Indiscernible Extraction in Theories with Bounded k-splitting

arXiv:2610.22002 · math.LO · Submitted 2026-10-15 · Read on arXiv

Listen

Radio episode about this paper

Transcript

Introduction to the show: ident: Mathematics. The day in mathematics: new theorems, new proofs, and what they open.

Ada: I'm Ada.

Emmy: And I'm Emmy, here with Kolya.

Ada: We're looking at the paper: "Indiscernible Extraction in Theories with Bounded k-splitting".

Emmy: Here's the gist of the paper. INDISCERNIBLE EXTRACTION FOR THEORIES WITH BOUNDED k-SPLITTING Hanson introduced the class of theories with bounded k-splitting and proved an indiscernible extraction result for those theories assuming the existence of a k-ineffable cardinal In this article, we show that the conclusion of Hanson’s theorem is equivalent to the weaker assertion that there exists a weakly compact cardinal, as well as prove additional indiscernible extraction results for these theories in ZFC Finally, we show that Hanson’s construction of total bounded ultraimaginary Morley sequences in theories with bounded k-splitting follows from ZFC.

Ada: The paper introduces a higher-arity notion of stability called bounded k-splitting, which is a natural generalization of both stability and k-arity A theory T is defined as k-ary if every formula is equivalent modulo T to a Boolean combination of formulas with at most k variables A type tp(a/B) in this context splits over C ⊆ B if there is a formula φ(x, y) and b, b′ ∈ A with b ≡C b′ such that both φ(a, b) and ¬φ(a, b′) are true A theory T has λ-bounded k-splitting if every such type with finite blocks has some C ⊆ A, C < λ, over which it does not k-split It is noted that stability is equivalent to having bounded 1-splitting, and k-arity implies bounded k-splitting.

Emmy: ZFC Results on Indiscernibles The authors establish two indiscernible extraction results in ZFC for theories with bounded k-splitting Theorem 2.1 states that if T has bounded k-splitting and λ ≥ T + B + γ, then every sequence of γ-tuples of length בk(λ) + has a B-indiscernible subsequence of order type λ+ This result is derived by applying Lemma 2.4, which extracts r-end-homogeneous sequences in theories with bounded k-splitting.

Ada: Theorem 2.6 improves this by showing that every sequence of n-tuples of length בk−1(λ) + has a B-indiscernible subsequence of length λ+ + (k − 1) Extraction from Weak Compactness The paper demonstrates that the large cardinal assumption can be weakened to the existence of a weakly compact cardinal Theorem 3.1 shows that if T has bounded k-splitting and κ > T + B + γ is weakly compact, every sequence of γ-tuples of length κ has a B-indiscernible subsequence of order type κ The optimality of this result is shown by considering the theory of the random graph with edge relation R.

Title and authors: Ada: The paper "Indiscernible Extraction in Theories with Bounded k-splitting" focuses on introducing this new class of theories, bounded k-splitting, which is a higher-arity generalization of stability.

Emmy: That generalization means every formula is equivalent modulo the theory to a Boolean combination of formulas with at most k variables, and it's a natural extension of what we already know about NIP theories or k-arity theories.

Kolya: So, what exactly is bounded k-splitting in terms of structure? Is it just about how complex the types are that we're looking at?

Ada: It's defined by requiring that every type with a finite number of blocks splits over some small subset of its parameter domain, specifically one whose size is less than lambda, where lambda is the bound on k-splitting.

Emmy: That definition links the splitting property directly to an ordinal lambda, which we then use to define bounded k-splitting as having a specific restriction on how types can split.

Kolya: So, if we look at the original Hanson theorem, it required a k-ineffable cardinal for the extraction result, and this paper says that's not necessary.

Ada: Exactly, because they show that the conclusion of Hanson’s theorem is equivalent to the weaker assertion that there exists a weakly compact cardinal instead of needing a k-ineffable cardinal.

The paper's summary: Ada: Wrapping up this discussion on "Indiscernible Extraction in Theories with Bounded k-splitting," the main implication is that we don't need those huge cardinals to get these types of extraction results; weak compactness is sufficient for the key result, and they also prove Hanson’s construction of total bounded ultraimaginary Morley sequences actually works just in ZFC.

Emmy: That final point is really interesting because it means that even the most complex constructions, like those involving ultraimaginary sequences, are achievable within standard set theory without relying on those high cardinal assumptions. The machinery they built is actually quite flexible for ZFC applications.

Kolya: If we take what you said about the total bounded ultraimaginary Morley sequences following from ZFC, how does that impact our ability to study these theories further? Can we use those tools more freely in other areas?

Ada: It means we have a robust set of tools available in ZFC for analyzing these theories. The results show that the complexity of bounded k-splitting isn't as high as some of the initial assumptions suggested when we first looked at it.

Emmy: It sets a clearer path for future research, showing that the connection between bounded k-splitting and extraction is more grounded in ZFC than we first thought about it.

The paper's improvements: Ada: So we looked at "Indiscernible Extraction in Theories with Bounded k-splitting," and the main thing is that you can get these strong extraction results in ZFC without needing any of those big large cardinal assumptions we usually have to assume, relying instead on weak compactness.

Emmy: Exactly, and they show that assuming weak compactness is enough to get the key result, which is a significant simplification for applying this kind of theory to broader mathematical structures.

Kolya: From a probabilistic view, that means the tail we're dealing with for these sequences isn't as heavy as we might think when you look at the constraints imposed by bounded k-splitting.

Ada: Right, and they even managed to prove that Hanson’s construction for total bounded ultraimaginary Morley sequences works just in ZFC. That’s a big win because it shows that even those very complex constructions aren't reliant on those high cardinal assumptions we were used to seeing.

Emmy: It means the structural logic behind these theories is more pliable than we thought, and the machinery they built is actually quite flexible for standard set theory applications.

Kolya: I just wonder if that flexibility translates directly into practical bounds for things like approximation or complexity in other areas of mathematics. What does this mean for our usual computational limits?

Ada: It suggests that the theoretical limits we usually impose are often stricter than what's actually needed to achieve these types of order control over sequences.

Emmy: So, we've established a solid foundation here for extracting patterns in theories with bounded k-splitting using only weak compactness and ZFC.

Ada: That’s the gist of "Indiscernible Extraction in Theories with Bounded k-splitting," showing that weaker assumptions can still yield powerful structural results.

Kolya: It certainly opens the door to exploring how these tools interact with things like those random graph theories we saw earlier. Maybe we can see if this extraction power works there too.

Emmy: That seems like a very natural next step, looking at where these ideas connect across different mathematical universes.

Conclusion: Ada: So we’ve been looking at "Indiscernible Extraction in Theories with Bounded k-splitting," and what we've seen is that you can get these strong extraction results in ZFC without needing any of those big large cardinal assumptions we usually have to assume.

Emmy: Exactly, and they show that assuming weak compactness is enough to get the key result, which is a significant simplification for applying this kind of theory.

Kolya: From a probabilistic standpoint, that means the tail we're dealing with for these sequences isn't as heavy as we might think when you look at the constraints imposed by bounded k-splitting.

Emmy: It means the structural logic behind these theories is more pliable than we thought, and the machinery they built is actually quite flexible for ZFC applications.

Ada: It suggests that the theoretical limits we usually impose are often stricter than what’s actually needed to achieve these types of order control over sequences.

Emmy: That seems like a very natural next step, looking at where these ideas connect across different mathematical universes.

math.LO

Submitted: 2026-10-15

Updated: 2026-11-11

License: http://creativecommons.org/licenses/by/4.0/

Importance score: 92/100

The gist: INDISCERNIBLE EXTRACTION FOR THEORIES WITH BOUNDED k-SPLITTING Hanson introduced the class of theories with bounded k-splitting and proved an indiscernible extraction result for those theories…

Key concepts

Bounded k-splitting
This is a higher-arity notion of stability where a theory is defined by how formulas behave with respect to variable count. A theory has λ-bounded k-splitting if any type with finite blocks splits over some subset C of the parameters, where the size of C is strictly less than λ. Stability corresponds to bounded 1-splitting.
Indiscernible Extraction
This refers to a process in model theory that extracts long sequences from a given sequence of tuples that possess specific structural properties, such as being B-indiscernible. The paper proves various extraction theorems for theories with bounded k-splitting, showing how many such subsequences can be found under different set-theoretic assumptions.
Weakly Compact Cardinal
This is a large cardinal assumption that is weaker than stronger ones like measurable cardinals. The authors demonstrate that the conclusion of Hanson's theorem regarding indiscernible extraction for bounded k-splitting theories holds if we assume the existence of such a cardinal, showing it is sufficient.
Total bu-Morley Sequences
These are sequences constructed within a theory that possess specific structural properties related to ultraimaginary elements. The paper proves that Hanson's construction for these sequences can be achieved entirely within ZFC, without needing the existence of a k-ineffable cardinal.

Terminology

Summary

INDISCERNIBLE EXTRACTION FOR THEORIES WITH BOUNDED k-SPLITTING Hanson introduced the class of theories with bounded k-splitting and proved an indiscernible extraction result for those theories assuming the existence of a k-ineffable cardinal In this article, we show that the conclusion of Hanson’s theorem is equivalent to the weaker assertion that there exists a weakly compact cardinal, as well as prove additional indiscernible extraction results for these theories in ZFC Finally, we show that Hanson’s construction of total bounded ultraimaginary Morley sequences in theories with bounded k-splitting follows from ZFC. p. 0

Where the paper says it

Hanson introduced the class of theories with bounded k-splitting, a higher-arity notion of stability, and proved an indiscernible extraction result for those theories assuming the existence of a k-ineffable cardinal. In this article, we show that the conclusion of Hanson’s theorem is equivalent to the weaker assertion that there exists a weakly compact cardinal, as well as prove additional indiscernible extraction results for these theories in ZFC.

Introduction to Bounded k-Splitting

The paper introduces a higher-arity notion of stability called bounded k-splitting, which is a natural generalization of both stability and k-arity A theory T is defined as k-ary if every formula is equivalent modulo T to a Boolean combination of formulas with at most k variables A type tp(a/B) in this context splits over C ⊆ B if there is a formula φ(x, y) and b, b′ ∈ A with b ≡C b′ such that both φ(a, b) and ¬φ(a, b′) are true A theory T has λ-bounded k-splitting if every such type with finite blocks has some C ⊆ A, C < λ, over which it does not k-split It is noted that stability is equivalent to having bounded 1-splitting, and k-arity implies bounded k-splitting.

ZFC Results on Indiscernibles

The authors establish two indiscernible extraction results in ZFC for theories with bounded k-splitting Theorem 2.1 states that if T has bounded k-splitting and λ ≥ T + B + γ, then every sequence of γ-tuples of length בk(λ) + has a B-indiscernible subsequence of order type λ+ This result is derived by applying Lemma 2. p. 2

Where the paper says it

Our first goal is the following theorem: Theorem 2. If T has bounded k-splitting and λ ≥ |T| + |B| + |γ|, then every sequence of γ-tuples of length ℶk(λ) + has a B-indiscernible subsequence of order type λ +.

4, which extracts r-end-homogeneous sequences in theories with bounded k-splitting. p. 2
Where the paper says it

□ As mentioned in the preliminary section, the notion of r-end-homogeneous sequences is central to indiscernible extraction in both stable theories and theories with bounded k-splitting. The following lemma extracts r-end-homogeneous sequences in theories with bounded k-splitting.

Theorem 2.6 improves this by showing that every sequence of n-tuples of length בk−1(λ) + has a B-indiscernible subsequence of length λ+ + (k − 1)

Extraction from Weak Compactness

The paper demonstrates that the large cardinal assumption can be weakened to the existence of a weakly compact cardinal Theorem 3.1 shows that if T has bounded k-splitting and κ > T + B + γ is weakly compact, every sequence of γ-tuples of length κ has a B-indiscernible subsequence of order type κ The optimality of this result is shown by considering the theory of the random graph with edge relation R. p. 6

Where the paper says it

□ It is clear from the definition of a weakly compact cardinal that this result is optimal. Let Trg be the theory of the random graph with edge relation R.

Total bu-Morley Sequences

The weaker indiscernible extraction result in ZFC suffices to prove the existence of total bu-Morley sequences Theorem 4.4 proves that for every parameter set B and every possibly infinite tuple a, there is a total bu-Morley sequence (ai)i<ω over B with a0 = a This is achieved by combining the extraction result with Lemma 4. p. 7

Where the paper says it

Suppose T has bounded k-splitting for some 1 ≤ k < ω. For every parameter set B and every possibly infinite tuple a, there is a total bu-Morley sequence (ai)i<ω over B with a0 = a.

3, which provides tuples (dξ)ξξ. p. 6
Where the paper says it

The next lemma is a version of Hanson’s tree construction used in [2]. For every infinite cardinal µ, parameter set B, and set-sized tuple a, there are tuples (dξ)ξ<µ, with dξ ≡B a, and, for each ξ < µ, a bu-Morley sequence (c ξ i )i<ω over B such that d<ξ ⊆ rng(c ξ 0 ), (c ξ i )i<ω is indiscernible over Bd>ξ.

The final construction ensures that the resulting sequence is total bu-Morley over B. p. 6
Where the paper says it

A B-indiscernible sequence (ai)i<ω is total bu-Morley over B if and only if there are infinite sequences I, J such that I + J ≡EM B (ai)i<ω and I ⌣| bu B J. The next lemma is a version of Hanson’s tree construction used in [2].

How the training signal is built

The proof of Theorem 2. p. 3

Where the paper says it

1 now follows. Proof of Theorem 2.

1 relies on Lemma 2. p. 2
Where the paper says it

If T has bounded k-splitting, then every k-partitioned type whose blocks have length γ does not k-split over some subset of its parameter domain of size at most |T| + |γ|. The following lemma is an analogue of a classical result of Shelah [3] bounding the number of complete types extending a given type in a stable theory, to the context of bounded k-splitting.

4, which constructs an end-homogeneous subsequence by recursively partitioning the index set For r > 1, Lemma 2. p. 5
Where the paper says it

The downward closure of a cofinal branch gives classes Xα ∈ Pα for all α < κ. Their minima strictly increase and index an end-homogeneous subsequence of length κ, by the same type equalities as in Lemma 2.

5 is used to extract a final segment that becomes r-end-homogeneous over B This process utilizes the fact that cf(θ) > λ in Theorem 3. p. 3
Where the paper says it

Suppose T has bounded k-splitting and λ ≥ |T| + |B| + |γ| is infinite. If cf(θ) > λ, every k-end-homogeneous sequence of γ-tuples of length θ over B has a B-indiscernible final segment.

1 to ensure the existence of an indiscernible final segment. p. 3
Where the paper says it

Suppose T has bounded k-splitting and λ ≥ |T| + |B| + |γ| is infinite. If cf(θ) > λ, every k-end-homogeneous sequence of γ-tuples of length θ over B has a B-indiscernible final segment.

Where the gains come from

The main gains stem from proving that a weakly compact cardinal is sufficient for the extraction results, and showing that this large cardinal assumption can be weakened to a weakly compact cardinal The paper also shows that Hanson’s construction of total bounded ultraimaginary Morley sequences follows from ZFC This demonstrates that the required large cardinal assumptions are not necessary for these results in ZFC.

Pruning redundant reasoning tokens

The paper explicitly states that the large cardinal assumptions here are not needed, either The equivalence between a weakly compact cardinal and the partition relation κ → (κ)2 holds for the random graph theory The proof of Theorem 4.4 uses a sequence of steps involving Lemma 4. p. 3

Where the paper says it

Then the sequence (eξ)ξ≤ρ (setting eρ = d) is r-end-homogeneous over B and has order type λ + + r. □ Lemma 2.

3, Fact 4.1, and Monotonicity to establish totality The construction in Lemma 4. p. 5
Where the paper says it

We first prove, by induction on r ≥ 1, that every sequence of γ-tuples of length κ has an r-end-homogeneous subsequence of order type κ. For r = 1, carry out the partition construction in Lemma 2.

p. 5
Where the paper says it

We first prove, by induction on r ≥ 1, that every sequence of γ-tuples of length κ has an r-end-homogeneous subsequence of order type κ. For r = 1, carry out the partition construction in Lemma 2.

3 involves a tree Tµ where fξ is identically zero on [ξ + 2, µ) The proof of Theorem 2. p. 7
Where the paper says it

The tree construction in [1] gives an s-indiscernible, bu-spread-out family (tf )f∈Tµ over B, with tf ≡B a for every f. For each ξ < µ, let fξ be identically zero on [ξ + 2, µ), and put dξ = tfξ .

1 applies Lemma 2. p. 3
Where the paper says it

This is a k-end-homogeneous sequence of length λ +. 5 applies since cf(λ +) = λ + > λ.

p. 3
Where the paper says it

This is a k-end-homogeneous sequence of length λ +. 5 applies since cf(λ +) = λ + > λ.

4 with r = k and uses the fact that cf(λ+) = λ+ > λ This ensures that the final segment has order type λ+. p. 3
Where the paper says it

5 applies since cf(λ +) = λ + > λ. A final segment of λ + has order type λ +.

p. 3
Where the paper says it

5 applies since cf(λ +) = λ + > λ. A final segment of λ + has order type λ +.

The construction in Lemma 2. p. 5
Where the paper says it

We first prove, by induction on r ≥ 1, that every sequence of γ-tuples of length κ has an r-end-homogeneous subsequence of order type κ. For r = 1, carry out the partition construction in Lemma 2.

p. 5
Where the paper says it

We first prove, by induction on r ≥ 1, that every sequence of γ-tuples of length κ has an r-end-homogeneous subsequence of order type κ. For r = 1, carry out the partition construction in Lemma 2.

4 for r=1 involves partitioning κ by i ∼ j if and only if ai ≡B aj The proof of Theorem 3. p. 3
Where the paper says it

1 now follows. Proof of Theorem 2.

p. 3
Where the paper says it

1 now follows. Proof of Theorem 2.

1 uses the strong inaccessibility of κ to show that every Pα is nonempty at a limit. p. 5
Where the paper says it

The strict inequality uses the strong inaccessibility of κ. Indeed, the right-hand side bounds the number of types over the parameters on a branch of length α, and at a limit it also bounds the number of preceding chains.

p. 5
Where the paper says it

The strict inequality uses the strong inaccessibility of κ. Indeed, the right-hand side bounds the number of types over the parameters on a branch of length α, and at a limit it also bounds the number of preceding chains.

The construction of an ultraimaginary sequence in Lemma 4.3 involves enumerations matched by the map that changes only the value at ξ + 1 The domains of fξ and gξi start at successor ordinals, so both are nodes of Tµ The s-indiscernibility then (c ξ i)iξ This also holds for the whole subtree tuples, since each formula uses only finitely many coordinates. p. 7

Where the paper says it

Use enumerations matched by the map that changes only the value at ξ + 1. The domains of fξ and g ξ i start at successor ordinals, so both are nodes of Tµ.

p. 7
Where the paper says it

Use enumerations matched by the map that changes only the value at ξ + 1. The domains of fξ and g ξ i start at successor ordinals, so both are nodes of Tµ.

The key improvement in Theorem 2.6 is removing a beth from the length of the original sequence and extracting an indiscernible sequence with an extra finite tail The equivalence between a weakly compact cardinal and the partition relation κ → (κ)2 holds for the random graph theory This establishes that weak compactness is optimal for Theorem 3. p. 4

Where the paper says it

This is an improvement of Theorem 2. 1 in two ways, first by removing a beth from the length of the original sequence, and second by extracting an indiscernible sequence with an extra finite tail.

1. The construction in Lemma 4.3 uses an s-indiscernible, bu-spread-out family (tf)f∈Tµ over B This family is used to build the total bu-Morley sequence. p. 7
Where the paper says it

Let Tµ be the tree of finitely supported functions f : [β, µ) → ω, where β < µ is zero or a successor ordinal, ordered by extension. The tree construction in [1] gives an s-indiscernible, bu-spread-out family (tf )f∈Tµ over B, with tf ≡B a for every f.

p. 7
Where the paper says it

Let Tµ be the tree of finitely supported functions f : [β, µ) → ω, where β < µ is zero or a successor ordinal, ordered by extension. The tree construction in [1] gives an s-indiscernible, bu-spread-out family (tf )f∈Tµ over B, with tf ≡B a for every f.

The paper shows that the conclusion of Hanson’s theorem is equivalent to the weaker assertion that there exists a weakly compact cardinal The equivalence between a weakly compact cardinal and the partition relation κ → (κ)2 holds for the random graph theory The proof of Theorem 3.1 uses the strong inaccessibility of κ to show that every Pα is nonempty at a limit The construction in Lemma 4. p. 5

Where the paper says it

The strict inequality uses the strong inaccessibility of κ. Indeed, the right-hand side bounds the number of types over the parameters on a branch of length α, and at a limit it also bounds the number of preceding chains.

3 involves an s-indiscernible, bu-spread-out family (tf)f∈Tµ over B This family is used to build the total bu-Morley sequence. p. 7
Where the paper says it

Let Tµ be the tree of finitely supported functions f : [β, µ) → ω, where β < µ is zero or a successor ordinal, ordered by extension. The tree construction in [1] gives an s-indiscernible, bu-spread-out family (tf )f∈Tµ over B, with tf ≡B a for every f.

The proof of Theorem 2.1 applies Lemma 2.4 with r = k and uses the fact that cf(λ+) = λ+ > λ This ensures that the final segment has order type λ+. The construction in Lemma 2.4 for r=1 involves partitioning κ by i ∼ j if and only if ai ≡B aj The proof of Theorem 3.1 uses the strong inaccessibility of κ to show that every Pα is nonempty at a limit.

The construction in Lemma 4.3 involves enumerations matched by the map that changes only the value at ξ + 1 The domains of fξ and gξi start at successor ordinals, so both are nodes of Tµ The s-indiscernibility then (c ξ i)iξ This also holds for the whole subtree tuples, since each formula uses only finitely many coordinates.

The equivalence between a weakly compact cardinal and the partition relation κ → (κ)2 holds for the random graph theory This establishes that weak compactness is optimal for Theorem 3.1 The construction in Lemma 4. p. 5

Where the paper says it

We first prove, by induction on r ≥ 1, that every sequence of γ-tuples of length κ has an r-end-homogeneous subsequence of order type κ. For r = 1, carry out the partition construction in Lemma 2.

3 uses an s-indiscernible, bu-spread-out family (tf)f∈Tµ over B This family is used to build the total bu-Morley sequence.

Green — the paper says it, and the page is printed. Red — no passage carries it, so treat that sentence as unverified. The same check runs on your own text: check a summary against its source, $5.

Improvements for AI systems

  1. Enhanced indiscernible extraction capabilities for theories with bounded k-splitting in ZFC by proving that the conclusion of Hanson’s theorem is equivalent to the weaker assertion that there exists a weakly compact cardinal. This allows for extracting B-indiscernible subsequences even without assuming the existence of a k-ineffable cardinal.

  2. Improved extraction results for sequences by showing that every sequence of gamma-tuples of length beth k plus has a B-indiscernible subsequence of order type beth plus r in Lemma 2.4, which provides more control over the resulting sequence lengths and orders compared to previous results.

  3. Increased power in total bu-Morley sequence extraction by demonstrating that Hanson’s construction of total bounded ultraimaginary Morley sequences follows from ZFC, allowing the system to generate sequences where I +J is B indiscernible based on the weaker ZFC results.

  4. Stronger control over sequence structure by showing that every sequence of n-tuples of length beth k minus one plus has a B-indiscernible subsequence of length beth plus k minus one in Theorem 2.6, which improves extraction by removing a beth from the original sequence length and adding an extra finite tail.

  5. Verification of optimality for large cardinal assumptions by showing that the results are optimal when assuming the existence of a weakly compact cardinal, as demonstrated by its equivalence to the partition relation κ -> (κ)2/2 in Trg.

Abstract

Hanson introduced the class of theories with bounded k-splitting, a higher-arity notion of stability, and proved an indiscernible extraction result for those theories assuming the existence of a k-ineffable cardinal. In this article, we show that the conclusion of Hanson's theorem is equivalent to the weaker assertion that there exists a weakly compact cardinal, as well as prove additional indiscernible extraction results for these theories in ZFC. Finally, we show that Hanson's construction of total bounded ultraimaginary Morley sequences in theories with bounded k-splitting follows from ZFC.

Sources

Related papers