Homogeneous Colored Linear Orderings

• Homogeneous colored linear orderings are countable, linearly ordered structures expanded by unary color predicates that guarantee any isomorphism between finite suborders extends to an automorphism.
• These structures are studied from multiple perspectives—including convex equivalence relations, permutation structure reducts, and dense shuffle constructions—linking them to Fraïssé homogeneity and ordered model theory.
• Recent work emphasizes combinatorial enumeration, the role of dense shuffles over the rationals, and partition properties, shedding light on the balance between dense mixing and convex block decomposition.

Homogeneous colored linear orderings are countable linearly ordered structures of the form $(M,<,(P_i)_{i\in I})$ in a relational language, typically with unary predicates as colors, such that every isomorphism between finite induced colored suborders extends to an automorphism. In the current literature, they are studied both as pure colored orders and as reducts or definitional companions of richer ordered structures involving convex equivalence relations, subquotient orders, or successor–predecessor structure. The subject sits at the intersection of Fraïssé homogeneity, ordered model theory, permutation-structure classification, and recent computability-theoretic and enumerative work (Tanović et al., 2018, Braunfeld et al., 2018, Gonzalez, 15 Apr 2026).

1. Basic model-theoretic setting

A colored linear order is naturally treated as a relational structure in a language $L=\{<,P_1,\dots,P_k\}$, where $<$ is a binary relation interpreting a linear order and the $P_i$ are unary predicates. In this framework, homogeneity is the usual Fraïssé-homogeneity: a countable structure $U$ is homogeneous if any isomorphism between finite induced substructures extends to an automorphism. Equivalently, in the formulation used for relational structures, for every finite $A\subseteq U$ and every embedding $f:U\restriction A\to U$, there is an automorphism $g\in \mathrm{Aut}(U)$ with $g\restriction A=f\restriction A$ (Sauer, 2020).

For pure linear orders, the classical situation is rigid: the only countable homogeneous linear order is the order type of the rationals $(\mathbb Q,<)$. Homogeneous colored linear orderings therefore arise only after expanding the order by additional unary or otherwise definable structure. A recurring theme is that such expansions remain homogeneous only when the extra structure is itself distributed in a highly symmetric manner, typically through dense colorings, convex partitions, or generic quotient constructions (Braunfeld et al., 2018).

The literature uses several closely related viewpoints. One treats colors literally as unary predicates. Another treats them as definable equivalence classes or convex layers. A third regards them as coding block types in more structured homogeneous orders. These perspectives are technically distinct, but much of the modern theory shows that they are mutually translatable under strong regularity hypotheses.

2. Convex equivalence relations and definitional equivalence

A major line of work studies when an arbitrary linearly ordered theory is, up to definitional equivalence, “not much more complex than some colored orders.” In this setting, a colored order is a structure $L=\{<,P_1,\dots,P_k\}$0 with only unary predicates beyond the order, while a ccel-structure is a linearly ordered structure expanded by unary predicates and convex equivalence relations. Two structures on the same domain are definitionally equivalent if they have exactly the same definable sets of tuples (Tanović et al., 2018).

Rubin’s binarity condition (RB) characterizes pure colored orders: an $L=\{<,P_1,\dots,P_k\}$1-saturated linearly ordered structure satisfies (RB) iff it is definitionally equivalent to a colored order. The stronger condition introduced as (SLB), strong linear binarity, characterizes the next layer up: a complete theory $L=\{<,P_1,\dots,P_k\}$2 satisfies (SLB) iff it is definitionally equivalent to its ccel-companion, equivalently to a theory of colored orders expanded by convex equivalence relations (Tanović et al., 2018).

This has strong geometric consequences for definable sets. If $L=\{<,P_1,\dots,P_k\}$3 satisfies (SLB), every formula is equivalent modulo $L=\{<,P_1,\dots,P_k\}$4 to a Boolean combination of u-convex formulas, i.e. formulas built from unary predicates together with comparisons to successor or predecessor classes of definable convex equivalence relations. In particular, unary definable sets are controlled by intervals, unary colors, and classes of convex equivalence relations; higher-arity definable sets are correspondingly reduced to low-arity order-and-convexity data (Tanović et al., 2018).

From the perspective of homogeneous colored linear orderings, this yields a structural principle: whenever the local automorphism behavior is strong enough to force (SLB), the resulting theory has no hidden higher-arity geometry beyond colors and convex blocks. A plausible implication is that many homogeneous colored orders of interest are best viewed not as arbitrary unary expansions, but as colored orders layered by convex equivalence relations.

3. Reducts of permutation structures and distributive lattices of colors

A second major viewpoint comes from homogeneous finite-dimensional permutation structures, i.e. countable structures in finitely many linear orders. The classification theorem states that a homogeneous finite-dimensional permutation structure $L=\{<,P_1,\dots,P_k\}$5 is exactly a structure interdefinable with an expansion of the generic $L=\{<,P_1,\dots,P_k\}$6-ultrametric space by generic subquotient orders, for some finite distributive lattice $L=\{<,P_1,\dots,P_k\}$7, such that every meet-irreducible of $L=\{<,P_1,\dots,P_k\}$8 is the bottom relation of some subquotient order (Braunfeld et al., 2018).

Here the lattice $L=\{<,P_1,\dots,P_k\}$9 records the $<$0-definable equivalence relations, and a subquotient order from $<$1 to $<$2 is a partial order on $<$3 whose comparability is confined to each $<$4-class. This is the natural language for multi-level convex partitions. For homogeneous colored linear orderings that arise as reducts of such permutation structures, the “colors” are therefore not arbitrary: they are controlled by a finite distributive lattice of definable equivalence relations, and the ordering between or within color classes is encoded by generic subquotient orders (Braunfeld et al., 2018).

The imprimitive case is especially relevant. In a homogeneous finite-dimensional permutation structure, every meet-irreducible definable equivalence relation is convex with respect to at least one basic linear order, and every definable equivalence relation is an intersection of such convex ones. This implies that definable color classes must sit inside a finite distributive lattice of convex partitions rather than appearing as arbitrary non-convex unary patterns (Braunfeld et al., 2018).

The three-dimensional classification makes these templates concrete. Every homogeneous 3-dimensional permutation structure is interdefinable with the Fraïssé limit of a well-equipped lift of finite $<$5-ultrametric spaces, and examples such as $<$6 can be read as dense linear orders of convex blocks, each block internally a dense order. As reducts to one order plus definable unary data, such examples become canonical homogeneous colored linear orderings built from lexicographic-style block systems (Braunfeld, 2017).

4. Dense colorings of $<$7 and shuffle constructions

A particularly transparent class of homogeneous colored linear orders is built over the rational order. Given a non-empty countable set $<$8 of linear orders, an $<$9-coloring $P_i$0 is dense if every color appears between any two rationals. Skolem’s theorem implies that for every non-empty countable $P_i$1 there exists a dense coloring $P_i$2, and any two such dense colorings are conjugate by an automorphism of $P_i$3 (Srivastava et al., 2024).

From such a coloring one forms the shuffle

$P_i$4

obtained by replacing each rational $P_i$5 with a copy of the linear order $P_i$6. This gives a two-layer structure: the base layer is the densely colored homogeneous order $P_i$7, and the fiber layer expands each color to a prescribed interval type. The construction is canonical up to isomorphism because the dense coloring is unique up to automorphism (Srivastava et al., 2024).

These shuffles behave as homogeneous colored linear orderings in a strong sense. The paper proves that if $P_i$8 and $P_i$9 are countable shuffles that embed convexly into each other, then $U$0. The proof reconstructs each shuffle from a colored tree whose frontier is again a densely colored copy of $U$1, showing that convex biembeddability collapses to isomorphism inside this class (Srivastava et al., 2024).

The same framework yields strong self-similarity identities, including

$U$2

and $U$3, where $U$4 is the one-point order. This suggests that homogeneous colored linear orderings built from dense rational shuffles are governed less by finite concatenation and more by persistent dense mixing of local interval types.

5. Indivisibility and partition properties

Partition properties provide a different measure of structural rigidity. A structure $U$5 is indivisible if for every coloring $U$6, there is a copy $U$7 of $U$8 such that $U$9 is constant on $A\subseteq U$0. For free amalgamation homogeneous structures, Sauer proves that indivisibility is equivalent to rank linearity, i.e. linearity of the poset of ages of typesets, under the hypotheses that $A\subseteq U$1 is countable, oligomorphic, and free amalgamation homogeneous (Sauer, 2020).

Homogeneous colored linear orders do not satisfy the crucial free amalgamation hypothesis. Linear orders cannot freely amalgamate, because any two points from different components must be comparable. Consequently, the full criterion “rank linear iff indivisible” does not transfer to homogeneous colored linear orders, and the paper explicitly states that Theorem 2.1 does not directly apply in that setting (Sauer, 2020).

One positive result does survive. If $A\subseteq U$2 is a countable, oligomorphic, homogeneous relational structure with $A\subseteq U$3, meaning all types have the same rank, then $A\subseteq U$4 is indivisible. Applied to homogeneous colored linear orders, this yields a direct sufficient condition: if the colored order is homogeneous, oligomorphic, and all typesets have the same age, then it is indivisible (Sauer, 2020).

The broader implication is negative as well as positive. The paper supplies a useful framework—types, ranks, bundles, and constructive families of types—but it does not give a complete rank-theoretic classification of indivisible homogeneous colored linear orderings. A plausible implication is that partition phenomena in ordered settings require invariants sensitive to ordered amalgamation, not only to free-amalgamation rank hierarchies.

6. Finite approximations and Ehrenfeucht–Fraïssé equivalence

Finite colored linear orders provide a local approximation theory for the infinite homogeneous case. A finite coloured linear ordering is a structure $A\subseteq U$5, where $A\subseteq U$6 is finite and $A\subseteq U$7 is a surjection onto a finite set of colours. Two such structures are $A\subseteq U$8-equivalent, written $A\subseteq U$9, if Duplicator has a winning strategy in the $f:U\restriction A\to U$0-move Ehrenfeucht–Fraïssé game on them (Mwesigye et al., 2017).

For $f:U\restriction A\to U$1, the classification is explicit. Every finite coloured linear order determines a T-configuration, built from the first and last occurrences at which new colors appear, together with a function $f:U\restriction A\to U$2 recording which colors occur between consecutive T-points. Two finite coloured linear orders are $f:U\restriction A\to U$3-equivalent iff they belong to the same class $f:U\restriction A\to U$4. Equivalently, T-configurations plus the interval-color data $f:U\restriction A\to U$5 classify all $f:U\restriction A\to U$6-classes (Mwesigye et al., 2017).

This yields canonical representatives. Any finite coloured string has a $f:U\restriction A\to U$7-equivalent optimal substring, and no $f:U\restriction A\to U$8-optimal $f:U\restriction A\to U$9-coloured string realizes the same $g\in \mathrm{Aut}(U)$0-character more than once. For $g\in \mathrm{Aut}(U)$1-equivalence, the behavior is subtler: the paper constructs a 2-coloured string of length $g\in \mathrm{Aut}(U)$2 in which all points have distinct $g\in \mathrm{Aut}(U)$3-characters and which is $g\in \mathrm{Aut}(U)$4-optimal, and also a longer $g\in \mathrm{Aut}(U)$5-optimal 2-coloured string of length $g\in \mathrm{Aut}(U)$6 where some $g\in \mathrm{Aut}(U)$7-characters repeat (Mwesigye et al., 2017).

These results are finite rather than homogeneous in the Fraïssé sense, but they are directly relevant to homogeneous colored linear orderings. They describe the local finite profiles that an infinite homogeneous limit must realize uniformly, and they show how quickly the combinatorics of local order-plus-color types become intricate even at quantifier rank $g\in \mathrm{Aut}(U)$8.

7. $g\in \mathrm{Aut}(U)$9-homogeneity, $g\restriction A=f\restriction A$0-homogeneity, and enumeration

Recent work connects homogeneous colored linear orderings to the successor–predecessor expansion of a linear order. An $g\restriction A=f\restriction A$1-linear ordering is a structure $g\restriction A=f\restriction A$2, where $g\restriction A=f\restriction A$3 is the successor of $g\restriction A=f\restriction A$4 if it exists and $g\restriction A=f\restriction A$5 otherwise, and similarly for $g\restriction A=f\restriction A$6. A linear order is $g\restriction A=f\restriction A$7-homogeneous if $g\restriction A=f\restriction A$8 is homogeneous. The classification states that $g\restriction A=f\restriction A$9 is $(\mathbb Q,<)$0-homogeneous iff it is the union of open intervals isomorphic to shuffle sums $(\mathbb Q,<)$1, together with isolated single blocks of types in $(\mathbb Q,<)$2 (Calvert et al., 29 Sep 2025).

The approximation hierarchy $(\mathbb Q,<)$3 replaces $(\mathbb Q,<)$4 and $(\mathbb Q,<)$5 by finitely many unary predicates $(\mathbb Q,<)$6 and finitely many adjacency predicates $(\mathbb Q,<)$7. For finite $(\mathbb Q,<)$8, a linear order is $(\mathbb Q,<)$9-homogeneous iff it is $L=\{<,P_1,\dots,P_k\}$00-homogeneous and every 1-block has size at most $L=\{<,P_1,\dots,P_k\}$01; if $L=\{<,P_1,\dots,P_k\}$02, blocks of type $L=\{<,P_1,\dots,P_k\}$03 are excluded; if $L=\{<,P_1,\dots,P_k\}$04, blocks of type $L=\{<,P_1,\dots,P_k\}$05 are excluded; and $L=\{<,P_1,\dots,P_k\}$06-homogeneity is exactly $L=\{<,P_1,\dots,P_k\}$07-homogeneity (Calvert et al., 29 Sep 2025).

The same papers show that both $L=\{<,P_1,\dots,P_k\}$08-homogeneous orders and homogeneous colored linear orders correspond to finite combinatorial objects. For homogeneous colored linear orders in $L=\{<,P_1,\dots,P_k\}$09 colors, the number $L=\{<,P_1,\dots,P_k\}$10 is finite and has exponential generating function

$L=\{<,P_1,\dots,P_k\}$11

Moreover,

$L=\{<,P_1,\dots,P_k\}$12

where

$L=\{<,P_1,\dots,P_k\}$13

and the first values are

$L=\{<,P_1,\dots,P_k\}$14

The proportion of homogeneous colored linear orders on $L=\{<,P_1,\dots,P_k\}$15 colors that actually use all $L=\{<,P_1,\dots,P_k\}$16 colors tends to $L=\{<,P_1,\dots,P_k\}$17 (Gonzalez, 15 Apr 2026).

For $L=\{<,P_1,\dots,P_k\}$18-homogeneous orders, writing $L=\{<,P_1,\dots,P_k\}$19, the number $L=\{<,P_1,\dots,P_k\}$20 is again finite, depends only on $L=\{<,P_1,\dots,P_k\}$21, and admits both recurrences and a closed form. The paper proves, in particular,

$L=\{<,P_1,\dots,P_k\}$22

with initial values

$L=\{<,P_1,\dots,P_k\}$23

and shows that

$L=\{<,P_1,\dots,P_k\}$24

Thus the $L=\{<,P_1,\dots,P_k\}$25-homogeneous orders form an asymptotically negligible subclass of all homogeneous colored linear orders with the same number of colors (Gonzalez, 15 Apr 2026).

Taken together, these results place homogeneous colored linear orderings in a sharply stratified landscape. At one extreme lie dense color shuffles over $L=\{<,P_1,\dots,P_k\}$26; at another lie reducts of generic ultrametric–subquotient systems; between them sit the $L=\{<,P_1,\dots,P_k\}$27- and $L=\{<,P_1,\dots,P_k\}$28-homogeneous classes, which are rigid enough to be classified by finite combinatorial data and counted explicitly. A plausible synthesis is that homogeneity in ordered colored settings is governed by a small number of recurring mechanisms—dense shuffling, convex block decomposition, and distributive-lattice organization of definable partitions—appearing in different technical guises across the literature.