I am reading Baby Rudin chapter 2 and came across some question on Theorem 2.41. When Rudin tries to prove that every infinite subset of $E$ has a limit points in $E$ implies $E$ is closed, he first supposes $E$ is not closed. Then there exists $x_0$ in $\mathbb{R}^k$ which is a limit point of $E$ but not a point in $E$. Then he constructs the set $S$. But later, he introduces $y$ and applies triangular inequality. I do not understand what $y$ is, and also, how he gets the last inequality of greater than or equal to $\frac12|x_0-y|$?

For reference, here is the proof given by Rudin:

2.41 $\ \ $ Theorem $\ \ $ If a set $E$ in ${\bf R}^k$ has one of the following three properties, then it has the other two:

$\quad(a)\ \ $ $E$ is closed and bounded.
$\quad(b)\ \ $ $E$ is compact.
$\quad(c)\ \ $ Every infinite subset of $E$ has a limit point in $E$.

Proof $\ \ $ If $(a)$ holds, then $E\subset I$ for some $k$-cell $I$, and $(b)$ follows from Theorems $2.40$ and $2.35$. Theorem $2.37$ shows that $(b)$ implies $(c)$. It remains to be shown that $(c)$ implies $(a)$.
$\qquad$ If $E$ is not bounded, then $E$ contains points ${\bf x}_n$ with $$|{\bf x}_n|>n\qquad(n=1,2,3,...).$$ The set $S$ consisting of three points ${\bf x}_n$ is infinite and clearly has no limit point in ${\bf R}^k$, hence has none in $E$. Thus $(c)$ implies that $E$ is bounded.
$\qquad$ If $E$ is not closed, then there is a point ${\bf x}_0\in {\bf R}^k$ which is a limit point of $E$ but not a point of $E$. For $n=1,2,3,...,$ there are points ${\bf x}_n\in E$ such that $|{\bf x}_n-{\bf x}_0|<1/n$. Let $S$ be the set of these points ${\bf x}_n$ Then $S$ is infinite (otherwise $|{\bf x}_n-{\bf x}_0|$ would have a constant positive value, for infinitely many $n$), $S$ has ${\bf x}_0$ as a limit point, and $S$ has no other limit point in ${\bf R}^k$. For if $\mathbf y\in{\bf R}^k$, ${\bf y}\neq{\bf x}_0$, then $$\begin{align}|{\bf x}_n-{\bf y}| & \geq |{\bf x}_0-{\bf y}| - |{\bf x}_n-{\bf x}_0| \\ & \geq |{\bf x}_0-{\bf y}|-\dfrac1n\geq \dfrac12|{\bf x}_0-{\bf y}|\end{align}$$ for all but finitely many $n$; this shows that $\bf y$ is not a limit point of $S$ (Theorem $2.20$).
$\qquad$ Thus $S$ has no limit point in $E$; hence $E$ must be closed if $(c)$ holds.

$\qquad$ We should remark, at this point, that $(b)$ and $(c)$ are equivalent in any metric space (Exercise $26$) but that $(a)$ does not, in general, imply $(b)$ and $(c)$. Examples are furnished by Exercise $16$ and by the space ${\scr L}^2$, which is discussed in Chap. 11.

  • $\begingroup$ I think it should be like this "Is a set $E$ does not contain all it's limit point then it is contained in $\Bbb R^k - E$. Let $x$ be such limit point. But $B(x; r)$ be an open ball interval centered at $x$. $B(x;r)$ contains points both from $E$ and $\Bbb R^k - E$. Let $y$ be a point of $E$ such that $y \in B(x;r)$. Now if you take $\delta = \frac 1 2\min\{||y - x||, r - ||y - x||\}$, there is an open ball centered at $y$ with length $\delta$ that is entirely contained in $S$. So $y$ is an interior point. Hence every point of $E$ is an interior point." $\endgroup$ – Santosh Linkha Mar 26 '13 at 8:41

In the proof, Rudin shows that $S$ has a limit point $x_0$. However, at the start of the proof we find that $x_0 \in \mathbb{R}^k - E$ and as $S \subset E$ we have found an infinite subset of $E$ which doesn't have a limit point in $E$. Well, not quite yet. It might be the case that $S$ has another limit point $y$ which is not equal to $x_0$ and is in $E$. So we must first exclude this possibility to be able to conclude that we have found an infinite subset $S \subset E$ which has no limit point in $E$. The string of inequalities that follow in Rudin's proof is to show that $x_n$ can't be simultaneously getting close to $x_0$ and $y$. I suppose the hard part is the last part:

$$\vert x_0 - y \vert - \frac{1}{n} \geq \frac{1}{2} \vert x_0 - y \vert.$$

It is important to read what follows: "for all but finitely many $n$". If $n$ is large enough (and therefore, $\frac{1}{n}$ is very small) we would essentially have

$$\vert x_0 - y \vert \geq \frac{1}{2} \vert x_0 - y \vert.$$

Which should be very clear to why it is true. In fact, we find that for $n \geq \frac{2}{\vert x_0 - y \vert}$ (well defined as $x_0 \neq y$) the first inequality holds and for $n < \frac{2}{\vert x_0 - y \vert}$ (which are finitely many) it doesn't hold. So what exactly have we shown about $y$? We have shown that $\vert x_n - y \vert$ isn't getting smaller and smaller as it should be if $y$ were a limit point of $S$. So $S$ doesn't have any other limit point besides $x_0$ and as $x_0 \notin E$ we have constructed an infinite set $S \subset E$ which has no limit point in $E$. However, this is a contradiction as we assumed c) to be true to prove a). Therefore, $E$ must be closed.

Hope this helps!


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.