Sp(3) structures on 14-dimensional manifolds (Q387261)

The article studies \(14\)-dimensional Riemannian manifolds on which the Lie group Sp(3) acts infinitesimally. The purpose is at least twofold. First, these objects befit the pattern laid out by \textit{P. Nurowski} comprising the rank-two symmetric spaces \(\mathrm{SU}(3)/\mathrm{SO}(3)\), \(\mathrm{SU}(3)\times\mathrm{SU}(3)/\mathrm{SU}(3)\), \(\mathrm{SU}(6)/\mathrm{Sp}(3)\), \(E_6/F_4\), of respective dimensions \(n=5,8,14,26\) [J. Geom. Phys. 58, No. 9, 1148--1170 (2008; Zbl 1215.53032)]. But the study is also intimately related to isoparametric hypersurfaces in spheres [\textit{E. Cartan}, Math. Z. 45, 335--367 (1939; Zbl 0021.15603)]. In fact, a hypersurface with \(3\) distinct principal curvatures immersed in \(S^{n-1}\subset \mathbb{R}^n\) is determined by a certain symmetric traceless \(3\)-tensor. The possible irreducible isotropy groups of the latter are \(\mathrm{SO}(3)\), \(\mathrm{SU}(3)\), \(\mathrm{Sp}(3)\), \(F_4\) and may occur only if \(n\) is one of the `distinguished' numbers above. The smallest dimension \(5\) was thoroughly studied in [\textit{M. Bobieński} and \textit{P. Nurowski}, J. Reine Angew. Math. 605, 51--93 (2007; Zbl 1128.53017); the first author et al., Ann. Global Anal. Geom. 40, No. 1, 67--84 (2011; Zbl 1218.53047)]. Also the \(8\)-dimensional case has been considered, for instance in [\textit{F. Witt}, Adv. Math. 219, No. 6, 1972--2005 (2008; Zbl 1170.53020)], so next on this agenda is \(n=14\). Here the authors describe the algebraic types of \(\mathrm{Sp}(3)\)-structures using representation theory, and determine some topological obstructions for their existence on compact oriented manifolds \(M^{14}\): these involve the vanishing of the Euler characteristic and most Stiefel-Whitney classes. As an interesting spin-off of sorts, the paper considers metric connections on compact connected Lie groups \(G\) with bi-invariant metric. The bi-invariant connection (with skew torsion), the canonical (metric) connection and the characteristic connection are shown to coincide on any such \(G\). The authors also prove a general uniqueness criterion for characteristic connections, thus pursuing the line of thought of [the first two authors, Math. Ann. 328, No. 4, 711--748 (2004; Zbl 1055.53031); \textit{C. Olmos} and \textit{S. Reggiani}, J. Reine Angew. Math. 664, 29--53 (2012; Zbl 1241.53045); \textit{P.-A. Nagy}, J. Lie Theory 23, No. 1, 1--33 (2013; Zbl 1264.53033)]. A big plus to the paper is the number of examples considered, including large families of homogeneous Riemannian manifolds admitting Sp(3)-structures that are described in depth, e.g. \(\mathrm{SU}(5)/\mathrm{Sp}(2)\), \(\mathrm{SU}(4)/S^1\) and quotients. As everything is explicit, this article is very valuable to anybody working with \(G\)-structures and the theory of connections on Riemannian manifolds.