Properties expressible in small fragments of the theory of the hyperfinite \(\mathrm{II}_1\) factor (Q2022201)

This paper lies at the intersection of logic and operator algebras. Specifically, it uses continuous model theory for \(\mathrm{II}_1\) factors to give partial answers to Popa's factorial embedding problem. \(\mathrm{II}_1\) factors are semifinite von Neumann algebras of trivial center. Chief among (separable) \(\mathrm{II}_1\) factors is the hyperfinite \(\mathrm{II}_1\) factor \(\mathcal R\), the only separable amenable \(\mathrm{II}_1\) factor (up to isomorphism). As factor represent the building block, through the direct integral decomposition, the study of the properties of \(\mathrm{II}_1\) factors, and specifically that of \(\mathcal R\), became a major theme in von Neumann algebra theory in the last 40 years. Connes' embedding problem (CEP) asked whether all separable (in von Neumann algebra terms) \(\mathrm{II}_1\) factors can be embedded into an ultrapower \(\mathcal R^\mathcal U\). CEP was given a negative answer, but many questions on the possible embedding of factors into ultrapowers are yet to be answered. Among these, Popa asked whether all \(\mathcal R^\mathcal U\)-embeddable factors admit a factorial embedding into \(\mathcal R^\mathcal U\), that is, an embedding whose relative commutant is a factor itself. This problem is known as Popa's factorial commutant embedding problem (FCEP), and it has a positive answer for \(\mathcal R\). Further partial answers have been provided in [\textit{J. Dixmier} and \textit{E. C. Lance}, Invent. Math. 7, 226--234 (1969; Zbl 0174.18701); \textit{S. Popa}, J. Funct. Anal. 266, No. 9, 5818--5846 (2014; Zbl 1305.46052)]. Recently, the FCEP has been studied through the lenses of model theory, notably in recent work of \textit{S. Atkinson}, the first author and \textit{S. Kunnawalkam Elayavalli} [``Factorial commutants and the generalized Jung property for \(\mathrm{II}_1\) factors'', Preprint, \url{arXiv 2004.02293}], who showed that if \(M\) is elementary equivalent to \(\mathcal R\) then \(M\) satisfies the FCEP. Further results were obtained in [\textit{I. Goldbring}, Proc. Am. Math. Soc. 148, No. 11, 5007--5012 (2020; Zbl 1456.03058)] for property (T) factors. A modification of FCEP is the study of the Brown property. A \(\mathrm{II}_1\) factor \(M\) has the Brown property if, for all separable subfactors \(N\) of \(M^\mathcal U\) there is subfactor \(P\) with \(N\subseteq P\subseteq M^\mathcal U\) such that the embedding of \(P\) inside \(M^\mathcal U\) has factorial relative commutant. Once again, the first author, Atkinson and Kunnawalkam Elayavalli [loc. cit.] showed that factors which are elementary equivalent to \(\mathcal R\) satisfy the Brown property. The authors work on both FCEP and the Brown property. Their main results (Theorems A and C) show that to satisfy FCEP (resp., the Brown property), it is enough to share only a fairly weak fragment of the theory of \(\mathcal R\). The main techniques used involve the quantifier analysis of the property of satisfying FCEP and of the Brown property to measure their `quantifier depth'. This is done in \S2 and \S4 respectively.