---
title: Universal Pasture in Matroid Theory
url: https://www.emergentmind.com/topics/universal-pasture
type: topic
---

# Universal Pasture in Matroid Theory

Universal pasture denotes the algebraic object that represents matroid or polymatroid representation data over pastures or tracts. In the matroid setting, the literature distinguishes the universal pasture \(k_M\), which classifies \(F\)-matroid structures on a fixed matroid \(M\), from the foundation \(k_M^f\) or \(F_M\), which classifies rescaling classes and is generated by cross-ratios. In the discrete polymatroid setting, the universal pasture \(P_J\) represents the weak thin Schubert cell of a polymatroid \(J\), while its foundation \(F_J\) represents the realization space. These constructions are formulated through universal mapping properties, explicit generators-and-relations presentations, and, for matroids, effective algorithms based on Tutte groups and cross-ratio data [1809.03542, 2307.14275, 2507.14718].

## 1. Universal property and categorical role

A pasture is a “field-like” algebraic object whose multiplicative units \(P^\times\) form an abelian group and whose addition is encoded by a distinguished set of triples
\[
N_P\subset P^3,\qquad (a,b,c)\in N_P \Longleftrightarrow a+b+c=0.
\]
A morphism of pastures \(f:P_1\to P_2\) is a monoid map sending \(0\mapsto 0\), units to units, \(\epsilon_{P_1}\mapsto \epsilon_{P_2}\), and preserving additive triples. Fields, hyperfields, and partial fields appear as special cases inside the category of pastures [2307.14275, 1809.03542].

For a matroid \(M\), two representation functors are singled out. One writes \(\mathcal X^I_M(P)\) for isomorphism-classes of \(P\)-representations of \(M\), and \(\mathcal X^R_M(P)\) for rescaling-equivalence-classes of \(P\)-representations of \(M\). The representability theorem states that there is a unique pasture \(P_M\), the universal pasture, representing \(P\mapsto \mathcal X^I_M(P)\), and a unique pasture \(F_M\), the foundation, representing \(P\mapsto \mathcal X^R_M(P)\). Equivalently,
\[
\mathcal X^I_M(P)\cong \Mor(P_M,P),
\qquad
\mathcal X^R_M(P)\cong \Mor(F_M,P).
\]
In particular, \(F_M\) is initial among pastures \(Q\) for which \(M\) admits a \(Q\)-representation up to rescaling, and every rescaling-class over any \(P\) arises by push-forward along a unique morphism \(F_M\to P\) [2307.14275].

The moduli-space formulation uses different notation. For a classical matroid \(M\), the residue pasture \(k_M\) is the universal pasture; the weak universal pasture \(k_M^w\) classifies weak \(F\)-matroid structures; and the foundation \(k_M^f\) is the sub-pasture generated by cross-ratios. The inclusions
\[
k_M^f\subset k_M^w\subset k_M
\]
encode the passage from rescaling classes to weak structures and then to strong structures [1809.03542].

## 2. Residue-pasture construction from the moduli space of matroids

Let \(E\) be a finite set with a fixed total order and let \(r\) satisfy \(0<r<\#E\). The moduli space of rank-\(r\) matroids on \(E\) is the ordered blue scheme
\[
\Mat(r,E)=\Proj\!\bigl([\,x_I\mid I\in\binom Er\,]/\cPl(r,E)\bigr),
\]
where \(\cPl(r,E)\) is generated by the Plücker relations
\[
0\le \sum_{k=0}^r \epsilon^k\,x_{I\setminus\{i_k\}}\,x_{J\cup\{i_k\}}
\]
for each \((r-1)\)-subset \(J\subset E\) and \((r+1)\)-subset \(I=\{i_0<\dots<i_r\}\). Its \(F\)-points classify strong \(F\)-matroids of rank \(r\) on \(E\) [1809.03542].

A classical matroid \(M\) corresponds to the unique \(\mathbb K\)-valued point
\[
\chi_M:\Spec\mathbb K\to \Mat(r,E),
\]
with support-determined image
\[
x_M=\bigl\langle x_I\mid I\notin\cB\bigr\rangle,
\]
where \(\cB\subset \binom Er\) is the set of bases of \(M\). The universal pasture \(k_M\) is the residue idyll at \(x_M\):
\[
k_M=\Bigl(\cO_{\Mat(r,E),x_M}\Bigr)_0^\pm
=
\Bigl(\bigl[\,x_I^{\pm1}\mid I\in\cB\,\bigr]/\cPl(r,E)\Bigr)_0^\pm.
\]
Here \((-)_{0}\) denotes the degree-zero part of the naturally graded blueprint, and \((-)^\pm\) is the pasteurization [1809.03542].

The universal mapping property is precise: for any idyll \(F\), there is a canonical bijection
\[
\Hom(k_M,F)\longrightarrow \{\text{strong }F\text{-matroid structures on }M\},
\]
functorial in \(F\). In the matroid representation-theoretic notation of Chen–Zhang, this is the same universal role assigned to \(P_M\) [1809.03542, 2307.14275].

## 3. Weak universal pasture, foundation, and cross-ratios

The weak version keeps only the 3-term Plücker relations. The weak matroid scheme \(\Mat^w(r,E)\) is defined by
\[
0\le x_{I,12}\,x_{I,34}
+\epsilon\,x_{I,13}\,x_{I,24}
+x_{I,14}\,x_{I,23},
\qquad
I\in \binom E{r-2},\; i_1<\cdots<i_4.
\]
The corresponding weak universal pasture is
\[
k_M^w
=
\bigl(B^w_{x_M}\bigr)_0^\pm
=
\Bigl(\bigl[x_I^{\pm1}\mid I\in\cB\bigr]/\cPl^w(r,E)\Bigr)_0^\pm.
\]
It classifies weak \(F\)-matroid structures on \(M\) [1809.03542].

The foundation is the sub-idyll of \(k_M^w\) generated over \(\mathbb F_1^\pm\) by all cross-ratios of \(M\). For a weak \(F\)-matroid \(\Delta:\binom Er\to F\) and a quadrangle
\[
\omega=(I;i_1,i_2,i_3,i_4)\in \binom E{r-2}\times E^4,\qquad I_{k\ell}=I\cup\{i_k,i_\ell\},
\]
the cross-ratio is
\[
\Cr_\Delta(\omega)
=
\epsilon^{\mu(i_1,i_2;\,i_3,i_4)}
\frac{\Delta(I_{13})\,\Delta(I_{24})}{\Delta(I_{14})\,\Delta(I_{23})}.
\]
Each \(\Cr_\Delta(\omega)\) is a fundamental element, meaning that it satisfies \(0\le a+b+\epsilon\). The foundation \(k_M^f\) is generated by these cross-ratios, and in Chen–Zhang’s notation the foundation \(F_M\) represents rescaling classes of matroid representations [1809.03542, 2307.14275].

The three universal objects are related as follows:

| Object | Description | Universal property |
|---|---|---|
| \(k_M\) | residue pasture at \(x_M\) | classifies strong \(F\)-matroid structures |
| \(k_M^w\) | weak universal pasture from 3-term relations | classifies weak \(F\)-matroid structures |
| \(k_M^f\) | sub-pasture generated by cross-ratios | classifies rescaling classes of weak \(F\)-matroids |

At the level of units, one has
\[
(k_M^w)^\times\cong \mathbb T_M,
\qquad
(k_M^f)^\times\cong \mathbb T_M^{(0)},
\]
where \(\mathbb T_M\) is the Tutte group and \(\mathbb T_M^{(0)}\) is the inner Tutte group, the kernel of the degree map to \(\mathbb Z^E\) [1809.03542].

## 4. Algorithmic computation of the foundation and of morphisms

For a matroid \(M\), Chen–Zhang describe an explicit algorithmic route to the foundation \(F_M\). The first step is to compute the outer Tutte group \(\mathbb T_M\). If \(\mathcal B\) is the set of bases and \(B_0\) is a reference basis, one forms the free abelian group
\[
G=\langle \epsilon,\; X_B\;(B\in\mathcal B)\rangle
\]
subject to \(\epsilon^2=1\), \(X_{B_0}=1\), and the cross-ratio relations \(\Cr(I,k_1,k_2,k_3,k_4)=1\) whenever \(I\subset[n]\), \(|I|=r-2\), and all four \(2\)-element enlargements are bases. These relations are assembled into a matrix \(R_1\) over \(\mathbb F_2\), with one \(\mathbb Z\)-row for \(\epsilon^2=1\), and
\[
\mathbb T_M\cong \coker(R_1).
\]
The second step computes the inner Tutte group
\[
\mathbb T_M^{(0)}=\ker\bigl(\deg:\mathbb T_M\to \mathbb Z^n\bigr)
\]
by choosing a spanning forest in the exchange bipartite graph and constructing a second relations-matrix \(R_2\) so that
\[
\mathbb T_M^{(0)}\cong \coker(R_1\mid R_2).
\]
By Baker–Lorscheid’s theory,
\[
F_M^\times\cong \mathbb T_M^{(0)},
\qquad
\epsilon_{F_M}=\text{image of }\epsilon,
\]
so the Smith normal form of \(\coker(R_1\mid R_2)\) yields the decomposition \(F_M^\times\simeq T\oplus \mathbb Z^r\) [2307.14275].

The additive structure is then recovered from “hexagons.” Each 3-term relation in a pasture is encoded by a fundamental pair \((x,y)\) with
\[
x+y+\epsilon=0.
\]
The only new additive relations in the foundation come exactly from modular quadruples of hyperplanes, namely corank-two flats that lie in four hyperplanes. For each corank-2 flat \(X\) and four hyperplanes \(H_1,\dots,H_4\supset X\), one chooses an independent set \(I\subset X\), elements \(a_i\in H_i\setminus X\), forms the relevant cross-ratios, projects them into \(\mathbb T_M^{(0)}\), and records the resulting pair \(\{x,y\}\). Declaring \(x+y+\epsilon=0\) for all such pairs completes the presentation of \(F_M\) by generators and hexagons [2307.14275].

Once \(F_M\) is known, all rescaling classes of \(P\)-representations are obtained by computing \(\Mor(F_M,P)\). Since \(\Mor(F_M,P)\approx\{\text{group homomorphisms }F_M^\times\to P^\times\}\) subject to hexagon-constraints, the procedure decomposes \(F_M^\times\), fixes \(\psi\in \Hom(T_1,P^\times)\) sending \(\epsilon\mapsto \epsilon\), uses a depth-first search to assign images to generators in \(\FE(P)\), and then extends to a full homomorphism. If \(P\) is a field, each morphism produces a unique reduced-row-echelon form with respect to a fixed basis \(B_0\), and the minors \(\Delta(B_0\setminus\{i\}\cup\{j\})\) are read off as entries of a standard \(r\times n\) matrix [2307.14275].

## 5. Internal structure, examples, and representation-theoretic criteria

The unit group of a foundation has the form
\[
F_M^\times\simeq T\oplus \mathbb Z^r.
\]
The torsion part \(T\) often encodes binary- or ternary-type behavior. Hyperfields and partial fields sit inside the category of pastures, and the cited examples include the sign hyperfield \(\mathbb S\), the near-regular partial field
\[
\mathbb U_1={}_{1}^{\pm}\langle x,y\rangle\sslash\{x+y=1\},
\]
the dyadic partial field
\[
\mathbb D={}_{1}^{\pm}\langle x\rangle\sslash\{2x=1\},
\]
and
\[
\mathbb H={}_{1}^{\pm}\langle x\rangle\sslash\{x+x^{-1}=1,\,x^3=-1\}.
\]
Every fundamental pair \((x,y)\) in \(F_M\) arises from one of four “universal” pastures, and each hexagon of \(F_M\) is one of these four types [2307.14275].

Several classical classification statements are reformulated through foundations. A matroid \(M\) is regular iff its foundation is the regular partial field \(\mathbb F_1^{\pm}={}_{1}^{\pm}\). A non-regular matroid is binary iff its foundation is \(\mathbb F_2\). Equivalently, the regular case and the binary non-regular case are separated by
\[
M\text{ regular} \Longleftrightarrow F_M\cong {}_{1}^{\pm},
\qquad
M\text{ binary (but non-regular)} \Longleftrightarrow F_M\cong \mathbb F_2.
\]
This yields the classical characterization
\[
M\text{ regular}\Longleftrightarrow M\text{ binary and orientable},
\]
because orientability is equivalent to the existence of a morphism \(F_M\to \mathbb S\) [1809.03542, 2307.14275].

Two standard examples illustrate the formalism. For the rank-2 uniform matroid on four points,
\[
F_{U_{2,4}}
=
{}_{1}^{\pm}\langle x,y\rangle\sslash\{x+y=1\}.
\]
A pasture morphism \(F_{U_{2,4}}\to P\) is equivalent to choosing a fundamental pair \((a,b)\in \FE(P)\), and therefore
\[
\#\Mor(F_{U_{2,4}},P)=\#\FE(P).
\]
Over a field \(k\), \(\FE(k)=k\setminus\{0,1\}\), so \(U_{2,4}\) has exactly \(|k|-2\) distinct rescaling classes of \(k\)-representations. The same algebraic object is identified in the cited accounts as the near-regular partial field, denoted \(\mathbb U_1\) in one presentation and \(\mathbb U_0\) in another [2307.14275, 2507.14718].

For the Fano matroid \(F_7\),
\[
F_{F_7}^\times\cong \mathbb Z/7\oplus \mathbb Z^3,
\qquad
\{\text{hexagons}\}=\{(x,x^3)\text{ and its cyclic companions}\}.
\]
The resulting pasture is the pasture attached to the finite field \(\mathbb F_7\), so
\[
\Mor(F_{F_7},P)\neq \varnothing
\quad\text{if and only if}\quad
P\text{ has characteristic }7.
\]
Hence \(F_7\) is representable exactly over fields of characteristic \(7\) [2307.14275].

The same framework yields practical criteria. \(M\) is orientable iff \(\Mor(F_M,\mathbb S)\ne \varnothing\). If \(1\in \FE(F_M)\), then no field \(k\) admits a pasture morphism \(F_M\to k\), so \(M\) is non-representable. Once \(\Mor(F_M,\mathbb F_q)\) is computed, one obtains the exact number of inequivalent \(q\)-representations of \(M\) [2307.14275].

## 6. Universal pastures for discrete polymatroids

Let
\[
\Delta_n^r=\{\alpha\in \mathbb Z_{\ge 0}^n:\sum_i\alpha_i=r\},
\]
and let \(J\subseteq \Delta_n^r\) be a finite M-convex set, i.e. a discrete polymatroid. For a commutative tract \(F\), a weak \(F\)-representation of \(J\) is a map
\[
\rho:\Delta_n^r\to F
\]
whose support is exactly \(J\), which is alternating, and which satisfies the 3-term Plücker relations
\[
\rho(\alpha+j+k)\rho(\alpha+i+l)-\rho(\alpha+i+k)\rho(\alpha+j+l)+\rho(\alpha+i+j)\rho(\alpha+k+l)\in N_F
\]
whenever all six relevant points lie in \(J\). The weak thin Schubert cell is the quotient
\[
\Gr^w_J(F)=\{\text{weak }F\text{-representations }\rho:\Delta_n^r\to F\}/F^\times
\]
by overall rescaling [2507.14718].

The universal pasture \(P_J\) is constructed from generators \(\{x_\beta\}_{\beta\in J}\) and exactly the 3-term Plücker null-relations. One first forms the quotient tract
\[
\widehat P_J
=
\bigl(\text{free tract on }\{x_\beta\}\bigr)\big/\{\text{all }3\text{-term Plücker sums}\},
\]
grades it by \(\deg(x_\beta)=1\), and defines the universal pasture as the degree-zero subtract
\[
P_J=\{a\in \widehat P_J:\deg(a)=0\}\subseteq \widehat P_J.
\]
For every tract \(F\), there is a natural bijection
\[
\Gr^w_J(F)\cong \Hom_{\Tracts}(P_J,F),
\]
functorial in \(F\) [2507.14718].

The full universal tract \(T_J\) is obtained by imposing all higher Plücker relations
\[
\sum_{k=0}^s(-1)^k\,x_{i_0\cdots \widehat{i_k}\cdots i_s}\,x_{i_kj_2\cdots j_s}\in N_{T_J}
\qquad (s=2,\dots,r).
\]
A theorem states that the natural quotient \(P_J\to T_J\) is a bijection on underlying sets, indeed on unit groups. Equivalently,
\[
\Gr^w_J(F)\cong \Gr_J(F)
\qquad\text{for every tract }F.
\]
The abstract further states that the canonical bijection between the universal tract and the universal pasture is new even in the case of matroids [2507.14718].

Inside \(P_J\), the foundation \(F_J\subseteq P_J\) is the subgroup of invertible degree-zero elements. It is generated, as a tract over the Krasner hyperfield \(\{0,1\}\), by cross-ratios attached to non-degenerate \(4\)-tuples:
\[
\Cr(\alpha;i,j,k,l)
=
\frac{x_{\alpha+i+k}\,x_{\alpha+j+l}}{x_{\alpha+i+l}\,x_{\alpha+j+k}}.
\]
These cross-ratios satisfy a list of “obvious” multiplicative relations, including symmetries, the relation
\[
\Cr(\alpha;i,j,k,l)+\Cr(\alpha;i,j,l,k)=1,
\]
degenerate relations \(\Cr=1\) when one of the six points is missing, and 3-term and 4-term cycle relations. These relations suffice to present \(F_J\) when \(J\) is a matroid; in general they remain a conjecturally complete generating list [2507.14718].

For \(J=U_{2,4}\subseteq \Delta_4^2\), all cross-ratios coincide with
\[
x=\frac{x_{13}x_{24}}{x_{14}x_{23}},
\]
and the single 3-term Plücker relation becomes \(1+x-1\in N_{P_J}\). Hence the foundation is identified with the near-regular partial field. The realization space is recovered as
\[
\ulineGr^w_J(F)\cong \Hom(F_J,F),
\]
making the bijection between \(F\)-realizations of \(J\) and morphisms from the foundation manifest [2507.14718].

Source: https://www.emergentmind.com/topics/universal-pasture