If $X,Y \subset \mathbb{R}$ are measure zero sets, how can I show that $X \times Y \subset \mathbb{R^2}$ is a measure zero set too?

My outline is the following:

Since $X,Y$ is a measure zero set, we can find open intervals $U_i$ and $V_j$ for $X$ and $Y$, respectively. Then, the union of the open intervals is a cover for $X$ and $Y$. That part I have been able to do. However, I need to show now that the sum of the measures of $(U_i \times V_j)$ can be written less than any arbitrary $\epsilon >0$. That is:

$$ \sum_{i,j}|(U_i \times V_j)| < \epsilon $$

Since by definition of $X,Y$ being a measure zero set, we have

$$ \sum_{i}|U_i| < \sqrt{\epsilon} \\ \text{and} \sum_{j}|V_j| < \sqrt{\epsilon} $$

It then makes logical sense that we want

$$ \sum_{i,j}|(U_i \times V_j)| < \sum_{i}|U_i| \cdot \sum_{j}|V_j| < \sqrt{\epsilon} \cdot \sqrt{\epsilon} < \epsilon $$

Does anyone know how I can get the relation:

$$ \sum_{i,j}|(U_i \times V_j)| < \sum_{i}|U_i| \cdot \sum_{j}|V_j| $$

by something simple like the triangle inequality without having to resort to heavy measure theory? Thanks!


$U_i \times V_i$ is an open rectangle. Use the formula for the measure of a rectangle, and apply the distributive law:

$$\sum_{i=1}^m\sum_{j=1}^na_ib_j = \left(\sum_{i=1}^ma_i\right)\left(\sum_{j=1}^jb_j\right)$$

to the resulting sum of products.

| cite | improve this answer | |

If $X\subseteq\mathbb{R}$ has measure zero, the product will have measure zero no matter how large $Y$ is. So morally, the proof should not depend on $Y$ having measure zero.

It clearly suffices to show that $X\times\mathbb{R}$ has measure zero. Since $$X\times\mathbb{R}=\bigcup_{n\in\mathbb{N}}\big(X\times [-n,n]\big),$$ and the countable union of measure zero sets is a set of measure zero, it suffices to show that $X\times [-n,n]$ has measure zero for all $n$.

So take any $\epsilon>0$. Since $X$ has measure zero, there is a sequence of open intervals $(O_k)$ such that $X\subseteq\bigcup_k O_k$ and the sum of the length of the intervals is less than $(2n+1)\epsilon$. But then the sum of the areas of the products of open intervals $O_k\times (-n-1/2,n+1/2)$ will be less than $\epsilon$ and $$X\times[-n,n]\subseteq\bigcup_kO_k\times (-n-1/2,n+1/2).$$

| cite | improve this answer | |

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.