Affine hovels (Q635871): Difference between revisions

\textit{S. Gaussent} and the author [Ann. Inst. Fourier 58, No. 7, 2605--2657 (2008; Zbl 1161.22007)] investigated certain Kac-Moody groups over ultrametric fields and associated with them incidence spaces \(\mathcal{I}\) that almost looked like Bruhat-Tits buildings for reductive groups over local fields. \(\mathcal{I}\), which they called a hovel, is the union of apartments but there may be two points in \(\mathcal{I}\) that are not contained in a common apartment. In the paper under review the author introduces affine hovels axiomatically. The axioms given are similar to \textit{J. Tits}' ones for affine buildings [Lect. Notes Math. 1181, 159--190 (1986; Zbl 0611.20026)]. The construction is based on the notion of apartment, which is a copy of a finite-dimensional real affine space on which a reflection group acts. One essentially has a Coxeter group extended by the coroot lattice. Motivated by the theory of affine buildings the author verifies that many familiar properties of affine buildings carry over to affine hovels. In particular, he shows that Tits cones confer a pre-order, that each germ of sectors admits a retraction centred at it and that one obtains at infinity a pair of twin buildings and two micro-affine buildings. In section 5 of the paper the author assumes that the hovel is ordered. Under this assumption he shows that the local space of directions about a point can be equipped with the structure of a twin building. The last section serves to verify that the incidence spaces from certain Kac-Moody groups by S. Gaussent and the author [loc. cit.] are indeed ordered affine hovels in the sense introduced in the present note (but they are not affine buildings).