Division pairs: a new approach to Moufang sets (Q2340224)

The description of Moufang sets initiated by \textit{T. De Medts} and \textit{R. M. Weiss} [Math. Ann. 335, No. 2, 415--433 (2006; Zbl 1163.17031)], was roughly speaking, given in terms of pairs \((U, \tau)\) where \(U\) is a group and \(\tau\) a bijection of the set of non-trivial elements of \(U\) onto itself. The pair \((U,\tau)\) has to satisfy suitable conditions for the construction to yield a Moufang set, denoted \(\mathbb M(U, \tau)\). However, as pointed out by De Medts (\url{http://cage.ugent.be/simtdemedts/notes/catmouf.pdf}), and by \textit{T. De Medts} and \textit{Y. Segev} [Innov. Incidence Geom. 9, 79--122 (2009; Zbl 1233.20028)], the Moufang set \(\mathbb M(U, \tau)\) does not determine the data \((U, \tau)\) uniquely. In this work, the concept of division pairs, a non-abelian version of Jordan division pairs, is introduced and a categorical equivalence between division pairs and Moufang sets is established. This is used to remedy the non-uniqueness occurring in the description of Moufang sets initiated by De Medts and Weiss. Let \(V = (V^+,V^-)\) be a pair of groups. We will denote \(\dot{V}^\sigma:=V^\sigma\setminus\{0\}\). A pre-division pair is a pair \(V\) as above together with a pair of maps \(j_\sigma: \dot{V}^\sigma\to\dot{V}^{-\sigma}\) which are inverses of each other: \(j_{-\sigma}j_\sigma = Id\) for \(\sigma\in \{+, -\}\). If \(V\) and \(W\) are pre-division pairs, a homomorphism \(h = (h^+, h^-) : V \to W\) is a pair of injective group homomorphisms \(h_\sigma: V^\sigma\to W^\sigma\) such that, for any \(\sigma\in \{+, -\}\) and all \(x \in\dot{V}^\sigma\), we have \(h_\sigma(x)\in\dot{W}^\sigma\) and \(h_{-\sigma}(j_\sigma(x)) = j_\sigma(h_\sigma(x))\). Pre-division pairs form a category, denoted \textbf{pre-div}. Then division pairs are introduced by requiring additional conditions and it is proved, using the quasi-inverse in Jordan pairs, that a Jordan division pair is a division pair in the sense of this paper. Denote by \(\text{Sym}(X)\) the group of permutations of the set \(X\) and by \(\text{sbgr}(G)\) the set of subgroups of the group \(G\). A pre-Moufang set is a pair \(M = (X,{\mathcal U})\) consisting of a non-empty set \(X\) and a map \({\mathcal U} : X \to \text{sbgr}(\text{Sym}(X))\) satisfying Tits's first axiom: (M1) for all \(p \in X\), the group \({\mathcal U} (p)\) acts simply transitively on \(X\setminus\{p\}\). If besides, \(M\) satisfies: (M2) for each \(p\in X\), the group \({\mathcal U} (p)\) normalizes the set \({\mathcal U} (X)\subset \text{sbgr}(\text{Sym}(X))\), \noindent we will say that \(M\) is a Moufang set. We can also consider based pre-Moufang sets with base points \(b^+\) and \(b^-\), and define a half-Moufang set of type \(+\) (respectively \(-\)) by requiring that \({\mathcal U} (b^+)\) (respectively \({\mathcal U} (b^-)\)) consist of automorphisms. In Section 3, the author define functors \(\mathbb D\) from half-Moufang sets to pre-division pairs and \(\mathbb P\) from pre-division pairs to half-Moufang sets, and show that \(\mathbb D\) and \(\mathbb P\) are, up to natural isomorphism, inverses of each other. By restricting these functors to division pairs and Moufang sets, respectively, the author obtains the main result (Theorem 3.5), the categorical equivalence of Moufang sets and division pairs. In the last Section of the work, the author relates his approach with the construction of Moufang sets from pairs \((U, \tau)\), as outlined above.