There is no tame triangulation of the infinite real Grassmannian (Q5933470)

It is an open question whether or not the infinite real Grassmanian \(G(k,\mathbb R^\infty)\), for \(k\geq 3\), can be triangulated. The authors prove that at least there is no triangulation which is, in a sense, combinatorially nice. More precisely, by choice of a fixed basis, to each \(k\)-dimensional subspace of~\(\mathbb R^n\) there is an associated oriented matroid. Hence, the isomorphism classes of oriented matroids form a partition of the Grassmannian~\(G(k,\mathbb R^n)\). The blocks of this partition are called matroid strata. A triangulation of \(G(k,\mathbb R^n)\) or \(G(k,\mathbb R^\infty)\) is tame if it refines this oriented matroid stratification. By the main result of the paper under review such a tame triangulation of \(G(k,\mathbb R^\infty)\) does not exist, provided that \(k\geq 3\). Assuming the contrary the authors arrive at a contradiction via an explicit construction in the real affine plane which would yield an infinite number of simplices in a compact subspace~\(G(3,\mathbb R^8)\).