[Math] Exercise: signed measures, total variation.

measure-theoryreal-analysis

I have this exercise:

Let $\nu_1$ and $\nu_2$ be finite signed measures on
$(\Omega,\mathcal{A})$. Prove that:

$|\nu_1+\nu_2|\le|\nu_1|+|\nu_2|$;

, that is $|\nu_1+\nu_2|(A)\le|\nu_1|(A)+|\nu_2|(A)$ for each $A \in
\mathcal{A}$.

My problem is that I get an equality and not an inequality, so that leads me to believe I have done something wrong, because that is a stronger resulat than they ask for. Can you see where I did wrong, or should there be an equality?

Proof:

We have that $(\nu_1+\nu_2)(A)=\nu_1(A)+\nu_2(A)$. Then use the Jordan decomposition theorem on each part:

$=\nu_1^+(A)-\nu_1^-(A)+\nu_2^+(A)-\nu_2^-(A)=(\nu_1^+(A)+\nu_2^+(A))-(\nu_1^-(A)+\nu_2^-(A))$. This gives us the Jordan-decomposition of $\nu$ since it's Jordan-decomposition is unique.

Hence we have that the total variation is:

$|\nu_1+\nu_2|(A)=(\nu_1+\nu_2)^+(A)+(\nu_1+\nu_2)^-(A)=_\text{by what showed above}$

$(\nu_1^+(A)+\nu_2^+(A))+(\nu_1^-(A)+\nu_2^-(A))=\nu_1^+(A)+\nu_1^-(A)+\nu_2^+(A)+\nu_2^-(A)$

$=|\nu_1|(A)+|\nu_2|(A)$

Do you see an error?, I get equality, but it should be inequality. I doubt they would have written the exercise like that if in fact it should be equality.

Best Answer

Consider the following lemma:

If $\nu$ is a signed measure and $\lambda$, $\mu$ are positive measures such that $\nu=\lambda-\mu$, then $\lambda\geqslant \nu^+$ and $\mu\geqslant \nu^-$.

To see this, let $(P,N)$ be a Hahn decomposition for $\nu$, then for $E\subset P$, $$\nu(E) = \nu^+(E) = \lambda(E) - \mu(E)\implies \lambda(E) = \nu^+(E)+\mu(E)\geqslant \nu^+(E), $$ and similarly if $E\subset N$, $$\nu(E) = -\nu^-(E) = \lambda(E) - \mu(E)\implies \mu(E) = \nu^-(E) + \lambda(E)\geqslant \nu^-(E), $$ since $\lambda,\mu\geqslant0$.

Now, we have \begin{align} \nu_1+\nu_2 &= (\nu_1^+ - \nu_1^-) + (\nu_2^+-\nu_2^-)\\ &= (\nu_1^++\nu_2^+)-(\nu_1^-+\nu_2^-), \end{align} so $(\nu_1^++\nu_2^+)\geqslant (\nu_1+\nu_2)^+$ and $(\nu_1^-+\nu_2^-)\geqslant (\nu_1+\nu_2)^-$. It follows that \begin{align} |\nu_1+\nu_2| &= (\nu_1+\nu_2)^+ + (\nu_1+\nu_2)^- \\ &\leqslant (\nu_1^++\nu_2^+) + (\nu_1^-+\nu_2^-)\\ &= (\nu_1^++\nu_1^-) + (\nu_2^++\nu_2^-)\\ &= |\nu_1| + |\nu_2|. \end{align}

Related Question