# Upper limits of the sequence of unions of sets [closed]

How to prove $$\limsup (A_k\cup B_k)=\limsup A_k \cup \limsup B_k \tag{1}$$ and does this remain true if I replace $\cup$ with $\cap$?

Here $A_k$ and $B_k$ are sets. The upper and lower limits are $$\liminf X_k=\bigcup_n\bigcap_{k>n}X_k,\qquad \limsup X_k = \bigcap_n\bigcup_{k>n}X_k$$

### Remark

Property (1) is similar to $$\limsup \max(a_k,b_k) = \max(\limsup a_k, \limsup b_k)$$ for numerical sequences.

-

## closed as too localized by The Chaz 2.0, Lord_Farin, rschwieb, Davide Giraudo, MicahApr 25 '13 at 18:03

This question is unlikely to help any future visitors; it is only relevant to a small geographic area, a specific moment in time, or an extraordinarily narrow situation that is not generally applicable to the worldwide audience of the internet. For help making this question more broadly applicable, visit the help center.If this question can be reworded to fit the rules in the help center, please edit the question.

$\limsup$ is a number. What do you mean by $\limsup A_k \cup \limsup B_k$? –  Dennis Gulko Apr 25 '13 at 15:29
@DennisGulko: See this page for some information on the $\limsup$ of sets. –  Clayton Apr 25 '13 at 15:32
Hint: pick up an element of RHS, show it stay in LHS, and vise versa. –  Ma Ming Apr 25 '13 at 15:34
@noname1014 If it is the case, you should better explicitly state that the $A_k, B_k$ are sets of real numbers (and not real numbers). –  Hagen von Eitzen Apr 25 '13 at 15:38
I just learned of a second notion for $\liminf$ and $\limsup$ of sets. On the one hand, we have $\limsup X_k$ is the set of limit points of sequences $(x_k)$ with $x_k\in X_k$ and $\liminf$ is the set of limits of converging such sequences. It is however likely that the OP's definition is the case I already knew (corresponding to discrete topology): $\liminf X_k=\bigcup_n\bigcap_{k>n}X_k$, $\limsup X_k = \bigcap_n\bigcup_{k>n}X_k$. –  Hagen von Eitzen Apr 25 '13 at 15:46

If $x \in \limsup A_k \cup \limsup B_k$, then $x$ is an at least in one of the two sets, say $\limsup A_k$. Then, for every $k$, there exists an $n \geqslant k$ such that $x \in A_n$, but $A_n \subset A_n \cup B_n$. Thus $\forall k$, $\exists n \geqslant k$, such that $x \in A_k \cup B_k$. We just proved that $x \in \bigcap_{k \geqslant 1} \bigcup_{n \geqslant k} \left( A_n \cup B_n \right) = \limsup \left( A_k \cup B_k \right)$.

Let's try the contrapositive proof of the other direction. Assume $x \not\in \limsup A_k \cup \limsup B_k$, then $x \not\in \limsup A_k \wedge x \not\in \limsup B_k$, then $\exists k_1, \forall n_1 \geqslant k_1, x \not\in A_{n_1}$, $\exists k_2$, $\forall n_2 \geqslant k_2$, $x \not\in B_{n_2}$. Take $k$ to be the max of $k_1$ and $k_2$, then $\forall n \geqslant k$, $x \not\in A_n \wedge x \not\in B_n$, which implies $x \not\in A_n \cup B_n$, thus $x \not\in \limsup \left( A_k \cup B_k \right)$

-
why $\bigcap_{k \geqslant 1} \left[ \left( \bigcup_{n \geqslant k} A_n \right) \bigcup \left( \bigcup_{n \geqslant k} B_n \right) \right]\\ = \left( \bigcap_{k \geqslant 1} \bigcup_{n \geqslant k} A_n \right) \bigcup \left( \bigcap_{k \geqslant 1} \bigcup_{n \geqslant k} B_n \right)$ I cannot understand. –  noname1014 Apr 25 '13 at 16:20
First direction is right,but I cannot understand the other direction. –  noname1014 Apr 25 '13 at 16:25
you means $\forall k \exists n \geqslant k s.t. x \in A or x \in B$ we can conclude that $\forall k \exists n \geqslant k s.t. x \in A$ or $\forall k \exists n \geqslant k s.t. x \in B$.but I think it is not right ,because $\forall x(P(x) or Q(x))$we cannot conclude that$\forall x P(x)$ or $\forall x Q(x)$ . –  noname1014 Apr 25 '13 at 16:38
@noname1014 I tried a new proof by contrapositive (instead of proving $A$ implies $B$, I tried to prove not $B$, implies not $A$). Verify if my reasoning holds. –  Learner Apr 25 '13 at 17:01
I think it is right. –  noname1014 Apr 25 '13 at 17:31

Assume $x\in\limsup_k (A_k\cup B_k)$. Then there is a sequence $x_k\in A_k\cup B_k$ that has a subsequence $x_{n_k}$ converging to $x$. If $x_{n_k}\in A_{n_k}$ holds infinitely often, these infinite sub-subsequence also converges to $x$ and shows that $x\in\limsup_k A_k$. Otherwise, we must have $x_{n_k}\in B_{n_k}$ for all but finitely many $k$, hence $x_{n_k}\to x$ and $x\in\limsup_k B_k$. All in all, $x\in\limsup_k A_k\cup \limsup_k B_k$.

Conversely assume $x\in\limsup_k A_k\cup \limsup_k B_k$. Then $x$ is limit point of a sequence $(x_k)$ with $x_k\in A_k$ for all $k$, or of a sequence with $x_k\in B_k$ for all $k$. But then also $x_k\in A_k\cup B_k$ for all $k$, i.e. $x\in\limsup_k (A_k\cup B_k)$.

If you replace $\cup$ with $\cap$, the caim is no longer valid. For example, we can have $\limsup A_k=\limsup B_k\ne\emptyset$ although $A_k\cap B_k=\emptyset$ for all $k$.

-