Let $\{ E_n \}_{n=1}^{\infty} $be a collection of countable sets and let

$$ F_k = E_k \setminus ( \bigcup_{j=1}^{k-1} E_j ) $$

Then $F_k$ are pairwise disjoint and $\bigcup^{\infty} F_k = \bigcup^{\infty} E_k $



$$ F_k = E_k \cap ( \bigcup_{j=1}^{k-1} E_j )^c $$

say $l \neq k$ we show $F_k \cap F_l = \varnothing $. WLOG we can assume $l < k $ and write

$$ F_k \cap F_l = E_k \cap ( \bigcup_{j=1}^{k-1} E_j )^c \cap E_l \cap ( \bigcup_{j=1}^{l-1} E_j )^c$$

By Demorgans we get all intersections:

$$ F_k \cap F_l = E_k \cap ( E_1^c\cap ... \cap E_{k-1}^c) \cap E_l \cap ( E_1^c \cap ... \cap E_{l-1}^c)$$

Since $l < k$, then $E_k^c $ must be somewhere in $( E_1^c \cap ... \cap E_{l-1}^c)$ and since $E_k \cap E_k^c = \varnothing $ we have

$$ F_k \cap F_l = \varnothing $$ as desired.

As for the other part of the problem, If $x \in \bigcup^{\infty} F_k$, then $x \in F_{k_0}$ for some $k_0$. But $F_{k_0} \subset E_{k_0} \subset \bigcup^{\infty} E_k $. So,

$$ \bigcup F_k \subset \bigcup E_k $$

Next, notice

$$ F_k = E_k \cap ( \bigcup_{j=1}^{k-1} E_j )^c \implies E_k = F_k \cap ( \bigcup_{j=1}^{k-1} E_k )$$

and so by the same argument as above we get the equality of the unions.

Is this a correct approach? or is it unnecessary long? I know it is a trivial result, but I would like to get feedback. Thanks

  • $\begingroup$ Related. $\endgroup$ – Git Gud Nov 25 '14 at 8:16
  • $\begingroup$ I think the first part would be shorter with a proof by contradiction. $\endgroup$ – H_B Nov 25 '14 at 8:23
  • $\begingroup$ Suppose there exists $x\in F_k, F_l$ wlog $k<l$ then $x\in F_k\subseteq ( \bigcup_{j=1}^{l-1} E_j )$ contradicting $x\in F_l$ $\endgroup$ – H_B Nov 25 '14 at 8:26

Your answer to the "disjoint" part is fine. You could perhaps make it just slightly quicker:

Assume $l<k$ and $x\in F_l$. Then $x\in E_l\subseteq\bigcup_{j=1}^{k-1}E_j$, so $x\notin F_k$. Hence $F_l\cap F_k=\emptyset$.

To prove the unions are equal, you would do exactly as you have done to show $\bigcup F_j\subseteq \bigcup E_j$, but unfortunately it does not work the same way for the reverse inclusion (indeed if your argument were correct, we would have $E_k=F_k$ for all $k$). To get you started:

Let $x\in\bigcup_{k=1}^\infty E_k$. Then $\{k\in\mathbb{N}\,|\,x\in E_k\}$ is a non-empty subset of $\mathbb{N}$, so it has a minimal element, say $m_0$. See if you can finish it off from here.


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.