Technical Summary: A Greedy Open-Orbit Criterion for Solvable Algebraic Group Actions

1. Problem Statement

The paper addresses two primary problems in the representation theory of algebraic groups and the geometry of nilpotent varieties:

1. The General Problem: Given a connected solvable algebraic group $G$ acting rationally on a finite-dimensional vector space $U$, determine whether $U$ admits an open (dense) $G$-orbit (i.e., whether $U$ is a prehomogeneous vector space, or PVS). If such an orbit exists, the paper seeks to construct a vector with an open orbit that has the minimum possible support cardinality and to characterize the generic stabilizer.
2. The Specific Application: Apply this criterion to the action of the automorphism group $\text{Aut}_Q(M)$ on the dual of the extension space $\text{Ext}^1_Q(M, M)^*$, where $M$ is a multiplicity-free representation of a Dynkin quiver $Q$. The goal is to provide a combinatorial and linear algebraic algorithm to decide the rigidity of the corresponding irreducible component of Lusztig's nilpotent variety $\Lambda(V)$. A component is rigid if it contains an open orbit.

2. Methodology

2.1. The Greedy Successive-Quotient Procedure

The core methodological contribution is a "greedy" algorithm that operates on a $G$-stable complete flag.

• Setup: By the Lie–Kolchin theorem, there exists a complete flag $U = V_0 \supset V_1 \supset \dots \supset V_d = 0$ of $G$-submodules. A basis of eigenvectors $e_1, \dots, e_d$ is chosen such that $e_i \in V_{i-1} \setminus V_i$, with associated characters $\chi_i \in X^*(T)$ (where $T$ is a maximal torus).
• The Algorithm: The procedure iterates through $k = 1, \dots, d$. At each step, it maintains a set of "active" indices $\Phi_{k-1}$.
  • Decision: It checks if the $G$-orbit of a vector with support $\Phi_{k-1}$ is dense modulo $V_k$.
  • Activation: If the orbit is not dense modulo $V_k$, the index $k$ is added to the active set ($\Phi_k = \Phi_{k-1} \cup \{k\}$).
  • Abort Condition: If adding $k$ causes the set of characters $\{\chi_i : i \in \Phi_k\}$ to become $\mathbb{Z}$-dependent, the procedure aborts.
  • Continuation: If the characters remain independent, the process continues.
• Outcome:
  • Success: If the procedure completes all steps, the resulting set $\Phi$ defines a vector with an open orbit. The size of $\Phi$ is the minimum support cardinality for such vectors.
  • Failure: If the procedure aborts, no open orbit exists.
• Infinitesimal Version: A linear algebra variant is formulated to detect separable open orbits. This replaces the geometric density check with a rank test on the differentials of the action maps and checks for linear independence of the differentials of the characters in the Lie algebra dual.

2.2. Application to Lusztig's Nilpotent Varieties

The paper applies the general procedure to the specific context of Dynkin quivers:

• Context: Let $M$ be a multiplicity-free module over the path algebra of a Dynkin quiver $Q$. The group $G = \text{Aut}_Q(M)$ is solvable. The space $E = \text{Ext}^1_Q(M, M)^*$ carries a natural $G$-action.
• Rigidity Connection: The irreducible component $Z_M$ of Lusztig's nilpotent variety corresponding to $M$ is rigid if and only if $G$ has a dense orbit on $E$.
• Combinatorial Reformulation:
  • The author constructs a "block basis" for $E$ adapted to a containment order on pairs of indecomposable summands $(i, j)$ where $\text{Ext}^1(M_i, M_j) \neq 0$.
  • The characters of the torus action on these blocks correspond to differences of weights $\epsilon_i - \epsilon_j$.
  • The abort condition (linear dependence of characters) translates combinatorially: the procedure aborts if and only if the set of active extension coordinates forms a cycle in a graph with vertices corresponding to the indecomposable summands.
  • Thus, rigidity is equivalent to the existence of a "forest" of active extension coordinates that satisfies a rank condition.

2.3. Specialization to Type A

For equioriented type $A_m$ quivers, the problem is encoded using incidence matrices.

• Multiplicity-free modules correspond to multisegments, which can be represented by incidence matrices (0-1 matrices).
• The rigidity of the component is determined by running the greedy procedure on the incidence matrix.
• The active set $\Phi$ corresponds to a forest on the set of "ones" in the matrix.
• The paper provides an explicit algorithm using the matrix $M(\lambda)$ (representing the commutator map) to test for linear independence of rows, effectively performing the rank test.

3. Key Contributions and Results

3.1. Theoretical Results

• Theorem 1.1 & 2.4: The greedy successive-quotient procedure correctly decides the existence of an open orbit for any rational representation of a connected solvable group. If successful, it constructs a vector with minimum support and identifies the image of the generic stabilizer in the maximal torus quotient.
• Theorem 2.12: The infinitesimal (linear algebra) version of the procedure correctly detects open separable orbits. In characteristic zero, this coincides with the group-level procedure.
• Theorem 3.4: For a multiplicity-free Dynkin quiver module $M$, the component $Z_M$ is rigid if and only if the greedy procedure on $\text{Ext}^1_Q(M, M)^*$ succeeds.
  • Combinatorial Criterion: Success is equivalent to the active edges forming a forest (acyclic graph) on the set of indecomposable summands.
  • Decomposition: The connected components of this active forest determine the generic decomposition of the corresponding $\Pi$-module (preprojective algebra module) into indecomposable summands.

3.2. Results for Type A and Incidence Matrices

• Theorem 4.4 & 4.5: An explicit algorithm is provided for incidence matrices. Rigidity is equivalent to the matrix $M(\lambda)$ (constructed from the active set) having full rank, and the active set forming a forest.
• Theorem 4.12: For permutation matrices (corresponding to smooth permutations), rigidity, strong rigidity, and the smoothness of the permutation (avoiding patterns 4231 and 3412) are equivalent conditions.
• Strong Rigidity: The paper defines "strong PVS" and "strong rigidity." It shows that for incidence matrices, strong rigidity holds if and only if the active set consists precisely of the "covers" in the containment order.
• Empirical Data: Table 1 presents the percentage of rigid and strongly rigid incidence matrices for $n \le 10$ ones, showing that while most small matrices are rigid, the percentage drops significantly as $n$ increases.

4. Significance and Claims

The paper claims the following significance:

• Uniformity: It provides a uniform, deterministic, and efficient algorithm for deciding prehomogeneity (existence of open orbits) for any rational representation of a connected solvable algebraic group, a class that includes many non-reductive cases often difficult to handle.
• Reduction to Combinatorics: In the context of Lusztig's nilpotent varieties for Dynkin quivers, the geometric problem of rigidity is reduced to a linear algebra rank test and a simple graph-theoretic condition (acyclicity/forest condition).
• Structural Insight: The algorithm does not merely decide rigidity; it constructs the generic indecomposable decomposition of the associated $\Pi$-module directly from the connected components of the active forest.
• Characteristic Independence: The author notes that for the specific case of incidence matrices with up to 10 ones, the rigidity verdict appears independent of the characteristic of the field, as the rank tests yield unimodular matrices over the integers.
• Connection to Known Results: The paper recovers and strengthens known results regarding smooth permutations and Schubert varieties (Theorem 4.12), linking the geometric notion of rigidity to the combinatorial notion of pattern avoidance.

The paper explicitly avoids claiming that the method solves the rigidity problem for general (non-multiplicity-free) modules or for non-Dynkin quivers, noting that the multiplicity-free case is the specific context where the solvable group action allows for this greedy flag-based approach. It also leaves the general combinatorial criterion for arbitrary incidence matrices as an open problem, offering a conjecture regarding secluded submatrices.