Entangled Linear Orders in Set Theory

• Entangled linear orders are uncountable ordered sets with a strong finite-dimensional configuration property that forces every large disjoint family to realize all coordinatewise order patterns.
• They form a strict n-entangled hierarchy where, for example, 3-entangled orders become separable and embed into ℝ, illustrating critical separability thresholds and finite-dimensional contrasts.
• The study of these orders integrates combinatorial invariants, descriptive set theory, and forcing techniques, impacting the theory of Suslin lines and the behavior of order-theoretic topology.

Entangled linear orders are uncountable linear orders with a strong finite-dimensional configuration property: every uncountable family of pairwise disjoint finite tuples must contain two tuples whose coordinates realize any prescribed comparison pattern. Introduced by Abraham and Shelah, the notion isolates a combinatorial form of order-theoretic complexity, and Todorčević showed under $\mathsf{CH}$ that fully entangled orders exist. Subsequent work has clarified the internal hierarchy of $n$-entangledness, its interaction with separability and the c.c.c., its failure to be a purely topological invariant on suborders of $\mathbb R$, and its role in the theory of Suslin lines and forcing axioms (Carroy et al., 23 Jul 2025, Martinez-Ranero et al., 30 Nov 2025).

1. Formal definition and combinatorial content

Let $(L,\le)$ be a linear order, let $\kappa$ be an infinite cardinal, and let $n>0$. The order $L$ is $(\kappa,n)$-entangled if whenever $F\subseteq\{e\in L^n:e\text{ is injective}\}$ has $|F|=\kappa$, the tuples in $n$0 are pairwise disjoint, and $n$1, there exist distinct $n$2 such that

$n$3

When $n$4, one simply says that $n$5 is $n$6-entangled; when this holds for every $n$7, $n$8 is entangled. An equivalent notation, used in later formulations, describes a “type” as a function $n$9, with the same meaning coordinatewise (Carroy et al., 23 Jul 2025, Martinez-Ranero et al., 30 Nov 2025).

The definition is sensitive to families of pairwise disjoint tuples rather than arbitrary tuples. This disjointness requirement eliminates trivial repetitions and forces genuinely finite-dimensional comparison phenomena. In the standard presentation, one often works with increasing tuples, but the central issue is the realization of all coordinatewise order patterns inside a large disjoint family.

A useful strengthening follows from the Erdős–Dushnik–Miller theorem: once $\mathbb R$0 is $\mathbb R$1-entangled, then for any sufficiently large $\mathbb R$2 and any pattern $\mathbb R$3, there is in fact an infinite subfamily of $\mathbb R$4 all of whose pairs realize $\mathbb R$5 (Carroy et al., 23 Jul 2025). This reformulation is often more effective in recursive or forcing arguments, because it converts a single witness requirement into an infinite homogeneous-pattern conclusion.

2. Weak entangledness and general structural constraints

A weaker variant restricts attention to separated tuples. For $\mathbb R$6, a sequence of pairwise disjoint increasing $\mathbb R$7-tuples $\mathbb R$8 is separated if there are points

$\mathbb R$9

such that for every $(L,\le)$0 and every $(L,\le)$1,

$(L,\le)$2

The order $(L,\le)$3 is weakly $(L,\le)$4-entangled if the usual conclusion is required only for separated sequences; it is weakly entangled if this holds for every $(L,\le)$5 (Krueger, 2019).

Several basic structural facts organize the subject. Every uncountable linear order is trivially $(L,\le)$6-entangled. If $(L,\le)$7 is $(L,\le)$8-entangled, then $(L,\le)$9 has the c.c.c. If $\kappa$0 is $\kappa$1-entangled, then $\kappa$2 is separable; in particular, no Suslin line can be genuinely $\kappa$3-entangled. Moreover, if $\kappa$4 is dense and separable, then for each $\kappa$5, weak $\kappa$6-entangledness is equivalent to ordinary $\kappa$7-entangledness (Krueger, 2019).

These facts divide the theory sharply by dimension. Dimension $\kappa$8 is compatible with strong non-separability phenomena, while dimension $\kappa$9 already forces separability. A standard corollary, stated explicitly in the literature, is that every $n>0$0-entangled order embeds into $n>0$1, since order-theoretic separability is equivalent to embeddability into $n>0$2 (Carroy et al., 23 Jul 2025, Martinez-Ranero et al., 30 Nov 2025).

3. The finite-dimensional hierarchy

A central question was whether $n>0$3-entangledness automatically propagates upward: must every $n>0$4-entangled order be $n>0$5-entangled? Carroy, Levine, and Notaro showed that, under $n>0$6 — more weakly, under $n>0$7 — the answer is negative. For every integer $n>0$8, there is a subset $n>0$9 which is $L$0-entangled but not $L$1-entangled (Carroy et al., 23 Jul 2025).

The construction proceeds by hereditary diagonalization of Sierpiński type. One begins with an auxiliary countable “bad” family $L$2 of pairwise disjoint $L$3-tuples that avoids the increasing type $L$4, while still guaranteeing that in every open region one can find patterns with a single $L$5 in any coordinate. One then enumerates all continuous maps $L$6 from $L$7 into $L$8 and builds by transfinite recursion of length $L$9 a family of $(\kappa,n)$0-tuples $(\kappa,n)$1 such that

$(\kappa,n)$2

satisfies the criterion of Proposition 2.10 in the paper: continuous maps cannot create a large counterexample to $(\kappa,n)$3-entangledness, but the embedded bad family blocks $(\kappa,n)$4-entangledness (Carroy et al., 23 Jul 2025).

A mild covering-by-meager hypothesis, $(\kappa,n)$5, enters the argument to ensure that diagonalization against fewer than $(\kappa,n)$6 meager bad sets succeeds in an open region; this hypothesis holds under $(\kappa,n)$7 (Carroy et al., 23 Jul 2025). The result establishes a strict finite hierarchy rather than a single undifferentiated notion.

4. Topological non-invariance and definability

Entangledness is not determined by the homeomorphism type of the underlying subspace of $(\kappa,n)$8. Under $(\kappa,n)$9, there exist two homeomorphic suborders $F\subseteq\{e\in L^n:e\text{ is injective}\}$0 such that $F\subseteq\{e\in L^n:e\text{ is injective}\}$1 is entangled, meaning $F\subseteq\{e\in L^n:e\text{ is injective}\}$2-entangled for all $F\subseteq\{e\in L^n:e\text{ is injective}\}$3, while $F\subseteq\{e\in L^n:e\text{ is injective}\}$4 is not even $F\subseteq\{e\in L^n:e\text{ is injective}\}$5-entangled (Carroy et al., 23 Jul 2025).

The proof uses a combinatorial invariant denoted $F\subseteq\{e\in L^n:e\text{ is injective}\}$6. Assuming $F\subseteq\{e\in L^n:e\text{ is injective}\}$7, one constructs a $F\subseteq\{e\in L^n:e\text{ is injective}\}$8-entangled, $F\subseteq\{e\in L^n:e\text{ is injective}\}$9-crowded subset $|F|=\kappa$0 that admits a nontrivial homeomorphism $|F|=\kappa$1. The argument enumerates countably many forbidden relations $|F|=\kappa$2, each closed on fibers and enforcing monochromaticity constraints, and then diagonalizes so as to preserve both entangledness and closure under $|F|=\kappa$3. From such a space one can carve two homeomorphic suborders of $|F|=\kappa$4 with different entangledness behavior (Carroy et al., 23 Jul 2025).

The same paper also places entangledness in descriptive set theory. Under $|F|=\kappa$5, there exists a lightface co-analytic, i.e. $|F|=\kappa$6, subset $|F|=\kappa$7 which is entangled. The construction codes each countable admissible $|F|=\kappa$8 by a unique real $|F|=\kappa$9, and lets $n$00 consist of those codes whose transitive collapse is $n$01 and which are minimal in $n$02 among reals coding that structure. A canonical well-ordering of the admissibles then yields an enumeration $n$03, and diagonalization against continuous maps again invokes the Proposition 2.10 criterion (Carroy et al., 23 Jul 2025).

The descriptive-set-theoretic significance is explicit in the literature: analytic sets are too “perfect-set”-rich to be entangled, so the existence of a $n$04 example is best possible in that sense (Carroy et al., 23 Jul 2025).

5. Suslin lines and non-separable phenomena

A linear order has the c.c.c. if every family of pairwise disjoint nonempty open intervals is countable. It is order-theoretically separable if there is a countable $n$05 meeting every nonempty interval; equivalently, $n$06 embeds into $n$07. A Suslin line is a c.c.c. linear order with no countable dense subset, and classical arguments show that every Suslin line contains an order-isomorphic copy of a lexicographically ordered Suslin tree (Martinez-Ranero et al., 30 Nov 2025).

Because every $n$08-entangled order is c.c.c. but every $n$09-entangled order is separable, Suslin lines are the natural setting for the borderline case $n$10. Krueger showed that there is an $n$11-Knaster forcing $n$12 which adds a Suslin line $n$13 that is weakly $n$14-entangled for every $n$15. The construction first adds a normal Suslin tree $n$16 with rational immediate successors, using a variation of Tennenbaum’s forcing; $n$17 is then defined as the lexicographical order on $n$18. Weak entangledness is verified by thinning to a stationary set of compatible finite approximations and amalgamating them so as to realize a prescribed type (Krueger, 2019).

The same work extends entangledness from linear orders to $n$19-trees and proves that an $n$20-tree is entangled if and only if it is free. It also forces, for any positive $n$21, a Suslin tree which is $n$22-entangled but all of whose derived trees of dimension $n$23 are special (Krueger, 2019).

Carroy, Levine, and Notaro obtained a stronger genuinely order-theoretic result under $n$24: there exists a $n$25-entangled linear order of size $n$26 which is non-separable, hence a Suslin line. Their construction starts from a tree order $n$27 on $n$28, anticipates countable increasing partial automorphisms $n$29 at stage $n$30, and defines the final order $n$31 as a lexicographic break-tie extension of the tree order. The resulting order is c.c.c. and $n$32-entangled, while remaining non-separable (Carroy et al., 23 Jul 2025).

6. OGA, iteration methods, and the current frontier

A later development shows that $n$33-entangled Suslin lines can coexist with strong graph-dichotomy principles. Under the assumption of a supercompact cardinal, there is a forcing extension in which the Open Graph Axiom holds, $n$34, and there exists a $n$35-entangled Suslin line $n$36. Consequently, in that model there is a $n$37-entangled uncountable linear order, but no such order is separable (Martinez-Ranero et al., 30 Nov 2025).

The construction begins with the $n$38-entangled Suslin line $n$39 built by Krueger as a lexicographically ordered Suslin tree and performs a countable support iteration

$n$40

of length $n$41, guided by a Laver function $n$42. Two technical notions are central. The first is weakly bi-entangled, a strengthening of weak $n$43-entangledness tailored to separated pairs whose heights go to $n$44. The second is $n$45-properness, a strengthening of properness designed to preserve both the Suslin-tree property of $n$46 and its weakly bi-entangledness through iteration. For open graphs $n$47 on separable metric spaces that fail countable chromaticity, one uses a Todorčević-style side-condition forcing $n$48 that adds an uncountable clique and can be refined to be $n$49-proper and $n$50-preserving (Martinez-Ranero et al., 30 Nov 2025).

The resulting compatibility theorem for countable support iterations shows that the order-theoretic entanglement property and the Suslin-tree structure can be preserved in tandem. This resolves a problem of Carroy, Levine, and Notaro and answers a question of McKenney recorded in the literature (Martinez-Ranero et al., 30 Nov 2025).

Several open problems remain explicit. Among them are whether the supercompact assumption can be eliminated, whether OGA alone already implies the existence of some $n$51-entangled uncountable order, whether there is an OGA model with a $n$52-entangled Suslin line, and what the exact relationship is between entangledness and other splitting or reflection principles in topology (Martinez-Ranero et al., 30 Nov 2025). In parallel, earlier work notes that $n$53 eliminates genuinely entangled orders of size $n$54 while leaving weak entangledness open to further analysis (Krueger, 2019). Together these results place entangled linear orders at a junction of infinitary combinatorics, descriptive set theory, Suslin-tree technology, and forcing axioms.