Complement of limsup E_n

4k Views Asked by At

Im reading Friedmans modern analysis, and im having some flaws in my logical thinking... the question is to prove that $\left(\limsup\limits_{n}E_n\right)^c = \liminf\limits_{n}E_n^c$. My first attempt was this:

$\left(\limsup\limits_{n}E_n\right)^c = \{x\in X \mid x \notin \limsup\limits_{n}E_n\} = \{x \in X \mid x \notin \{x\in X \mid x \in E_n \text{ for infinitely many n}\}\} = \{x \in X \mid x \in E_n^c \text{ for infinitely many n}\} = \limsup\limits_{n}E_n^c$, which is clearly wrong. The second attempt was:

$\left(\limsup\limits_{n}E_n\right)^c = \{x\in X \mid x \notin \limsup\limits_{n}E_n\} = \{x \in X \mid x \notin \{x\in X \mid x \in E_n \text{ for infinitely many n}\}\} = \{x \in X \mid x\in E_n \text{ for at most finitely many n} \} = \{x \in X \mid x\in E_n^c \text{ for all but finitely many n}\} = \liminf\limits_{n}E_n^c$. The second attempt gives the right answer, but im not sure its correct. Also, whats the main difference between these two "solutions"? I believe the hard part is to find "the complement of infinitely many"? If anyone could guide me some in my thinking it would be much appreciated.

2

There are 2 best solutions below

1
On BEST ANSWER

The second attempt is correct. The first attempt contains no mistakes, it is just that you can say more.

If $x\in E_n$ does not hold for infinitely many $n$, then you can infere that $x\in E_n^C$ for infinitely many $n$. But you can infere something even stronger too: $x\in E_n^C$ for all but finitely many $n$.

You have to make use of the stronger conclusion to prove what you want to prove.

1
On

Could one also use De Morgan's laws, i. e.

\begin{align} (\lim \sup E_k)^C &= \mathbb{R} \bigcap (C (\lim \sup E_k)) \\ & = \mathbb{R} \bigcap (\lim \bigcup (C \bigcup E_k)) \\ & = \mathbb{R} \bigcap (\lim \bigcup (\bigcap C E_k)) \\ & = \mathbb{R} \bigcap \lim \inf (C E_k) \\ &= \lim \inf (E_k^C) \end{align} since intersecting $\lim \inf (E_k^C)$ with $\mathbb{R}$ won't change

\begin{equation} \lim \inf (E_k^C) \end{equation} $ \forall x \in \lim \inf (E_k^C)$?