For each of the following metric spaces $(M_i , d_i)$ and subsets $V_i ⊂ M_i$ decide whether $V_i$ is open with respect to the metric topology

1) $M_1= \Bbb R^2,d_1((a,b),(x,y))=|a-x|+|b-y|,V_1=\{(x,y)\in\Bbb R^2|xy>1\}$

2)$M_2=\Bbb R^\Bbb N, d_2(a,b)=$$\sum_{n=1}^{\infty} \frac{2^{-n} |a_n-b_n|}{1+|a_n-b_n|},V_2=\{a \in M_2| \{n \in N \mid a_n \neq 0\} \text{ finite}\}$

Here is my attempt (by the way this is not a homework question its from a past exam I'm using to help me study)

1)if $V$ is in open in $M \implies \exists B_e(a,b) \subset V $ so choosing an arbitrary $(a,b) \in V$ we choose (2,1) as this satisfies $ab>1$ now we have to see if there is an open ball from this point to any other point in $V$. i.e $d((a,b),(x,y))<e$

$d((2,1),(x,y))=|2-x|+|1-y|$ so we need to see is there a finite epsilon s.t. $d<e$ and $x,y$ satisfy $xy<1$

$|2-x|+|1-y|<e$ $\Rightarrow$$\sqrt{(2-x)^2}+\sqrt{(1-y)^2}<e \Rightarrow (2-x)^2+(1-y)^2<e^2=e$

$4-4a+a^2+1-b+b^2<e \Rightarrow a^2-4a-b+b^2<e \Rightarrow a^2(1-\tfrac{4}{a})+b^2(1-\tfrac{1}{b})<e$

and then as we increased the distance between these two points to the furthest apart they can be this implies that $a^2+b^2<e$ and if we suppose these points are still in V this implies that $a^2>\tfrac{1}{b^2}$ and so $a^2+\tfrac{1}{a^2}<a^2+b^2<e \Rightarrow a^2<e \Rightarrow a<e$ and so we have found that epsilon is greater than any real number and so there is no open ball and hence V is not open in M

2) let $a_n=1 \forall n \in \Bbb N $ $d(a,b)=$$\sum_{n=1}^{\infty} \frac{2^{-n} |1-b_n|}{1+|1-b_n|}$ the sum converges to .5 $\Rightarrow$ e>.5 so there is an open ball of radius .5 and so the set is open.

Is this close to right or completely wrong ?, I don't expect it to be totally correct as I'm not very good at this topic yet so any help would be much appreciated.


For 1)
Suppose $X=\mathbb{R}^2$ and $d(x,y)=|x_1-y_1|+|x_2-y_2|$ and let $V:=\{(x,y)\in X: xy>1\}$.
Let $x=(x_1,x_2)\in V$. Then $\exists\varepsilon>0$ such that $x_1 x_2-1>\varepsilon$. Then $y_1 y_2>1$ for all $y=(y_1,y_2)\in X$ with $d(x,y)<\frac{\varepsilon}{|x_1|+|x_2|}$, since then $$y_1 y_2\geq\left(|x_1|-\frac{\varepsilon}{|x_1|+|x_2|}\right)\left(|x_2|-\frac{\varepsilon}{|x_1|+|x_2|}\right)=|x_1x_2|-\varepsilon+\frac{\varepsilon^2}{(|x_1|+|x_2|)^2}>1$$ Then $V$ is open.
Your mistake was that you did not show it for arbirary $\varepsilon>0$. It is not sufficient to find one $\varepsilon$ such that $y_1y_2<1$ with $d(x,y)<\varepsilon$!
For a contradiction you have to prove this for all $\varepsilon>0$!
I might post something about 2) later but I am a bit in a hurry right now.

edit again: For 2): you cannot just consider $1$ sequence to show that $V$ is open, especially since $a\notin V$.
Let $X=\mathbb{R}^\mathbb{N}$ and $d(x,y)=\sum_{n=1}^\infty\frac{2^{-n}|x_n-y_n|}{1+|x_n-y_n|}$ and $V:=\{x\in X: \#\{x_n\neq 0\}<\infty\}$.
Let $x\in V$ and $I:=\{i\in\mathbb{N}:x_n\neq 0\}$ and suppose $V$ is open. Then $\exists\varepsilon>0$ such that $B_\varepsilon(x)\subseteq V$ (the open ball with radius $\varepsilon$ w.r.t. $d$).
So let $y\in B_\varepsilon(x)$. Then $\exists\delta>0$ such that $\varepsilon-d(x,y)>\delta$.
Then we can find $N\in\mathbb{N}$ such that $\sum_{n=N}^\infty 2^{-n}<\delta$. Then the sequence $(z_n)$ with $z_n=y_n$ for $n<N$ and $z_n=1$ if $n\geq N$ and $y_n=0$. Therefore $V$ is not open.

I hope this helps.

  • $\begingroup$ Thank you very much , your post was very informative , however I don't see how you got the inequality$ y_1 y_2\geq\left(|x_1|-\frac{\varepsilon}{|x_1|+|x_2|}\right)\left(|x_2|-\frac{\varepsilon}{|x_1|+|x_2|}\right)$ ? $\endgroup$ – excalibirr Jan 9 '18 at 17:52
  • $\begingroup$ sorry my mistake It was supposed to be dependant on $M_2$ I've edited the question to reflect that now :) $\endgroup$ – excalibirr Jan 9 '18 at 17:54
  • 1
    $\begingroup$ First we see that $x_1x_2>0$ for all $(x_1,x_2)\in V$. First suppose $x_1,x_2>0$. Let $\delta:=\frac{\varepsilon}{x_1+x_2}$. If we suppose that $d(x,y)<\delta$ then $|x_1-y_1|<\delta$ and also $|x_2-y_2|<\delta$. Then the inequality follows. If $x_1,x_2<0$ then we have a similar argument, since then $y_1,y_2<0$, because always at least $|x_1|>1$ or $|x_2|>1$. feel free to ask further if something is not clear. $\endgroup$ – Pink Panther Jan 9 '18 at 18:02
  • $\begingroup$ It took me a while but I see how the inequality was obtained now ( my brain is slow from 10 hours straight study haha). when you decided to choose $d(x,y) < \tfrac{e}{|x_1|+|x_2|} $ did you know you would get this inequality or what thoughts went into choosing it ? $\endgroup$ – excalibirr Jan 9 '18 at 18:23
  • 1
    $\begingroup$ Well I pretty much started with guessing that bound for $d(x,y)$. I first i tried with $\frac{\varepsilon}{4}$, which did not work apparently. Then i started to look at that product in the middle term and looked for a bound so that $|x_1|+|x_2|$ cancels with the bound so that the whole term becomes greater than $1$. $\endgroup$ – Pink Panther Jan 9 '18 at 18:28

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.