On Kim-independence (Q2178257)

In the article under review, the authors provide an analogous characterisation of the main lemma for simple theories for the independence relation of Kim-dividing in NSOP\(_1\) theories, and deduce many known results of simple theories to the context of Kim-forking for NSOP\(_1\) theories. Recall that a complete theory is simple if no formula \(\varphi(x,y)\) has the tree property with respect to some natural number \(k\), that is, if there is no sequence \((a_\alpha)_{\alpha \in {}^{<\omega}\omega}\) such that every branch of the tree \(\{\varphi(x, a_{\eta\!\upharpoonright n})\}_{n\in\mathbb N}\) is consistent (for every \(\eta\) in \( {}^{\omega}\omega\)), but at every node \(\alpha\) in \({}^{<\omega}\omega\), the collection \(\{\varphi(x, a_{\alpha\frown n})\}_{n\in\mathbb N}\) is \(k\)-inconsistent (that is, every subcollection of size \(k\) is inconsistent). The complete theory is NSOP\(_1\) if no formula \(\varphi(x,y)\) admits a sequence \((a_\alpha)_{\alpha \in {}^{<\omega}2}\) witnessing a tree of a particular kind: every branch of the tree is consistent, yet at every node \(\alpha\) in \({}^{<\omega}2\), whenever we consider any left-extension \(\beta\) of \(\alpha\frown 0\), the set \(\{\varphi(x, a_\alpha), \varphi(x, a_{\beta\frown 1})\}\) is inconsistent. This wider class of theories strictly contains all simple theories, yet many meaningful examples of theories which are not simple are NSOP\(_1\), among others, the collection of Frobenius fields (Section 9.3) or the model companion of vector spaces equipped with a symmetric bilinear form (Section 9.4). In the seminal work of \textit{B. Kim} and \textit{A. Pillay} [Ann. Pure Appl. Logic 88, No. 2--3, 149--164 (1997; Zbl 0897.03036)], it was shown that a theory is simple if and only if it is equipped with an invariant independence relation satisfying extension, symmetry, local and finite character, monotonicity and transitivity, as well as the independence theorem over models (or more generally for Lascar types). Furthermore, the independence relation must agree with non-dividing independence. The key feature in order to prove this result is the so-called Kim's Lemma: An instance of a formula \(\varphi(x, a)\) divides over a set \(B\) if and only if for some (or equivalently, for every) Morley sequence \((a_n)_{n\in \mathbb N}\) with \(a_0\equiv_B a\), we have that the intersection \[ \bigcap_{n \in \mathbb N} \varphi(x, a_n) \] is inconsistent. Whilst this result is no longer true for NSOP\(_1\) theories, it is true if one adapts the notion of a Morley sequence and of dividing. Fix a \(B\)-invariant type \(q\). A formula \(q\)-divides over \(B\) if the intersection \[ \bigcap_{n\in \mathbb N} \varphi(x, a_n) \] is inconsistent, for some (or equivalently for every) Morley sequence \((a_n)_{n\in \mathbb N}\) of realizations of \(q\). The authors then show (Theorem 3.16) that a theory is NSOP\(_1\) if and only if, for every type \(p\) over a model \(M\), if the formula \(q\)-divides for some global \(M\)-invariant type \(q\) extending \(p\), then it \(r\)-divides for every global \(M\)-invariant type \(r\) extending \(p\). They denote this by Kim-dividing of the instance \(\varphi(x, a)\), with \(a\) realizing \(p\) and observe that many of the known arguments for simple theories carry over naturally, such as Lemma 3.18. In particular, Kim-dividing equals Kim-forking over models (Proposition 3.19), which is a symmetric notion over models (Theorem 5.16) and satisfies the corresponding version of the independence theorem over models (Theorem 6.5). They deduce a characterization à la Kim-Pillay of NSOP\(_1\) theories (Theorem 9.1) along the lines of [\textit{A. Chernikov} and \textit{N. Ramsey}, J. Math. Log. 16, No. 2, Article ID 1650009, 41 p. (2016; Zbl 1402.03043)].