1
$\begingroup$

A direct line from a point $p$ to a set $X$ is a line segment with one endpoint at $p$ and one endpoint in $X$, which is as short as any other line segment from $p$ to $X$. Given a closed set $X$ and open set $Z$ with $X \subset Z \subset \mathbb{R}^n$, must there exist an open set $Y$ in between $X$ and $Z$ which includes a direct line to $X$ from each point in $Y$?


My thoughts:

(1) My guess is the answer is yes, and in fact I think the following stronger statement is true: there is a $Y$ such that all direct lines to $X$ from each point in $Y$ are included in $Y$.

(2) If $X$ is compact, then we know the answer is yes because in this case there exists a $\tau > 0$ such that $Y := \{y : d(X, y) < \tau\} \subset Z$. (Of course, we define $d(X, y) = \min_{x \in X} d(y, x)$.)

Originally posted on MSE without any responses.

$\endgroup$
0

1 Answer 1

1
$\begingroup$

This is to confirm the stronger claim mentioned in (1).

Set $X_r=X\cap \overline{B_r(0)}$ and $X^r=X\setminus B_r(0)$ (both are closed, for any $t$). Define $$ f(t)=\min\left\{1,\sup\left\{r\colon U_r(X_{t})\subseteq Z\right\}\right\}, $$ where $U_r(A)$ is the (open) $r$-neighborhood of $A$.

Define $$ Y_t=U_{f(t)}(X_t)\cap\{y\colon d(y,X_t)<d(y,X^t). $$ Notice that $X\cap B_t(0)\subseteq Y_t$, the set $Y_t$ is open and is contained in $Z$. Hence the set $$ Y=\bigcup_{t>0} Y_t $$ is also open, is contained in $Z$ and contains $X$. It remains to check the principal condition.

Consider any $y\in Y$ (say, $y\in Y_t$), and let $x\in X$ be a closest to $y$ point in $X$. By definition of $Y_t$, we have $x\in X\setminus X^t$. Now, if $y'\in(x,y)$, then $y'\in U_{f(t)}(X_t)$. Moreover, $y'$ has a unique closest point in $X$ --- namely, $x$. Hence $y'\in Y_t\subseteq Y$.

$\endgroup$
3
  • $\begingroup$ @MattF. Thanks a lot! The answer really contained serious flaws. Now it is completely rewritten (down to the initial notation). Hopefully, everything is correct now. $\endgroup$ May 17, 2022 at 20:37
  • $\begingroup$ This looks promising but I am again confused about why the interior points of $(x,y)$ (like $y'$) are in $U_{f(t)}(X_t)$. $\endgroup$
    – user44143
    May 20, 2022 at 5:45
  • $\begingroup$ @MattF. But $y$ is in $Y_t$ which is included into $U_{f(t)}(X_t)$, and $y'$ is closer to $x$ (and hence to $X_t$) than $y$... $\endgroup$ May 20, 2022 at 5:52

Your Answer

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

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