Tell me more ×
Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. It's 100% free, no registration required.

Let $\{E_k\}^{\infty}_{k=1}$ be a countable disjoint collection of measurable sets. Prove that for any set A, $m^*(A \cap \cup^{\infty}_{k=1})$ = $\sum^{\infty}_{k=1} m^*(A\cap E_k)$

My answer:

I tried to prove this by induction:

We know that if k=1, $m^*(A \cap E) = m^*(A \cap E)$. So now if we assume that for some n, $m^*(A \cap \cup^{n}_{k=1})$ = $\sum^{n}_{k=1} m^*(A\cap E_k)$. Now we have to prove that it is true for n+1. $m^*(A \cap \cup^{n+1}_{k=1}) \leq m^*((A \cap \cup^{n}_{k=1} E_k) + m^*(A \cap E_{n+1})$

So now I have to prove the reverse inclusion, right?...But I'm kind of stuck here...

share|improve this question
1  
And you should also use that $E_k$ are measurable. – Berci Feb 11 at 15:27

2 Answers

up vote 2 down vote accepted

I assume $m^*$ is an outer measure; and will use the fact that a set $E$ is measurable if and only if
$$m^*(E)=m^*(E\cap T)+m^*(E\cap T')$$ for all sets $T$.

Since $\cup_{k=1}^\infty E_k$ is measurable, your result is equivalent to proving that $$\tag{1} m^*(A)=\sum_{k=1}^\infty m^*(E_k\cap A) +m^*\Bigr(A\cap \bigl(\ \bigcup_{k=1}^\infty E_k\ \Bigr)'\Bigr). $$

Towards that end, note, by the the countable subadditivity of $m^*$ $$\eqalign{ m^*(A) &\le m^*\Bigr(A\cap \Bigl(\ \bigcup_{k=1}^\infty E_k\ \Bigr) \Bigr) +m^*\Bigr(A\cap \Bigl(\ \bigcup_{k=1}^\infty E_k\ \Bigr)'\Bigr)\cr &\le\sum_{k=1}^\infty m^*(E_k\cap A) +m^*\Bigr(A\cap \Bigl(\ \bigcup_{k=1}^\infty E_k\ \Bigr)'\Bigr). } $$

With this in hand, we see that the desired result is trivial if $m^*(A)=\infty$.

Now assume $m^*(A)\ne\infty$. To prove the reverse inequality needed, we will prove by induction that for each $p\ge1$ and any set $A$ $$\tag{2} m^*(A)=\sum_{k=1}^p m^*(E_k\cap A) +m^*\Bigr(A\cap \Bigl(\ \bigcup_{k=1}^p E_k \Bigr)'\ \Bigr). $$ Once this is done, since the sequence $\Bigl(m^*\Bigr(A\cap \bigl(\cup_{k=1}^p E_k \bigr)'\Bigr)\Bigr)_{p=1}^\infty$ is nonincreasing and bounded below by $m^*\Bigr(A\cap \bigl(\cup_{k=1}^\infty E_k \bigr)' \Bigr)$, it will follow that $$ m^*(A)\ge\sum_{k=1}^\infty m^*(E_k\cap A) +m^*\Bigr(A\cap \Bigl(\ \bigcup_{k=1}^\infty E_k\ \Bigr)'\Bigr), $$ as desired.


So, on to the proof of $(2)$.

The base case follows from the fact that $E_1$ is measurable: $$ m^*(A)=m^*(A\cap E_1)+m^*(E\cap E_1'). $$

Assume $(2)$ is true for $p=k$.

Since $E_{p+1}$ is measurable, $$m^*(A)=m^*(A\cap E_{p+1})+ m^*(A\cap E_{p+1}')$$ Now, using $(2)$ applied to the set $A\cap E_{p+1}'$,

$$ \eqalign{ m^*(A)&=m^*(A\cap E_{p+1})+ m^*(A\cap E_{p+1}')\cr &=m^*(A\cap E_{p+1})+\sum_{k=1}^p m^*(A\cap E_{p+1}' \cap E_k) +m^*\Bigl( A\cap E_{p+1}'\cap \Bigl( \bigcup_{k=1}^p E_k\ \Bigr)'\Bigr)\cr &=m^*(A\cap E_{p+1})+\sum_{k=1}^p m^*(A\cap E_k) +m^*\Bigl( A\cap E_{p+1}'\cap \Bigl(\ \bigcup_{k=1}^p E_k\ \Bigr)'\Bigr)\cr &=\sum_{k=1}^{p+1} m^*(E_k\cap A) +m^*\Bigr(A\cap \Bigl(\ \bigcup_{k=1}^{p+1} E_k\ \Bigr)'\Bigr); } $$ where, in the third equality, we used the fact that ${E_k\subset E_{p+1}'}$. This is exactly what we need to finish the proof by induction. $(2)$ holds for all $p$.

share|improve this answer

I don't think you can use induction to prove a property for $\infty$.

Why not using that measures are completely additive and the equality $A\cap \bigcup_{k=1}^\infty E_k = \bigcup_{k=1}^\infty(A\cap E_k)$

share|improve this answer
Actually, this is what I thought of doing before I used induction... $m^*(A\cap \bigcup_{k=1}^\infty E_k) = m^*(\bigcup_{k=1}^\infty(A\cap E_k))$ and we know that, since each of these is countable and disjoint we can say that it is equal to $\sum^{\infty}_{k=1}m(A \cap E_k)$ But before I use this theorem, I need to show that $A \cap E_k$ is measurable for each k, right? But there is a theorem in my textbook that says that we cannot conclude that a subset of a measurable set is measurable. – Artus Feb 11 at 16:12
Since $A\cap E_k$ is a subset of $\{E_k\}^{\infty}_{k=1}$, we cannot say taht $A \cap E_k$ is measurable either, right? That's what confused me... – Artus Feb 11 at 16:13

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.