3
$\begingroup$

I am a graduate student and this is not something related to my work but I was just wondering and did not find an answer on the Internet. I asked this on the other math site two weeks ago and no one responded. Question:

Let $g$ and $h: R \rightarrow R$, where $R$ is the set of all real numbers, be two increasing functions and $\theta_g$, $\theta_h$ be the associated Lebesgue--Stieltjes outer measures on $R$. We can also associate to $g+h$ the Lebesgue--Stieltjes outer measure $\theta_{g+h}$.

Is it possible to show that $\theta_g + \theta_h = \theta_{g+h}$?

Thank you so much

$\endgroup$
  • 1
    $\begingroup$ @ChristianRemling : Since the question is, not about Lebesgue--Stieltjes measures, but about the corresponding outer measures, it seems to me not altogether trivial. $\endgroup$ – Iosif Pinelis Aug 1 '16 at 20:25
4
$\begingroup$

A more general question is as follows. Let $\mu$ and $\nu$ be two measures on a measurable space $(S,\Sigma)$. Does it then always follow that $(\mu+\nu)^*=\mu^*+\nu^*$, where ${}^*$ denotes the corresponding outer measure?

The answer to this question is yes. Indeed, for any $E\subseteq S$, let $\mathcal{A}_E$ denote the set of all sequences $(A_j)$ of pairwise disjoint members of $\Sigma$ such that $\bigcup_j A_j\supseteq E$. Then \begin{equation} \mu^*(E)+\nu^*(E)=\inf_{(A_j)\in\mathcal{A}_E}\sum_j\mu(A_j) +\inf_{(A_j)\in\mathcal{A}_E}\sum_j\nu(A_j) \end{equation} \begin{equation} \le\inf_{(A_j)\in\mathcal{A}_E}\Big(\sum_j\mu(A_j)+\sum_j\nu(A_j)\Big) =\inf_{(A_j)\in\mathcal{A}_E}\sum_j(\mu+\nu)(A_j) = (\mu+\nu)^*(E). \end{equation}

To prove the reverse inequality, $(\mu+\nu)^*(E)\le\mu^*(E)+\nu^*(E)$, take any real numbers $a$ and $b$ (if they exist) such that $\mu^*(E)<a$ and $\nu^*(E)<b$. Then for some sequences $(A_i)$ and $(B_j)$ in $\mathcal{A}_E$ one has $\sum_i\mu(A_i)<a$ and $\sum_j\mu(B_j)<b$. Let $C_{i,j}:=A_i\cap B_j$. Somehow enumerating the pairs $(i,j)$, we may assume that the double-sequence $(C_{i,j})$ is in $\mathcal{A}_E$. So, \begin{equation} (\mu+\nu)^*(E)\le\sum_{i,j}(\mu+\nu)(C_{i,j}) =\sum_i\sum_j\mu(C_{i,j})+\sum_j\sum_i\nu(C_{i,j}) \end{equation} \begin{equation} =\sum_i\mu(A_i)+\sum_j\nu(B_j)<a+b, \end{equation} for any real $a$ and $b$ such that $\mu^*(E)<a$ and $\nu^*(E)<b$. Hence, $(\mu+\nu)^*(E)\le\mu^*(E)+\nu^*(E)$. QED

| cite | improve this answer | |
$\endgroup$

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.