1
$\begingroup$

The weak Whitney embedding theorem states that any continuous function from an $n$-dimensional manifold to an $m$-dimensional manifold may be approximated by a smooth embedding provided $m > 2n$.

Some differential topology textbooks state that this approximate version of Whitney embedding theorem cannot be improved. However, they only gave simple counterexamples from 1 to 2 dimensions. ('if $S^1$ is mapped into $\mathbb{R^2}$ so that the image curve crosses itself like a figure 8, no sufficiently close approximation can be injective [Hirsch1976, Differential Topology].')

My question is, can we construct such a counterexample to show that $m=2n+1$ is the lower bound? In other words, is it possible to construct a continuous mapping $f: [0,1]^n →R^m$, for any $\epsilon>0$, there is no smooth embedding $g: [0,1]^n →R^m$ such that $|f(x)-g(x)|<\epsilon,\forall x \in [0,1]^n, n < m <=2n$?

Improve this question
$\endgroup$
3
  • $\begingroup$ What do you mean by the "weak Whitney theorem"? What does mean that a continuous function cannot be embedded? If you mean that the manifold cannot be embedded, you did not ay into what. "Hirsh [...] said [...] that the approximation version cannot be improved". Approximation version of what? Your question should be self-contained and if possible (which is possible here) should not require the readers to read external sources to understand the question. $\endgroup$
    – Piotr Hajlasz
    Sep 25 at 11:33
  • $\begingroup$ I was in a hurry to write and didn't pay attention to these issues. Thank you for your advice. I will update my questions and elaborate in detail later $\endgroup$
    – li ang Duan
    Sep 25 at 12:38
  • 2
    $\begingroup$ Any continuous map where you have two discs which map linearly to a neighborhood of 0 in $0 \times \Bbb R^n$ and $\Bbb R^n \times 0$, respectively, cannot be perturbed to be injective by a small perturbation. (Consider the linking number of two small spheres around 0; this will be 1, but if a small perturbation made this map injective, the linking number would be 0.) It suffices to take $T^n \to \Bbb R^{2n}$ by taking products of your map with itself. $\endgroup$
    – mme
    Sep 25 at 14:14

0

Reset to default

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.