Representation and classification of Coxeter monoids (Q915871)

Coxeter monoids are introduced in this paper to study finiteness conditions for chamber systems and amalgams of groups. A chamber system \({\mathcal C}=(C,(p_ i)_{i\in I})\) can be viewed as a monoid with set of generators \((p_ i)_{i\in I}\), where the product operation is the usual product of equivalence relations and the unit element is the identity relation. This perspective is enlightnening when the chamber system admits something like a Coxeter diagram; namely, there is a matrix \((m_{ij})_{i,j\in I}\), with \(m_{ii}=2\) (i\(\in I)\), such that every residue of \({\mathcal C}\) of type \(\{\) i,j\(\}\) has pair of diameters \((m_{ij},m_{ji})\), \(m_{ij}\) (respectively, \(m_{ji})\) being computed starting from panels of cotype i (of cotype j, respectively). If that is the case, then the following hold (1) \(p_ ip_ jp_ ip_ jp_ i...=p_ jp_ ip_ jp_ ip_ j...=p_ i\vee p_ j\) (i,j\(\in I\), \(i\neq j)\); \(m_{ij}\) factors; \(m_{ji}\) factors,\ (2) \(p^ 2_ i=p_ i\) (i\(\in I).\) If \({\mathcal C}=(G,(p_ i)_{i\in I})\) is a parabolic system for a group G, then (1) and (2) above become as follows: \[ p_ ip_ jp_ ip_ jp_ i...=p_ jp_ ip_ jp_ ip_ j...=<p_ i,p_ j>;\quad m_{ij}\quad factors;\quad m_{ji}\quad factors\quad p^ 2_ i=p_ i. \] Coxeter monoids are defined taking the set of relations (1) and (2) as a presentation of the monoid, where \(p_ ip_ j\) is substituted by a symbol \(p_{ij}\) characterized by the properties \(p_{ij}p_ j=p_{ij}p_ i=p_ ip_{ij}=p_ jp_{ij}=p_{ji}\). A Coxeter monoid is uniquely determined by its `generalized Coxeter matrix' \(M=(m_{ij})_{i,j\in I}.\) The problem which most naturally arises in this approach is finding a property of Coxeter monoids such that, if a Coxeter monoid \({\mathcal M}\) satisfies that property, then all chamber systems obtained as homomorphic images of \({\mathcal M}\) have finite diameters (of course, the chamber systems obtained in this way are precisely those having the same generalized Coxeter matrix as \({\mathcal M})\). The author shows `admitting an attractor' is the property we are looking for, where an element a of a monoid \({\mathcal M}\) is an attractor in \({\mathcal M}\) if \(ax=xa=a\) for every element x of \({\mathcal M}\). Roughly speaking, Coxeter monoids admitting an attractor are those of `spherical type', in some sense. However, it is worth remarking that a Coxeter monoid may admit an attractor without being finite. Finding a good explanation of what this phenomenon means in `real life' is one of the most interesting open problems in the paper under review. Finally the author studies some generalized Coxeter matrices, checking which of them give rise to Coxeter monoids admitting attractors, namely which of them are `spherical'. Of course, all classical Coxeter matrices of spherical type are also `spherical' in this new sense.