I found this definition for the total variation distance for measures over countable sets:

The total variation distance between two measures $\mu$, $\nu$ on a countable set $S$ is defined as: $||\mu - \nu|| \equiv 0.5\sum_{z\in S}|\mu(z) - \nu(z)| = \sup_{A \subset S} |\mu(A) - \nu(A)|$.

Here $z$ is every singleton of $S$ and $A$ an arbitrary set from $S$. The proof is straightforward, but uses this statement:

$|\mu(A) - \nu(A)| + |\mu(A^c) - \nu(A^c)| = 2|\mu(A) - \nu(A)|$

which I can prove if both $\mu$ and $\nu$ are probability measures. If, for example, $\mu(S) \neq \nu(S)$ or if both of these measures is not finite, I cannot see the proof holding.

Given that the statement was found on a probability book, I believe this proof is supposed to hold for probability measures. But, as I quoted, it is stated for measures in general. This might just be poor phrasing in the book, so I am asking if this proof holds in general for an arbitrary measure (not necesarily a probability one).

Thank you very much in advance.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.