Jordan types for pairs of commuting nilpotent matrices: A survey

Technical Summary: Jordan Types for Pairs of Commuting Nilpotent Matrices: A Survey

Problem Statement
The paper addresses the classification of pairs of integer partitions $(P, Q)$ that arise as the Jordan types of two commuting nilpotent matrices over an infinite field. While the set of all possible Jordan types for a single nilpotent matrix is well-understood (corresponding to all partitions of $n$), the characterization of pairs $(P, Q)$ such that there exist commuting nilpotent matrices $A$ and $B$ with $P_A = P$ and $P_B = Q$ remains a complex, largely open problem. The paper surveys existing results, distinguishes between positive and negative cases, and provides a detailed outline of the proof for the "Box Conjecture," which describes the structure of the inverse image of the dominant partition map.

Methodology and Framework
The survey utilizes tools from linear algebra, combinatorics, and algebraic geometry. Key methodological components include:

• The Nilpotent Commutator: For a nilpotent matrix $B$ of type $P$, the set $N_B = \{A \in M_n(F) : BA=AB, A \text{ nilpotent}\}$ is an irreducible variety.
• The Dominant Map ($D$): Introduced by Panyushev, the map $D: \mathcal{P} \to \mathcal{P}$ assigns to a partition $P$ the Jordan type of the dense orbit within $N_B$. The image of $D$ is the set $\mathcal{Q}$ of "Rogers-Ramanujan" (RR) partitions (partitions where parts differ by at least 2).
• The Burge Correspondence: A bijection between the set of all partitions $\mathcal{P}$ and a set of binary words (using operators $\alpha$ and $\beta$). This correspondence is central to the proof of the Box Conjecture.
• The Descent Map: A combinatorial map derived from the Burge code that characterizes the structure of the inverse image $D^{-1}(Q)$.
• Invariant Subspaces: The proof relies on Shayman's description of invariant subspaces of nilpotent matrices to link the algebraic structure of the commutator to the combinatorial properties of the Burge code.

Key Contributions and Results

1. Survey of Commuting Pairs:
   The paper aggregates scattered results regarding specific classes of commuting partitions:

   • One Jordan Block: A partition $P$ commutes with $(n)$ if and only if $P$ is an "almost rectangular" (AR) partition (parts differ by at most 1).
   • Universally Commuting: For $n \ge 4$, only partitions of the form $(2^k, 1^l)$ commute with all other partitions.
   • Conjugates: A partition $P$ always commutes with its conjugate $P^T$, though conjugates of commuting pairs do not necessarily commute.
   • Refinements: If two partitions are AR refinements of a common partition, they commute.
   • Negative Results: Two distinct RR partitions never commute. The paper notes a new observation that two non-equal RR partitions are not commuting.

2. The Box Theorem (Proof Outline):
   The central contribution is the outline of the proof for the Box Conjecture (formulated by Iarrobino et al.), which characterizes the set $D^{-1}(Q)$ for any $Q \in \mathcal{Q}$.

   • Structure: The set $D^{-1}(Q)$ forms a "box" (a rectangular array) of size $\delta_1 \delta_2 \cdots \delta_k$, where $\delta_i$ are derived from the parts of $Q$.
   • Bijection: There is a bijection between the points in this box (indexed by coordinates $(i_1, \dots, i_k)$) and the partitions in $D^{-1}(Q)$.
   • Burge Code Characterization: A partition $P$ belongs to $D^{-1}(Q)$ if and only if its Burge code $\Omega(P)$ has a specific form determined by the coordinates of the box. Specifically, the code consists of alternating blocks of $\alpha$ and $\beta$ where the lengths are determined by the box dimensions and the coordinates.
   • Equivalence of Maps: The proof establishes that the geometric map $D$ (dominant orbit) coincides with the combinatorial "descent map" derived from the Burge correspondence. This is achieved by showing that the Jordan type of the restriction of a nilpotent matrix to a generic invariant subspace corresponds to the operator $\partial$ in the Burge chain.

3. Limitations of the Burge Correspondence:
   The paper highlights that while the Burge correspondence successfully describes the set $D^{-1}(Q)$, it does not fully determine commutativity within that set. Recent examples (joint work with Khatami) show that distinct partitions within the same "box" (sharing the same $D(P)$) do not necessarily commute. Thus, the Burge correspondence is a necessary but not sufficient tool for characterizing all commuting pairs.

Significance and Open Questions
The paper positions the Box Theorem as the central pillar of current understanding regarding commuting nilpotent matrices. It clarifies the structure of the inverse image of the dominant map, resolving a conjecture that had been open since approximately 2012.

However, the author emphasizes that the problem of a complete characterization of all commuting pairs $(P, Q)$ remains unsolved. The paper concludes by listing open questions:

• Complete Characterization: No general algorithm exists to test commutativity for arbitrary pairs.
• Asymptotics: The behavior of the proportion of commuting pairs as $n \to \infty$ is unknown.
• Algebraic Structure: The relationship between the Burge correspondence and the specific non-commuting pairs within a "box" requires further investigation, potentially involving inverse semigroup theory.
• Geometric Stratification: A generalization of the Gerstenhaber-Hesselink Theorem to the stratification of the nilpotent commutator variety is an open geometric problem.
• Other Lie Algebras: The extension of these combinatorial frameworks to simple Lie algebras of types B, C, and D is an open area, complicated by the fact that the variety of commuting pairs is not irreducible in those cases.

The survey serves as a complement to Khatami's earlier work, specifically by including the proof of the Box Conjecture and a broader collection of negative results and open problems.