Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

Consider the two coins (i.e., probability measures on the discrete set ${0,1}$) $C_{0.9}$ and $C_{0.99}$, where $C_{x}$ is the coin having probability of turning head equal to $x$.

Let $\mu_{0.9}$ and $\mu_{0.99}$ be the probability measures on $2^{\omega}$ (Markov chains) obtained by flipping infinitely often the two coin $C_{0.9}$ and $C_{0.99}$ respectively. So that, for instance, the probability assigned by $\mu_{0.9}$ to the (open) set of sequences starting with two zeros is $0.81$.

Question: I would like to calculate $D(\mu_{0.9}, \mu_{0.99})$ defined as:

$D(\mu_{0.9}, \mu_{0.99})$ = $\displaystyle \bigsqcup_{B} \{ |\mu_{0.9}(B) - \mu_{0.99}(B)| \ \ $ ; $B$ a Borel subset of $2^{\omega} \}$.

I don't know how to approach this problem. I have the feeling that the above supremum should be strictly less than $1$ and that the maximizing event should be the set of sequences starting with about 25 ones. The number 25 comes from the function $f : \mathbb{R}\rightarrow \mathbb{R}$ defined as $f(x)= {0.99}^x - {0.9}^x$ which attains its maximum around 25.

Question 2: More generally, is the function $D$ (mapping pairs of probability measures to the unit interval) as defined above well known?

Thank you in advance for any comment!

share|cite|improve this question
Regarding Question 2: Total variation distance. – martini Mar 27 '12 at 13:30
@martini: Thanks! – MM83 Mar 27 '12 at 14:09

Call $B$ the event that the asymptotic proportion of heads exists and is 0.9, then the law of large numbers shows that $\mu_{0.9}(B)=1$ and $\mu_{0.99}(B)=0$. Hence $D(\mu_{0.9},\mu_{0.99})=1$. One sees that in fact the measures $\mu_{0.9}$ and $\mu_{0.99}$ are mutually singular.

share|cite|improve this answer

Your Answer


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.