Moufang twin trees of prime order (Q317289)

Twin trees are precisely the non-spherical twin buildings of rank 2. The Moufang condition in spherical buildings characterizes those buildings of algebraic origin. As the authors argue a classification of all Moufang twin trees does not seem to be feasible, but there is hope to achieve a classification of locally finite Moufang twin trees. The paper under review is a major step in the determination of all Moufang twin trees of prime order \(p\). Since Moufang sets of cardinality \(p+1\) are known and because the problem can be reduced to the classification of RGD-systems in which all root groups \(U_\alpha\) have order \(p\), it remains to classify the possible commutation relations. The authors introduce \(\mathbb{Z}\)-systems of prime order \(p\). These consist of a group \(X\), a set of generators \(x_k\) of \(X\) indexed by \(\mathbb{Z}\) such that the subgroup generated by the \(x_n,\ldots,x_m\) where \(n\leq m\) has order \(p^{m-n+1}\) and such that the shift map \(x_k\mapsto x_{k+2}\) extends to an automorphism of \(X\). Such \(\mathbb{Z}\)-systems arise from Moufang twin trees of order \(p\). In this case, \(X\) is the unipotent horocyclic group of the twin tree, which in fact encodes the commutation relations of the Moufang twin tree, and the \(x_n\) are generators of the root groups. Much of the paper is devoted to the investigation of \(\mathbb{Z}\)-systems, their shift-invariant normal subgroups, and to establish that \(X\) has infinite abelianization \(X/X'\). The main result of the paper is purely group theoretical, namely that a \(\mathbb{Z}\)-system of prime order is nilpotent of class at most 2. As a consequence the unipotent horocyclic group of a Moufang twin tree of prime order is nilpotent of class at most 2.