Projected and near-projected embeddings (Q2027327): Difference between revisions

A smooth map \(f:N\to M\) is said to be a \(k\)-codimensionally projected embedding (a \(k\)-prem) if it factors into the composition of a smooth embedding \(N\hookrightarrow M \times \mathbb{R}^k\) and the projection \(M \times\mathbb{R}^k\to M\). The abbreviation `prem' was introduced by \textit{A. Szűcs} [Math. Proc. Camb. Philos. Soc. 100, 331--346 (1986; Zbl 0625.57017)]. In this paper, the authors study the connections between two notions saying that a stable smooth map \(f:N\to M\) is \(k\)-realizable if its composition with the inclusion map \(M\subset M\times\mathbb{R}^k\) is \(C^0\)-approximable by smooth embeddings; and that it is a \(k\)-prem if the same composition is \(C^{\infty}\)-approximable by smooth embeddings. It is obvious that if \(f\) is a \(k\)-prem, then it is \(k\)-realizable. In a main result, the authors show that the converse does not hold, thereby refuting the so-called `prem conjecture', by showing that for each \(n = 4k+ 3\ge 15\) there exists a stable smooth immersion of \(S^n\) into \(\mathbb{R}^{2n-7}\) that is \(3\)-realizable but not a \(3\)-prem. The authors give in fact such examples in a wide range of cases.