Kim's Lemma in Model-Theoretic Independence

• Kim's Lemma is a family of transfer principles in model-theoretic independence theory that ensures inconsistency of formulas propagates along every sufficiently generic Morley sequence.
• It is pivotal in distinguishing simple, NTP₂, and NSOP₁ frameworks by linking the notion of Kim-dividing with key stability-theoretic and combinatorial properties.
• Recent extensions involving hyperimaginaries and bi-invariant types refine its application, clarifying precise conditions for consistency and inconsistency across models.

Kim’s Lemma is a family of transfer principles in model-theoretic independence theory. Its core assertion is that once a formula divides, or Kim-divides, along one sufficiently generic Morley sequence, the same inconsistency must persist along every Morley sequence of the appropriate kind. In simple theories this principle is equivalent to simplicity itself; in the modern stability-theoretic landscape it has distinct $NTP_{2}$ and $NSOP_{1}$ variants, a hyperimaginary extension, and more recent formulations involving bi-invariant, strongly bi-invariant, extendibly invariant, and reliably invariant global types. These later forms connect Kim’s Lemma to the comb tree property, the antichain tree property, local-character phenomena, and higher-$k$ combinatorial consistency patterns (Kruckman et al., 2023, Bossut, 2022, Hanson, 2023, Hanson, 28 Jul 2025).

1. Classical form in simple theories

In the classical setting, let $T$ be complete, $A$ a small parameter set, and $\varphi(x;b)$ a formula. The formula divides over $A$ if there is an $A$-indiscernible sequence $(b_i)_{i<\omega}$ with $b_0=b$ such that $NSOP_{1}$0 is inconsistent. It forks over $NSOP_{1}$1 if it implies a finite disjunction of formulas each of which divides over $NSOP_{1}$2. A global type $NSOP_{1}$3 is $NSOP_{1}$4-invariant if membership of a formula $NSOP_{1}$5 in $NSOP_{1}$6 is preserved under automorphisms fixing $NSOP_{1}$7, and a Morley sequence of a global $NSOP_{1}$8-invariant type $NSOP_{1}$9 is a sequence $k$0 with $k$1 for all $k$2; by invariance, such a sequence is automatically $k$3-indiscernible (Kruckman et al., 2023).

Kim’s Lemma for simple theories is presented as an equivalence due to Kim and Pillay: $k$4 is simple if and only if, for every set $k$5 and formula $k$6, whenever $k$7 divides over $k$8, then for every Morley sequence $k$9 in a non-forking extension of $T$0, the set $T$1 is inconsistent. The central content is the replacement of “there exists an indiscernible witness” by “every sufficiently generic Morley sequence witnesses the same inconsistency” (Kruckman et al., 2023).

This formulation makes Kim’s Lemma a structural criterion rather than merely a technical tool. The lemma governs how dividing propagates across invariant extensions, and its later generalizations preserve precisely this transfer-of-inconsistency pattern while altering what counts as the relevant generic sequence.

2. Variants in $T$2 and $T$3

In $T$4 theories, the statement persists over models, but the relevant invariant types must be stricter. A global type $T$5 is strictly $T$6-invariant if it is $T$7-invariant and, for every set $T$8 and every realization $T$9, one has $A$0. Chernikov–Kaplan’s version states that $A$1 is $A$2 if and only if, for every model $A$3 and formula $A$4, whenever $A$5 divides over $A$6, then for every strictly $A$7-invariant global extension $A$8 and every Morley sequence of $A$9 over $\varphi(x;b)$0, the set $\varphi(x;b)$1 is inconsistent. The quantifier shift from “some” to “every” strictly invariant Morley sequence is meaningful because every type over a model admits a strictly invariant extension in an $\varphi(x;b)$2 theory (Kruckman et al., 2023).

In $\varphi(x;b)$3 theories, the notion of Kim-dividing replaces ordinary dividing. A formula $\varphi(x;b)$4 Kim-divides over a model $\varphi(x;b)$5 if it divides along some Morley sequence for some global $\varphi(x;b)$6-invariant extension of $\varphi(x;b)$7. Kaplan–Ramsey’s Kim’s Lemma for $\varphi(x;b)$8 says that $\varphi(x;b)$9 is $A$0 if and only if, for every model $A$1 and every formula $A$2, if $A$3 Kim-divides over $A$4, then it divides along every Morley sequence for every global $A$5-invariant extension of $A$6 (Kruckman et al., 2023).

These two variants are formally parallel but conceptually orthogonal. The $A$7 version retains ordinary dividing and strengthens the genericity of the sequence; the $A$8 version changes the independence notion itself from dividing to Kim-dividing while allowing arbitrary invariant Morley sequences. This contrast is the starting point for later attempts to unify the lemma.

3. Hyperimaginaries and model reduction

A hyperimaginary is a class $A$9 of tuples modulo a type-definable equivalence relation $A$0, and $A$1 denotes the home sort of hyperimaginaries. Bossut studies Kim’s Lemma in a complete $A$2-theory with the additional assumption of existence for hyperimaginaries, namely that for any $A$3, one has $A$4 in the sense of forking. Over a hyperimaginary base $A$5, a partial type $A$6 Kim-divides if there is an $A$7-Morley sequence $A$8 in $A$9 such that $(b_i)_{i<\omega}$0 is inconsistent; Kim-forking is defined by finite disjunction as usual (Bossut, 2022).

The hyperimaginary version of Kim’s Lemma states that, assuming $(b_i)_{i<\omega}$1 is $(b_i)_{i<\omega}$2 with existence for hyperimaginaries, for any formula $(b_i)_{i<\omega}$3 and any hyperimaginary parameter $(b_i)_{i<\omega}$4 over $(b_i)_{i<\omega}$5, the following are equivalent: first, $(b_i)_{i<\omega}$6 Kim-divides over $(b_i)_{i<\omega}$7; second, there is a finite $(b_i)_{i<\omega}$8 such that for every $(b_i)_{i<\omega}$9-Morley sequence $b_0=b$0 in $b_0=b$1, the set $b_0=b$2 is $b_0=b$3-inconsistent. In particular, once $b_0=b$4 Kim-divides along one $b_0=b$5-Morley sequence, it Kim-divides along every such sequence (Bossut, 2022).

The proof strategy is explicitly parallel to the model-based $b_0=b$6 argument. It reduces to a model base by embedding a long $b_0=b$7-Morley sequence into a model $b_0=b$8 with $b_0=b$9, and then derives a contradiction from an array built out of two competing Morley sequences. The argument uses witnessing, symmetry, transitivity, and the independence theorem for Kim-independence over models. This indicates that the obstruction is not the passage from real tuples to hyperimaginaries as such, but the need for the existence hypothesis to perform lifting and extension (Bossut, 2022).

4. Bi-invariant and reliably invariant formulations

Recent work broadens the ambient class of invariant types. Let $NSOP_{1}$00 be a parameter set. A global type is $NSOP_{1}$01-bi-invariant if it is $NSOP_{1}$02-invariant and whenever $NSOP_{1}$03, the type $NSOP_{1}$04 extends to some global $NSOP_{1}$05-invariant type. It is strongly $NSOP_{1}$06-bi-invariant if $NSOP_{1}$07 is $NSOP_{1}$08-bi-invariant for every $NSOP_{1}$09. It is extendibly $NSOP_{1}$10-invariant if whenever $NSOP_{1}$11 extends $NSOP_{1}$12, the union $NSOP_{1}$13 extends to some global $NSOP_{1}$14-invariant type. It is reliably $NSOP_{1}$15-invariant if it belongs to the largest subclass of $NSOP_{1}$16-invariant types closed under restriction to fewer variables, extension along any $NSOP_{1}$17 over the same base, and amalgamation along any finite invariant sequence over $NSOP_{1}$18. An invariance base is a set $NSOP_{1}$19 such that every type over $NSOP_{1}$20 admits an $NSOP_{1}$21-invariant global extension (Hanson, 2023).

Fix a base $NSOP_{1}$22, an $NSOP_{1}$23-invariant global type $NSOP_{1}$24, and a Morley sequence $NSOP_{1}$25 generated by $NSOP_{1}$26 over $NSOP_{1}$27. A formula $NSOP_{1}$28 Kim-divides over $NSOP_{1}$29 with respect to $NSOP_{1}$30 if, whenever $NSOP_{1}$31 is such a Morley sequence with $NSOP_{1}$32, the set $NSOP_{1}$33 is inconsistent. The corresponding Kim’s Lemma for bi-invariant types says that if $NSOP_{1}$34 Kim-divides over $NSOP_{1}$35 with respect to some $NSOP_{1}$36-bi-invariant type extending $NSOP_{1}$37, then it must Kim-divide with respect to every $NSOP_{1}$38-bi-invariant extension of $NSOP_{1}$39 (Hanson, 2023).

The central theorem identifies the exact obstruction: a complete theory $NSOP_{1}$40 has the comb tree property if and only if Kim’s Lemma for bi-invariant types fails. Equivalently, $NSOP_{1}$41 has CTP precisely when there are a model $NSOP_{1}$42, a formula $NSOP_{1}$43, an $NSOP_{1}$44-heir-coheir $NSOP_{1}$45, and an $NSOP_{1}$46-coheir $NSOP_{1}$47 with $NSOP_{1}$48 such that $NSOP_{1}$49 Kim-divides over $NSOP_{1}$50 with respect to $NSOP_{1}$51 but does not Kim-divide with respect to $NSOP_{1}$52. The same paper proves that if Kim’s Lemma fails for a reliably $NSOP_{1}$53-invariant type against an extendibly $NSOP_{1}$54-invariant one, then $NSOP_{1}$55 has CTP (Hanson, 2023).

A second major result concerns existence. If $NSOP_{1}$56 is an invariance base, then every type $NSOP_{1}$57 admits a global extension $NSOP_{1}$58 that is reliably $NSOP_{1}$59-invariant. As a consequence, if $NSOP_{1}$60 has no CTP, then over any invariance base $NSOP_{1}$61, Kim-forking coincides with Kim-dividing for formulas with respect to extendibly, hence reliably, invariant types. In the special case where $NSOP_{1}$62 is a model, every $NSOP_{1}$63 extends to a reliable $NSOP_{1}$64-coheir, sharpening the usual statement that every type over a model extends to an $NSOP_{1}$65-strictly invariant type (Hanson, 2023).

5. Unified formulations and examples

Kruckman–Ramsey propose a unified reformulation designed to subsume both the $NSOP_{1}$66 and $NSOP_{1}$67 variants. A global $NSOP_{1}$68-invariant type $NSOP_{1}$69 is Kim-strict if for every $NSOP_{1}$70 and every realization $NSOP_{1}$71, one has $NSOP_{1}$72. A formula $NSOP_{1}$73 Kim-strictly divides if it divides along some Morley sequence of a Kim-strict invariant type, and universally Kim-strictly divides if it divides along every Morley sequence of every Kim-strict $NSOP_{1}$74-invariant extension of $NSOP_{1}$75. The New Kim’s Lemma is the statement that whenever $NSOP_{1}$76 Kim-divides over $NSOP_{1}$77, it in fact universally Kim-strictly divides over $NSOP_{1}$78 (Kruckman et al., 2023).

The reductions are immediate from the definitions. In an $NSOP_{1}$79 theory, Kim-forking equals forking over models, so Kim-strict coincides with strict invariant, and the new lemma specializes to the $NSOP_{1}$80 version. In an $NSOP_{1}$81 theory, every invariant type is automatically Kim-strict, so the new lemma specializes to the $NSOP_{1}$82 version. The paper also notes a “vacuum-cleaner lemma” asserting that in any theory every type over a model admits a Kim-strict invariant extension, so the universal quantification is non-vacuous (Kruckman et al., 2023).

The same work records examples and non-examples. $NSOP_{1}$83 and $NSOP_{1}$84 satisfy New Kim’s Lemma by direct analysis of $NSOP_{1}$85 in those theories. The Henson triangle-free graph $NSOP_{1}$86 fails it: there is a formula which strictly divides but does not universally strictly divide. The paper further introduces the Bizarre Tree Property $NSOP_{1}$87, shows that $NSOP_{1}$88 implies New Kim’s Lemma, and states that $NSOP_{1}$89 strictly contains both the $NSOP_{1}$90 and $NSOP_{1}$91 classes and lies inside the $NSOP_{1}$92 class (Kruckman et al., 2023).

6. Combinatorial characterizations, tree properties, and open directions

A later refinement studies higher-$NSOP_{1}$93 analogues for pairs of bi-invariant types. For $NSOP_{1}$94, the scheme $NSOP_{1}$95 says: for every parameter set $NSOP_{1}$96, every formula $NSOP_{1}$97, and any two $NSOP_{1}$98-bi-invariant types $NSOP_{1}$99, if $k$00 $k$01-divides along $k$02, then it also $k$03-divides along $k$04. In the special case $k$05, this is identified with the usual Kim’s Lemma for $k$06; the higher-$k$07 form tracks $k$08-inconsistency rather than plain inconsistency (Hanson, 28 Jul 2025).

The failure of $k$09 is characterized by explicit consistency–inconsistency configurations called weaves. A $k$10-weave is a family of parameters indexed by $k$11 such that every finite up-$k$12-comb is $k$13-inconsistent while every finite right-$k$14-comb is consistent. Hanson proves that if there are $k$15-invariant types $k$16 with $k$17 $k$18-strongly bi-invariant and $k$19 $k$20-strongly bi-invariant, or the corresponding semi-reliable variants when $k$21 or $k$22, and $k$23 $k$24-divides along $k$25 but not along $k$26, then $k$27 admits a strong $k$28-weave of depth $k$29. Conversely, for each $k$30, the following are equivalent: $k$31 satisfies $k$32-Kim’s Lemma; $k$33 satisfies $k$34-Kim’s Lemma over models; and $k$35 admits no strong $k$36-weave of depth $k$37 (Hanson, 28 Jul 2025).

The combinatorics become richer for $k$38. The “antichain” side of a $k$39-weave can be replaced by an arbitrary $k$40-free graph, that is, a cograph, and the existence of all finite cograph patterns is equivalent to a $k$41-weave of depth $k$42. A coarser obstruction is a $k$43-grid indexed by a linear order $k$44, where every strict chain in $k$45 yields a consistent family and every antichain yields a $k$46-inconsistent one. If an infinite $k$47-grid exists, then $k$48 fails several asymmetric Kim-lemma schemes over models involving coheir and strong-heir-coheir types, and under $k$49 it also fails generic stationary local character (Hanson, 28 Jul 2025).

These developments interact with other tree properties. The earlier bi-invariant analysis shows that the failure of Kim’s Lemma for strongly bi-invariant types is equivalent to the antichain tree property $k$50. It also shows that $k$51 theories satisfy Kim’s Lemma for strongly bi-invariant types and satisfy generic stationary local character; under a measurable cardinal, CTP and $k$52 admit corresponding dual local-character formulations. Open questions remain about the exact cardinals required, possible club-local character strengthenings, and whether New Kim’s Lemma is equivalent to a known syntactic property such as $k$53, $k$54, or $k$55 (Hanson, 2023, Kruckman et al., 2023).