non-degeneracy

Mutation is an elementary operation $Q$8 at a vertex $Q$9 (provided $W$0 is not contained in a two-cycle or loop). The mutated quiver and potential $W$1 are constructed via specific combinatorial rules: arrows incident to $W$2 are reversed, new arrows ("composite arrows") are introduced to connect sources and targets of 2-step paths through $W$3, and the potential is transformed by substitution and addition of new cycles composed of the new arrows; trivial two-cycles are then split off to yield the reduced mutated pair (Mizuno, 2012, Abrikosov, 2017).

A pair $W$4 is called non-degenerate if under all admissible sequences of iterated mutations, neither loops nor two-cycles are ever created in the reduced part. Non-degeneracy is critical: it ensures that mutation is always defined, underpins the construction of well-behaved Ginzburg DG-algebras, and guarantees the finiteness and structure of cluster categories (Völcsey et al., 2010).

3. Structural Theorems and Classification of Non-degenerate Potentials

A central result is Theorem 3.1 (Völcsey et al., 2010), providing a powerful criterion for non-degeneracy. Suppose $W$5 is a $W$6-graded, connected quiver with a reduced homogeneous potential $W$7 of degree $W$8, and let $W$9 be the graded-complete Jacobian algebra. If:

1. $Q$0 has at least three vertices;
2. $Q$1 for all $Q$2, and $Q$3 for $Q$4;
3. $Q$5 is a 3-Calabi–Yau order over its noetherian center and almost Azumaya;
4. Hochschild 0-homology $Q$6 has no nonzero homogeneous elements of degree $Q$7;

then $Q$8 contains neither loops nor two-cycles, and all iterated mutations preserve these properties—so $Q$9 is non-degenerate.

Applications:

• Cyclic McKay quivers in dimension 3 with freely acting cyclic subgroups of $Q$0 and cubic potential have non-degenerate potentials. The Jacobian algebra is the skew group algebra $Q$1 and is 3-CY (Völcsey et al., 2010).
• Del Pezzo helix quivers with potentials from geometric helices also satisfy the criteria, yielding a broad class of non-degenerate potentials.

These results extend non-degeneracy to previously inaccessible families beyond classic acyclic or surface-type examples.

4. Derived and Cluster Category Implications

For non-degenerate quivers with potential, one can always form associated Ginzburg DG-algebras, which are 3-Calabi–Yau. The derived (perfect) categories of these DG-algebras, modulo finite-dimensional complexes, realize the cluster categories central to cluster algebra theory. Non-degeneracy ensures mutations induce derived equivalences and that Jacobian algebras across the mutation class have compatible Calabi–Yau structures (Völcsey et al., 2010, Mizuno, 2012).

Specifically, for selfinjective Jacobian algebras $Q$2, Okuyama–Rickard (silting) complexes provide an explicit mechanism for passing between derived equivalent algebras. Mutation of the complex mirrors mutation of the quiver-with-potential, and their endomorphism rings are isomorphic to the mutated Jacobian algebra. If the Nakayama permutation fixes the set $Q$3 of mutated vertices, these are actual tilting complexes yielding explicit derived equivalences (Mizuno, 2012).

5. Classification in Surface and Braid Semigroup Contexts

For quivers $Q$4 arising from triangulations of closed oriented surfaces with punctures, non-degenerate potentials can be classified:

• For a once-punctured surface of genus $Q$5, each triangulation quiver admits infinitely many non-degenerate potentials $Q$6 (where $Q$7 is the puncture cycle), all pairwise not weakly right-equivalent (Geuenich et al., 2020).
• For twice-punctured surfaces, there is exactly one weak right-equivalence class: $Q$8.

The braid semigroup construction generalizes the assignment of quivers with potentials to elements of a simply-laced Coxeter group, with primitive potentials classified up to right-equivalence by the second cohomology of an explicit 2-dimensional CW-complex associated to the quiver and its cycles (Abrikosov, 2017).

6. Applications to Invariants and Representation Theory

Quivers with non-degenerate potentials occupy a central role in the computation and wall-crossing of Donaldson–Thomas (DT) invariants, cohomological Hall algebras, and the structure and combinatorics of cluster algebras. The non-degeneracy condition enables:

• Consistent definition of scattering diagrams and identification of their wall-crossings with DT-invariant generating functions; for many cases, the DT and cluster scattering diagrams coincide (Mou, 2019).
• Mutation-invariance of DT invariants, explicit mutation formulas, and compatibility with wall-crossing formulas in Hall algebras.
• Construction and categorification of CY$Q$9 $Q_0$0-categories, explicit realization of cluster exchange relations, and anti-cluster algebra structures (Hua et al., 2019).
• Representation-theoretic encoding via the combinatorics of finite-codimension ideals in the Jacobian algebra, connection to BPS algebras, and realization of vacuum characters as DT series (Li, 2023).

These features unify the algebraic, geometric, and enumerative aspects of 3-CY categories, categorified cluster structures, and supersymmetric gauge theoretical systems.

7. Future Directions and Broader Impact

The technical framework for establishing non-degeneracy, grounded in graded CY-orders and Hochschild degree constraints, paves the way for the classification of non-commutative crepant resolutions, advances higher-dimensional cluster theory, and influences the study of stability conditions and wall-crossing structures in derived categories. Further applications are expected in the context of mirror symmetry, enumerative Calabi–Yau geometry, and representation theory of quantum groups via their associated cohomological Hall algebras (Völcsey et al., 2010, Davison, 2013).

The unification and extension of mutation-theoretic, categorical, and enumerative methods through non-degenerate quivers with potential continue to provide rich structure and avenues for exploration across algebra, geometry, and mathematical physics.