Cette thèse se propose d'étudier sous divers points de vue les groupes d'automorphismes d'immeubles. Un de ses objectifs est de mettre en valeur les différences autant que les analogies entre les immeubles affines et non affines. Pour appuyer cette dichotomie, on y démontre que les groupes d'automorphismes d'immeubles non affines n'ont jamais de paire de Gelfand, contrairement aux immeubles affines. Dans l'autre sens, pour souligner l'analogie entre immeubles affines et non affines, on définit une nouvelle notion de bord combinatoire d'un immeuble. Dans le cas des immeubles affines, ce bord s'identifie au bord polyédral. On relie la construction de ce bord à d'autres constructions déjà existantes, par exemple, la compactification de Busemann du graphe des chambres. La compactification combinatoire est également isomorphe à la compactification par la topologie de Chabauty de l'ensemble des chambres, sous des hypothèses de transitivité. On relie aussi le bord combinatoire à un autre espace, généralisant une construction de F. Karpelevic pour les espaces symétriques : celle du bord raffiné d'un espace CAT(0).On démontre alors que les points du bord paramètrent les sous-groupes moyennables maximaux de l'immeuble, à indice fini près. Enfin, on prouve que l'action du groupe d'automorphismes d'un immeuble localement fini sur le bord combinatoire de ce dernier est moyennable, fournissant ainsi des résolutions en cohomologie bornée et des applications bord explicites. Ceci donne aussi une nouvelle preuve que ces groupes satisfont la conjecture de Novikov. / The object of this thesis is the study, from different point of views, of automorphism groups of buildings. One of its objectives is to highlight the differences as well as the analogies between affine and non-affine buildings. In order to support this dichotomy, we prove that automorphism groups of non-affine buildings never have a Gelfand pair, contrarily to affine buildings.In the other direction, the analogy between affine and non-affine buildings is supported by the new construction of a combinatorial boundary of a building. In the affine case, this boundary is in fact the polyhedral boundary. We connect the construction of this boundary to other compactifications, such as the Busemann compactification of the graph of chambers. The combinatorial compactification is also isomorphic to the group-theoretic compactification, which embeds the set of chambers into the set of closed subgroups of the automorphism group. We also connect the combinatorial boundary to another space, which generalises a construction of F. Karpelevic for symmetric spaces : the refined boundary of a CAT(0) space.We prove that the maximal amenable subgroups of the automorphism group are, up to finite index, parametrised by the points of the boundary. Finally, we prove that the action of the automorphism group of a locally finite building on its combinatorial boundary is amenable, thus providing resolutions in bounded cohomology and boundary maps. This also gives a new proof that these groups satisfy the Novikov conjecture.

For a locally compact group G, the papers [13] and [7] have many results about G-invariant subsets of G-modules, and the relationship between G-module maps, L1(G)-module maps and M(G)-module maps. In both papers, the results were given for one specific module action. In this thesis we extended many of their results to arbitrary Banach G-modules. In addition, we give detailed proofs of most of the results found in the first section of the paper [21].

For a locally compact group G, the papers [13] and [7] have many results about G-invariant subsets of G-modules, and the relationship between G-module maps, L1(G)-module maps and M(G)-module maps. In both papers, the results were given for one specific module action. In this thesis we extended many of their results to arbitrary Banach G-modules. In addition, we give detailed proofs of most of the results found in the first section of the paper [21].