The center conjecture for spherical buildings of types \(F _{4}\) and \(E _{6}\) (Q719683)

The center conjecture for buildings states that a convex subcomplex \(K\subseteq B\) of a spherical building \(B\) is either a sub-building or the action of \(\mathrm{Stab}_{\mathrm{Aut}(B)}(K)\) on \(K\) of the automorphisms of \(B\) preserving \(K\) has a fixed point. It was first proved for irreducible spherical buildings of classical types \(A_n\), \(B_n\) and \(D_n\) by \textit{B. Mühlherr} and \textit{J. Tits} [J. Algebra 300, No. 2, 687--706 (2006; Zbl 1101.51004)] using combinatorial methods and the incidence geometries of the respective buildings. Recently, \textit{C. Parker} and \textit{K. Tent} [Arch. Math. 97, No. 2, 125--128 (2011; Zbl 1226.51003)] gave a short proof for these classical types. In the paper under review, the authors prove the center conjecture for irreducible spherical buildings of types \(F_4\) and \(E_6\). They also add a new proof for the classical types by using the methods developed for the \(F_4\) and \(E_6\) types. The second author [``The center conjecture for thick spherical buildings'' (2009), \url{arXiv:0909.2761}] dealt with types \(E_7\) and \(E_8\). Hence, the center conjecture holds for all spherical buildings without factors of type \(H_4\), and, in particular, for all thick spherical buildings. The methods used by the authors are differential-geometric and based on the theory of CAT(1) spaces and on the geometric approach to buildings from the perspective of comparison geometry as in [\textit{B. Kleiner} and the first author, Publ. Math., Inst. Hautes Étud. Sci. 86, 115--197 (1997; Zbl 0910.53035)]. The arguments rely on the specific features of \(F_4\)- and \(E_6\)-geometries. The authors start by giving detailed descriptions of Coxeter complexes of types \(F_4\) and \(E_6\) and by collecting results that allow them to recognize when a convex subcomplex \(C\) of a spherical building \(B\) is a sub-building or when the stabilizer of \(C\) in the automorphism group of \(B\) has a fixed point in \(C\). They then consider a counterexample \(K\subseteq B\) to the center conjecture, that is, \(K\) is a convex subcomplex but is not a sub-building and the action of \(\mathrm{Stab}_{\mathrm{Aut}(B)}(K)\) on \(K\) has no fixed point. Such a counterexample must have dimension at least 2 and cannot contain a singular sphere of codimension one (in \(K\)). Furthermore, \(K\) can neither contain a \(\mathrm{Stab}_{\mathrm{Aut}(B)}(K)\)-invariant subset with circumradius \(< \pi/2\) nor one with diameter \(\leq \pi/2\). The strategy is to show that all vertices of \(K\) have antipodes in \(K\). It follows then by [\textit{J.-P. Serre}, Sém. Bourbaki, Astérisque 299, 195--217, Exp. No. 932 (2005; Zbl 1156.20313)] that \(K\) is a sub-building, and a contradiction is obtained. In case of type \(F_4\) the pattern of 1- and 4-vertices in \(K\) is investigated, because vertices of these types are better separated from each other than 2- and 3-vertices. The crucial case is when \(K\) contains vertices of type 1 or 4 without antipodes in \(K\). It requires a detailed analysis of interior points and points whose links contain certain 2-dimensional hemispheres. The case of type \(E_6\) is even more involved.