A 15-vertex triangulation of the quaternionic projective plane (Q2316796)

\textit{U. Brehm} and \textit{W. Kühnel} [Math. Ann. 294, No. 1, 167--193 (1992; Zbl 0734.57017)] found a triangulated 8-manifold \(M^8_{15}\) with 15 vertices which is not a sphere. They gave a heuristic argument supporting that \(M^8_{15}\) is homeomorphic with the quaternionic projective plane, but that was not a proof. From the decomposition into 9:6 vertices (8-simplex vs. boundary of 5-simplex) it was clear that it is a triangulated manifold ``like a projective plane'' in the sense by \textit{J. Eells jun.} and \textit{N. H. Kuiper} [Publ. Math., Inst. Hautes Étud. Sci. 14, 181--222 (1962; Zbl 0109.15701)] and \textit{L. Kramer} [J. Differ. Geom. 64, No. 1, 1--55 (2003; Zbl 1068.57019)]. The Hirzebruch formula for the signature implies that it is sufficient to compute the first or second Pontryagin class but that was out of range for a long time, even though theoretically there was a formula by Gabrielov-Gelfand-Losik. But that could not practically be evaluated. In the remarkable paper under review the author manages to compute the first Pontryagin class of \(M^8_{15}\) in a purely combinatorial way. This is based on previous work by \textit{A. A. Gaifullin} [Izv. Math. 68, No. 5, 861--910 (2004; Zbl 1068.57022); translation from Izv. Ross. Akad. Nauk Ser. Mat. 68, No. 5, 13--66 (2004)]. Roughly speaking Gaifullin expressed the first Pontryagin class in terms of bistellar moves between all 3-dimensional links in the triangulated 8-manifold. This defines cycles in the graph \(\Gamma_2\) consisting of all oriented 2-spheres as vertices and bistellar moves as edges. Since \(M^8_{15}\) has \(\binom{15}{5}= 3003\) 4-simplices (with more than 60 combinatorial types even after regarding the automorphism group) one has to examine as many 3-dimensional links. Ultimately the calculation depends on a computer algorithm. This in turn is based on the mathematical software BISTELLAR which is available from the homepage of Frank H. Lutz.