Variance of the random variable with Cantor distribution

155 Views Asked by At

I'm currently looking through Oliver Knill's book Probability Theory and Stochastic Processes and am trying to understand Knill's proof that the variance of a random variable $X$ with the Cantor distribution is 1/8 (Corollary 3.1.6., pp. 129-130).

Taking $X$ to be a r.v. with the Cantor distribution and $Y = 1_{[0,1/3]}$, we have by the law of total variance that

$V[X]=E[V[X|Y]]+V[E[X|Y]]$.

Knill provides an argument that $E[V[X|Y]]=\frac{1}{9}V[X]$, which is the part I don't understand. The specific part that throws me is his assertion that (with $Z=E[X|Y]$),

$V[X|Y]=E[X^2|Y]-E[X|Y]^2=E[X^2|Y]-Z$.

If anyone could provide some insight into how the last equality above works (i.e., how $E[X|Y]^2=Z$) or more generally an explanation of how $E[V[X|Y]]=\frac{1}{9}V[X]$, I'd really appreciate it.