Construct a sequence of measureable sets $E_1\supseteq E_2 \supseteq E_3 \supseteq \cdots$ such that $\mu(E_n)=\infty$ for each $n$ but ...

Question

Construct a sequence of measureable sets $E_1\supseteq E_2 \supseteq E_3 \supseteq \cdots$ such that $\mu(E_n)=\infty$ for each $n$ but $$\mu\left(\bigcap_{n=1}^\infty E_n\right)=0$$

Claim: Let \begin{equation*} \begin{aligned} E_1= & \left(\frac{1}{i},1\right]\cup \left(\frac{1}{i+1},2\right] \cup \cdots \\ E_2= & \left(\frac{1}{i},2\right]\cup \left(\frac{1}{i+1},3\right] \cup \cdots \\ \vdots & \vdots \\ E_n= & \left(\frac{1}{i},n\right]\cup \left(\frac{1}{i+1},n+1\right] \cup \cdots \\ \vdots & \vdots \\ \end{aligned} \end{equation*} where $i$ is an arbitrary positive integer.

I believe that this sequence of sets satisfies the conditions above, but I want to formally write it out.

Answer 1

This is way too complicated. Take $E_i = [i,\infty)$. Then, $\mu(E_i) = \infty$ and $E_j \subset E_i$ for any $j>i$, clearly.

$\bigcap_i E_i = \emptyset$ since if $x \in \bigcap_i E_i$, that means that $x\geq i$ for all $i \in \mathbb{N}$ (and there is no such $x$).

Thus, $\mu\left(\bigcap_i E_i\right) = 0$.

Answer 2

The question is actually whether the example given works. Notice that $(1/i,1]\subset \bigcap_{n=1}^\infty E_n$ and so it is doesn't work.

Answer 3

Take the measure defined by $\mu(X)=\text{Card}(X)$ and $E_n=]0,\frac{1}{n}]$.