Take the 2-minute tour ×
Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. It's 100% free, no registration required.

I'm working on a question from Topology by Munkres on p. 127 (Exercise #6). Here it is:

Let $\bar{\rho}$ be the uniform metric on $\mathbb{R}^\omega$. Given $\mathbb{x}= (x_1, x_2, ...) \in \mathbb{R}^\omega$ and given $0 < \epsilon < 1$, let $U(\mathbb{x}, \epsilon):= (x_1 - \epsilon, x_1 + \epsilon) \times \cdots \times (x_n - \epsilon, x_n + \epsilon) \times \cdots$.

The question is divided into three parts, of which I have been able to answer the first and last parts of the problem. For the second part, i.e. part (b), I think I'm on the right track but I'm not too sure about that. Part (b) asks to show that $U(\mathbb{x}, \epsilon)$ is not even open in the uniform topology.

In my attempt to do this part, I let $\mathbb{x}= \mathbb{0}= (0, 0, 0, 0, ...)$ and $\mathbb{y}= (0, 1/2, 2/3, 3/4, ...)$. Then I computed the distance $\displaystyle\bar{\rho}(\mathbb{x}, \mathbb{y})= \sup_{i \geq 0}(\min\{|x_i - y_i|, 1\})= 1$. So that means the point $\mathbb{y}$ would not be in the open ball of radius $1$, i.e. $B_\bar{\rho}(\mathbb{x}, 1)$. So wouldn't this mean that we can't fit an $\epsilon$-ball around some point $\mathbb{y}$ that stays in $U(\mathbb{x}, \epsilon)$, and hence, $U(\mathbb{x}, \epsilon)$ is not open in the uniform topology induced by $\bar{\rho}$? Any guidance or advice would be greatly appreciated. I'm just starting to learn topology, and still have a lot to learn! Interesting stuff, though.

share|improve this question
2  
The point $0$ is not the problem here, since the ball for the uniform norm centered at $0$ of radius $\varepsilon /2$ is contained in $U(0,\varepsilon)$. The idea is the following: consider the sequence $y=\{y_n\}$ with $y_n:=\frac n{n+1}\varepsilon$. Then $y\in U(0,\varepsilon)$. Now show that given a $r>0$, we can find a sequence $z$ which is in $B_{\bar{\rho}}(y,r)$ but not in $U(0,\varepsilon)$. –  Davide Giraudo Oct 10 '11 at 21:47
1  
An other thing: even if we can answer the question b) without using the previous one, can you write this one (and the last question)? I guess it's an interesting exercise. –  Davide Giraudo Oct 10 '11 at 21:58

1 Answer 1

up vote 3 down vote accepted

You have the right idea: it’s true that no open ball centred at your point $y$ is contained in $U(0,1)$. To finish the argument, you have to show this, and then you have to generalize the argument to all sets of the form $U(x,r)$.

Consider the ball $B_{\bar\rho}(y,r)$, where $r>0$; we want to show that it’s not a subset of $U(0,1)$ no matter how small $r$ is. Clearly $y$ is nudging the ‘top edge’ of $U(0,1)$, so we should try to pick something ‘bigger than’ $y$, but still ‘small’ enough to fit in $B_{\bar\rho}(y,r)$; since $$y=\left(\frac{n-1}{n}\right)_{n\in\mathbb{Z}^+}=\left(1-\frac1n\right)_{n\in\mathbb{Z}^+},$$ a natural idea is $$z=\left(\frac{n-1}{n}+\frac{r}2\right)_{n\in\mathbb{Z}^+} = \left(1-\frac1n+\frac{r}2\right)_{n\in\mathbb{Z}^+}.$$ Then $\bar\rho(y,z) = \sup\limits_{k\in\mathbb{Z}^+}\min\{|y_k-z_k|,1\}=\min\{r/2,1\}\le r/2$, so $z\in B_{\bar\rho}(y,r)$. But if $n\ge \dfrac2r$, then $d\frac1n \le \dfrac{r}2$, $z_n = 1 - \dfrac1n + \dfrac{r}2 \ge 1$, and hence $z\notin U(0,1)$.

This shows that $U(0,1)$ is not open, and the same idea generalizes very easily. To show that an arbitrary $U(x,\epsilon)$ is not open, just scale this example by a factor of $\epsilon$ and translate it from the origin to $x$. (This is what Davide was suggesting in his comment.) That is, instead of taking $$y=\left(\frac{n-1}{n}\right)_{n\in\mathbb{Z}^+},$$ let $$y=x+\left(\frac{n-1}{n}\epsilon\right)_{n\in\mathbb{Z}^+}=\left(x_n+\frac{n-1}{n}\epsilon\right)_{n\in\mathbb{Z}^+}.$$ Then $y\in U(x,\epsilon)$, but no ball centred at $y$ is a subset of $U(x,\epsilon)$: the argument used above works with only very minor modifications, which I’ll leave to you.

share|improve this answer

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

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