Fold maps on 4-manifolds (Q1417432)

In this very nice paper, the author gives a complete characterization of those closed orientable 4-manifolds which admit fold maps into \({\mathbb R}^3\), i.e., smooth maps into \({\mathbb R}^3\) with only fold singularities. Recall that a singular point of a smooth map from an \(n\)-dimensional manifold to a \(p\)-dimensional manifold with \(n\geq p\) is of fold type if the map can be written in the form \((x_1,\ldots, x_n) \mapsto (x_1, \ldots, x_{p-1},\pm x_p^2, \ldots, \pm x_n^2)\) for some local coordinates. The motivation for this study goes back to the work of Thom who tried to generalize the notion of Morse function to generic maps into Euclidian spaces. Folds maps replace submersions when the target manifold is of dimension less than the source manifold, and singularities of fold type are the simplest among all generic singularities. The main result af the paper is given by the following theorem: Let \(M\) be a closed connected 4-manifold. Then the following three are equivalent. (i) There exists a fold map of \(M\) into \(\mathbb{R}^3\). (ii) The intersection form of \(M\) is not isomorphic to \(\pm (1)\), \(\pm \left(\begin{smallmatrix}1&0\\ 0&1\end{smallmatrix}\right)\). (iii) There exists a cohomology class \(v\in H^2(M;{\mathbb Z})\) such that \(v\cup v= p_1(M)\in H^4(M;{\mathbb Z})\), where \(p_i\) denotes the \(i\)-th Pontrjagin class. This result shows that there are a lot of 4-manifolds which admit fold maps into \({\mathbb R}^3\), but which have no fiberwise epimorphisms \(TM\oplus\varepsilon^1\to\varepsilon^3\), or no fold maps into \({\mathbb R}^2\). Since there exists a fiberwise emiporphism \(TM\oplus\varepsilon^1\to\varepsilon^p\) if and only if there exist \(p\) smooth sections of \(TM\oplus\varepsilon^1\) which are everywhere linearly independent, then the existence problem of fold maps is related to the (stable) vector field problem. The author discusses their relationship in the paper.