On non-abelian \(C\)-minimal groups (Q1400606)

A \(C\)-structure is a structure \({\mathcal M}=(M;C,\dots)\), where \(C\) is a ternary relation on \(M\) satisfying the following axioms: (1) \(C(x,y,z)\to C(x,z,y)\), (2) \(C(x,y,z)\to \neg C(y,x,z)\), (3) \(C(x,y,z)\to (C(w,y,z)\vee C(x,w,z))\), (4) \(x\neq y\to \exists z(y\neq z\wedge C(x,y,z))\). A prototypical example: if \((T,<)\) is an infinite tree such that there are incomparable elements above any node, \(M\) is the set of all maximal chains of \(T\), and \(C(x,y,z)\) means ``\(y\) and \(z\) branch above where \(x\) and \(y\) branch'' then the relation \(C\) satisfies (1)--(4). Such relations \(C\) were considered in the 80s in the context of permutation group theory by P. Cameron and then, in the axiomatic form, by S. Adeleke and P. Neumann. An expansion of a field \(F\) by a ternary relation \(C\) is called a \(C\)-field if it is a \(C\)-structure, and the maps \(x\mapsto x+a\) and \(x\mapsto xb\) preserve \(C\), for any \(a\in F\) and nonzero \(b\in F\). Any non-trivially valued field, with a valuation \(v\), becomes a \(C\)-field if we define \(C(x,y,z)\) as \(v(z-x)<v(z-y)\). Similarly, an expansion of a group \(G\) by a ternary relation \(C\) is called a \(C\)-group if it is a \(C\)-structure, and the map \(x\mapsto axb\) preserves \(C\), for any \(a,b\in G\). In his earlier papers the author considered the notion of valued group (with values in a chain with a maximal element without immediate predecessor) and noted that, like a valued field, any valued group can be naturally endowed with a structure of a \(C\)-group. \textit{H. D. Macpherson} and \textit{C. Steinhorn} [Ann. Pure Appl. Logic 79, 165-209 (1996; Zbl 0858.03039)] introduced \(C\)-minimality as a variant of o-minimality: a \(C\)-structure \(\mathcal M\) is called \(C\)-minimal if in any structure \(\mathcal N\) elementarily equivalent to \(\mathcal M\) every definable set is quantifier-free definable in the \(C\)-reduct of \(\mathcal N\). They showed that if a non-trivially valued field is algebraically closed then it is \(C\)-minimal; \textit{D. Haskell} and \textit{D. Macpherson} [Ann. Pure Appl. Logic 66, 113-162 (1994; Zbl 0790.03039)] proved the converse. In Ann. Pure Appl. Logic 110, 1-22 (2001; Zbl 0985.03023) the author investigated \(C\)-minimal abelian groups; an example of such a group is the additive group of any valued field. The paper under review deals with \(C\)-minimal valued groups that need not be virtually abelian. The main result: any \(C\)-minimal valued group is virtually nilpotent. All known examples of \(C\)-minimal valued groups are virtually abelian or virtually nil-2. The author gives some information about \(C\)-minimal valued groups that are neither virtually abelian nor virtually nil-2; for example, all of them are of finite exponent, and every definable subgroup has a connected component.