Regular Matroid Generalization

Updated 11 April 2026

Regular matroid generalization is a framework that extends classical regular matroids by incorporating sixth-root-of-unity, P-regular, and orthogonal matroid classes with diverse algebraic representations.
It unifies combinatorial, geometric, and topological perspectives by generalizing invariants such as the Tutte polynomial, Jacobians, and symmetric edge polytopes into functorial and categorical settings.
The generalization supports robust algorithmic and polyhedral applications, enabling efficient decomposition, matching, and spanning structure analysis in complex combinatorial systems.

A regular matroid generalization extends the robust theory of regular matroids—those representable over every field—into wider combinatorial, algebraic, and topological frameworks. This generalization encompasses classes such as sixth-root-of-unity matroids, regular orthogonal matroids, and $P$ -regular matroids defined via foundations, and connects to generalizations of invariants (e.g., Jacobians, Tutte polynomials, symmetric edge polytopes) and operation-closed classes from a functorial and structural perspective. The resulting landscape captures the structural essence of classical regular matroids and their extensions, and provides unified machinery for combinatorics, geometry, and optimization.

1. Sixth-Root-of-Unity Matroids and Complex-Unimodular Generalization

A primary regular matroid generalization is the class of sixth-root-of-unity matroids (√[6]{1}-matroids), also known as complex-unimodular matroids. A matroid $\mathcal{M}$ of rank $r$ on a ground set $E$ of size $n$ is a sixth-root-of-unity matroid if there exists a matrix $M\in\mathbb{C}^{r\times n}$ such that every maximal minor (determinant of an $r\times r$ submatrix) is a sixth root of unity or zero, i.e., $\det M_I\in\mu_6\cup\{0\}$ for all $I\subset E$ , $|I|=r$ , where $\mathcal{M}$ 0.

Regular matroids correspond to the case where every maximal minor is in $\mathcal{M}$ 1; thus, the √[6]{1} generalization replaces the constraint $\mathcal{M}$ 2 for every basis $\mathcal{M}$ 3 by $\mathcal{M}$ 4, i.e., $\mathcal{M}$ 5 for all bases. This characterization is strictly broader than regular matroids while preserving analogous combinatorial and representational properties (Baker et al., 2023).

2. Foundations, Pastures, and $\mathcal{M}$ 6-Regular Classes

The use of pastures and foundations provides a categorical generalization of regularity. A matroid $\mathcal{M}$ 7 is associated with a foundation $\mathcal{M}$ 8—a canonical pasture encoding universal cross-ratios and Plücker-type relations—such that $\mathcal{M}$ 9 is $r$ 0-representable if and only if there exists a morphism $r$ 1 in the category of pastures.

Regular matroids are precisely those with $r$ 2 (the sign-hyperfield pasture), i.e., they are representable over all pastures, and thus all fields. This functorial framework also defines near-regular, dyadic, and other regular-type classes by specifying $r$ 3 up to tensor products of atomic pastures, closing these classes under operations like direct sums, 2-sums, parallel connections, and segment-cosegment exchanges (Baker et al., 2024, Baker et al., 2020).

This foundation approach unifies partial-field and hyperfield perspectives and enables a comprehensive classification of matroids without large uniform minors into precisely twelve representation classes, three of which are not representable over any field (Baker et al., 2020).

3. Regular Orthogonal Matroids and Generalized Torsor Structures

Regular orthogonal matroids (even Δ-matroids, Lagrangian orthogonal matroids) further generalize regular matroids by admitting representations via principally-unimodular, skew-symmetric matrices. Central to their structure is the association of a canonical lattice and a Jacobian group $r$ 4 (with $r$ 5 generated by projected signed circuits). In the case of ribbon graphs, the set of spanning quasi-trees is canonically a $r$ 6-torsor, generalizing the classical planar graph correspondence of spanning trees and sandpile groups to higher-genus surfaces and models (Baker et al., 15 Jan 2025).

This generalization unites the classical Jacobian–spanning tree bijection with invariants from algebraic geometry, combinatorics, and topology, highlighting the deep interplay between combinatorial objects and algebraic structures.

4. Generalized Invariants: Jacobians, Tutte Polynomials, and Edge Polytopes

Regular matroid generalization supports natural extensions of classical invariants:

Jacobian groups/sandpile groups: For √[6]{1}-matroids, the Jacobian is defined as $r$ 7 with $r$ 8, and its cardinality equals the square of the number of bases; regular matroids are a special case, and the construction generalizes to regular orthogonal matroids as outlined above (Baker et al., 2023, Baker et al., 15 Jan 2025).
Tutte polynomial and A-polynomial: The A-polynomial $r$ 9 for regular oriented matroids generalizes the classical Tutte polynomial by encoding coflow enumeration, acyclicity, total cyclicity, and supporting deletion-contraction recurrences. For regular matroids, a specialization of $E$ 0 recovers the ordinary Tutte polynomial (Awan et al., 2022).
Symmetric edge polytopes: For every regular matroid, there exists a canonical generalized symmetric edge polytope, unimodularly determined by the matroid, and with facet and triangulation structure controlled by the dual matroid's flows and cuts. This identifies a bridge between Ehrhart theory, flow polynomials, and matroid structure (D'Alì et al., 2023).

5. Algorithmic and Polyhedral Ramifications

The regular matroid generalization enables robust algorithmic implications. For instance, the weighted linear matroid parity problem, which subsumes matching and matroid intersection, admits a deterministic, strongly polynomial augmenting-path algorithm in the linearly representable (and hence regular) case. This unifies matching, matroid intersection, and weighted T-join packing within a single algebraic-combinatorial paradigm (Iwata et al., 2019).

Moreover, the structure of regular matroids supports polyhedral generalizations—e.g., every symmetric edge polytope of a regular matroid is centrally symmetric, reflexive, and supports regular unimodular triangulations directly tied to matroid bases and signed circuits (D'Alì et al., 2023).

6. Structural and Decomposition Theories

Seymour's decomposition theorem for regular matroids extends to broader classes via the foundation framework. Any regular matroid decomposes (polynomially explicitly) into basic building blocks—graphic, cographic matroids, or $E$ 1—via 1-, 2-, and 3-sum operations. The categorical viewpoint further extends compositional closure to $E$ 2-regular classes for any pasture $E$ 3 [(Dinitz et al., 2012); (Baker et al., 2024)].

Additionally, generalizations such as max-flow min-cut matroids inherit these decomposition and algorithmic properties, emphasizing that regular matroid theory is a paradigmatic base case for broader algorithmic and structural results in matroid theory (Dinitz et al., 2012).

7. Connections to Activity Classes, Reversal Systems, and Enumeration

In regular matroids, reversal class enumeration (circuit–cocircuit reversal, minimal activity classes) is tightly controlled by evaluations of the Tutte polynomial. The equivalence of the number of reversal classes and Tutte polynomial values provides a purely combinatorial characterization of regularity, distinguishing regular matroids among their generalizations by the existence and breakdown of certain enumerative coincidences (Gioan et al., 2017).

This direct enumeration-theoretic criterion reinforces the primacy of regular matroids within the larger family while suggesting precise boundaries of generalizability.

References: