Cluster automorphism group of braid varieties

Technical Summary: Cluster Automorphism Group of Braid Varieties

Problem Statement
Braid varieties, denoted $X_{u,\beta}$, are affine algebraic varieties indexed by a permutation $u \in S_n$ and a positive braid word $\beta$. Recent works have established that the coordinate rings of these varieties admit cluster algebra structures, constructed via combinatorial models such as 3D plabic graphs. While the cluster structure is known, the explicit description of the cluster automorphism group, $\text{Aut}(A)$, and its action on the braid variety remains a central object of study. The cluster automorphism group consists of algebra automorphisms that scale cluster variables. Understanding this group is significant for the study of cluster deep loci and the "no mysterious point" conjecture. The primary challenge addressed in this paper is to provide an explicit, computable description of $\text{Aut}(A(X_{u,\beta}))$ and its action on the initial seed variables, specifically by characterizing the kernel of the extended exchange matrix associated with the braid variety.

Methodology
The paper proceeds by leveraging the combinatorial framework of 3D plabic graphs introduced in prior literature (specifically [17]) to construct the necessary algebraic objects.

1. Combinatorial Construction: The author utilizes the 3D plabic graph $G_{u,\beta}$ associated with the pair $(u, \beta)$. From this graph, "soap films" are defined, which correspond to the vertices of an initial quiver $Q_{u,\beta}$. These vertices are partitioned into mutable ($m$) and frozen ($f$) sets.
2. Matrix Definitions:
   • The half-weight matrix $H_{u,\beta}$ is constructed based on half-arrows in the quiver derived from the soap films.
   • A boundary correction matrix $D_{u,\beta}$ is defined using a boundary map $\partial_{u,\beta}$ that assigns vectors in $\mathbb{Z}^{n-1}$ to soap films based on the boundary regions they cover.
   • The central object of the study is the square matrix $\tilde{B}_{u,\beta} = H_{u,\beta} + D_{u,\beta}$, which serves as an invertible extension of the standard rectangular extended exchange matrix $\tilde{B}$ (of size $m \times (m+f)$).
3. Inductive Proofs: The properties of $\tilde{B}_{u,\beta}$ are established via induction on the length of the braid word $\beta$. The proof distinguishes between two cases:
   • Case A: The first letter of $\beta$ is not part of the positive distinguished subexpression for $u$.
   • Case B: The first letter is part of the subexpression, corresponding to the leftmost frozen bridge.
     In both cases, the author relates the matrices of the current configuration to those of a reduced configuration, utilizing row/column operations and properties of reflection transformations on the boundary vectors.
4. Group Action Derivation: Using the inverse matrix $A_{u,\beta} = \tilde{B}_{u,\beta}^{-1}$, the paper applies results from Lam and Speyer [13] to describe the cluster automorphism group. Specifically, the group is identified with an algebraic torus whose action is determined by the column vectors of $A_{u,\beta}$ corresponding to the frozen vertices.

Key Contributions and Results

• Theorem 1 (Integrality and Determinant): The matrix $\tilde{B}_{u,\beta}$ is proven to be an integer matrix with determinant $\det(\tilde{B}_{u,\beta}) = (-1)^{m+f}$. Consequently, its inverse $A_{u,\beta}$ is also an integer matrix. This confirms that the integer span of the columns of the original exchange matrix $\tilde{B}$ is $\mathbb{Z}^m$, implying the quiver has "really full rank."
• Explicit Description of the Automorphism Group (Corollary 1.1 / Corollary 4.5): The cluster automorphism group $\text{Aut}(A(Q_{u,\beta}))$ is shown to be an algebraic torus $(\mathbb{C}^\times)^f$. The action on the initial cluster variables $x = (x_1, \dots, x_{m+f})$ is explicitly given by:
  $$x_i \mapsto \left( \prod_{j=1}^f t_j^{a_{i, m+j}} \right) \cdot x_i$$
  where $t_1, \dots, t_f$ are the torus coordinates and $a_{i, m+j}$ are the entries of the inverse matrix $A_{u,\beta}$.
• Inductive Factorization (Lemma 4.7): The paper provides an inductive formula for computing the inverse matrix $A_{u,\beta}$ based on the structure of the braid word, facilitating practical computation.
• The "Sign Phenomenon": Through computational examples, the author observes that in many cases (specifically when $u$ is the identity), all non-zero entries of $A_{u,\beta}$ are non-positive. However, the paper provides a counterexample (Example 4.10) demonstrating that this phenomenon does not hold universally, as mixed signs can occur.

Significance and Claims
The paper claims to provide the first explicit, combinatorial description of the cluster automorphism group for general braid varieties, moving beyond the specific case where $u$ is the identity. By translating the construction into the language of 3D plabic graphs, the author offers a computational advantage, allowing for the use of existing software (such as Galashin's program) to generate examples.

The work highlights that the dimension of the cluster automorphism group ($f$) can be strictly larger than $n-1$ (the dimension of the standard torus action on braid varieties), indicating a richer symmetry structure than previously accounted for in general settings. The paper concludes by posing open problems regarding the geometric meaning of the "sign phenomenon" and the request for a combinatorial description of the non-zero entries of $A_{u,\beta}$, framing these as directions for future inquiry rather than resolved results. The author acknowledges that the matrix construction is motivated by and essentially equivalent to extensions defined in related works (e.g., [3]), but emphasizes the utility of the current formulation for computation and inductive description.