Decomposing manifolds into Cartesian products (Q721478): Difference between revisions

Let \(\mathcal{M}^n\) and \(\mathcal{N}^k\) be two closed, oriented \(n\)-dimensional manifolds which cannot be split into products of closed, orientable manifolds (\(\neq\) point), where \(k \leq n\). This paper deals with a natural question: Can \(\mathcal{M}^n \times \mathcal{N}^k\) be decomposed into products of manifolds of dimension \(\leq n - 1\)? If such an \(\mathcal{N}^k\) exists, then \(\mathcal{M}^n\) is called stably decomposable by \(\mathcal{N}^k\). Otherwise, \(\mathcal{M}^n\) is called stably nondecomposable. For \(n=1,2\), every \(\mathcal{M}^n\) is stably nondecomposable. That is due to an old result of Borsuk saying that a closed, \(n\)-dimensional manifold has at most one decomposition into the Cartesian product of indecomposable factors of dimension \(\leq 2\). This paper shows that this is also true for \(n = 3\). However, there are plenty of oriented, closed, nondecomposable \(\mathcal{M}^4\) which can be stably decomposed by 2- and 3-spheres. More precisely, it is shown that there are infinitely many \(\mathcal{M}^4\) such that \(\mathcal{M}^4 \times S^k =S^1 \times S^k \times \mathbb{RP}^3\) (\(k=2,3,4\)). The key ingredient is the example first constructed by \textit{S. Weinberger} [Isr. J. Math. 59, 1--7 (1987; Zbl 0638.57009)] (or see [the authors, Topology 27, No. 4, 443--457 (1988; Zbl 0664.57018)]).