If $A$ and $B$ are bounded subsets of $\mathbb{R}$, I need to show that, $\sup{A} = \inf{B} \iff \forall \epsilon > 0, \, \exists a \in A, \, b \in B \, : \, b - a < \epsilon$.
I have begun the proof in the "$\implies$" direction for now, but I am finding there are many cases. I ended up giving up a few times before I decided to ask here for help. Here is my proof so far.
$A$, $B$ bounded $\implies$ $\inf{A}$ and $\sup{B}$ exist. Also let $c = \sup{A} = \inf{B}$.
Without loss of generality, suppose $c - \inf{A} \leq \sup{B} - c$. That is, the "length" of $A$ is smaller than or equal to the "length" of $B$.
We will consider three cases for $\epsilon > 0$.
- $\epsilon / 3 < c - \inf{A}$
- $c - \inf{A} \leq \epsilon / 3 < \sup{B} - c$
- $\epsilon / 3 \geq \sup{B} - c$.
For case 1., let $a = c - \epsilon / 3 \in A$. Then let $b = c + \epsilon / 3 \in B$.
Then, $b - a = (2/3) \epsilon < \epsilon$.
For case 2., let $a \in A$ be fixed, and let $b = a + \epsilon / 3 \in B$.
Then, $b - a = \epsilon / 3 < \epsilon$.
I am unsure how to go about case 3.
As you can see, case 3 is the difficult one to think about. If anybody could help me thinking of an $a \in A$ and $b \in B$ for case 3 that satisfied $b - a < \epsilon$ it would be greatly appreciated. Or... If someone can think of a better/more elegant proof, feel free to share.
Thanks.
It's somewhat easier to follow the following approach:
Let $c=\sup(A)=\inf(B)$. Let $\varepsilon>0$.
Since $c=\sup(A)$, there is some $a\in A$ such that $a>c-\varepsilon/2$.
Since $c=\inf(B)$, there is some $b\in B$ such that $b<c+\varepsilon/2$.
Then, $b-a<\varepsilon$.
The other direction does not appear to be correct without additional assumptions, if $A=B=[0,1]$, the given conditions are trivially true since one can choose $a=b$, but the infimum and supremum are not equal. One way to correct the statement is to try to prove that:
$ \sup(A)\geq\inf(B)$ iff $\forall\varepsilon>0$, $\exists a\in A,b\in B$ such that $b-a<\varepsilon$.