12
$\begingroup$

Asked once on SE-mathematics.

Let $U$ be an open subset in $\mathbb{R}^n$, $m\in\mathbb{N}$, $1\leq m<n$ and let $$\mathcal{C}^k_{\leq m}(U,\mathbb{R}^n):=\lbrace g\in\mathcal{C}^k(U,\mathbb{R}^n)\mid\dim \operatorname{im} Df(x)\leq m\:\forall x\in U\rbrace,$$ where $\mathcal{C}^k(U,\mathbb{R}^n)$ mean $k-$times continuously differentiable mappings from U to $\mathbb{R}^n$. Is it true that $$\mathcal{C}^\infty_{\leq m}(U,\mathbb{R}^n)\overset{\text{dense}}{\subset}\mathcal{C}^1_{\leq m}(U,\mathbb{R}^n),$$ with the usual $\left(\mathcal{C}^1,d(\cdot,\cdot)_{\mathcal{C}^1}\right)$ distance $$d(f,g)_{\mathcal{C}^1}=\sup\limits_{x\in U}\left|f(x)-g(x)\right|+ \sup\limits_{x\in U}\left\|Df(x)-Dg(x)\right\|.$$

$|\cdot|$ is length of a vector from $\mathbb{R}^n$ and $\|\cdot\|$ is length of vector from $\mathbb{R}^{n^2}$. Link to mathSE question https://math.stackexchange.com/q/1876303/357336

$\endgroup$
  • 1
    $\begingroup$ You should link to your question on MSE. $\endgroup$ – Michael Albanese Aug 4 '16 at 14:35
  • $\begingroup$ It seems that the answer is ``No'' due to the Brouwer Fixed Point Theorem and Sard Theorem. $\endgroup$ – Taras Banakh Aug 12 '16 at 19:35
  • $\begingroup$ @TarasBanakh Thank you for the comment! Do you want to use BFPT to mapping with rank restriction and assume that domain U is a ball or apply this theorem to a ball in function space some transformation of mappings ? Would you please elaborate. $\endgroup$ – Polatucha Aug 16 '16 at 16:29
  • $\begingroup$ @Polatucha By Sard Theorem, any differentiable map $f:U\to R^n$ whose differential $Df$ has rank $<n$ at each point $x\in U$ has image $f(U)$ of zero Lebesgue measure. On the other hand, BFPT implies that any map $f:U\to R^n$ which is sufficiently close to the identity in $C^0$-topology has image $f(U)$ with non-empty interior. These two facts imply that maps with differentials of small rank cannot be dense in $C^\infty(U,R^n)$ even in the $C^0$-topology. $\endgroup$ – Taras Banakh Aug 18 '16 at 10:12
  • $\begingroup$ This is true, but I do not understand how not-density of small rank in whole $\mathcal{C}^\infty(U,\mathbb{R}^n)$ imply non-density for the subset $\mathcal{C}^1_{\leq m}(U,\mathbb{R}^n)$. As far as I understand identity mapping is far away from any map from the set I am interested in i.e. $$d_{\mathcal{C}^1}(\text{id},\mathcal{C}^1_{\leq m}(U,\mathbb{R}^n))>c$$ for some positive constant $c$ depending probably on the dimension $n$. $\endgroup$ – Polatucha Nov 15 '16 at 13:00
13
$\begingroup$

There is a counterexample.

Example. There is $f\in C^1(\mathbb{R}^5,\mathbb{R}^5)$ with ${\rm rank}\, Df\leq 3$ that cannot be approximated in the supremum norm by mappings $g\in C^2(\mathbb{R}^5,\mathbb{R}^5)$ satisfying ${\rm rank}\, Dg\leq 3$.

Example. There is $f\in C^1(\mathbb{R}^7,\mathbb{R}^7)$, ${\rm rank}\, Df\leq 4$, that cannot be approximated in the supremum norm by mappings $g\in C^3(\mathbb{R}^7,\mathbb{R}^7)$ satisfying ${\rm rank}\, Dg\leq 4$.

Both examples are special cases of the following result proved in [1]:

Theorem. Suppose that $m+1\leq k<2m-1$, $\ell\geq k+1$, $r\geq m+1$, and the homotopy group $\pi_k(S^m)$ is non-trivial. Then there is a map $f\in C^1(\mathbb{R}^\ell, \mathbb{R}^r)$ with ${\rm rank}\, Df\leq m $ in $\mathbb{R}^\ell$ that cannot be approximated by maps of class $ C^{k-m+1}$.

You can find infinitely many more examples.

[1] P. Goldstein, P. Hajłasz, $C^1$ mappings in $\mathbb{R}^5$ with derivative of rank at most $3$ cannot be uniformly approximated by $C^2$ mappings with derivative of rank at most $3$. . J. Math. Anal. Appl. 468 (2018), 1108–1114. (arXiv:1804.08289).

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.