0

Refer to my previous question: Can every positive real number $\leqslant\frac{\pi^2}6$ be expressed in this form?

Let $$A = \left\{\sum_{j\in S}f(j) : S\subset\mathbb N\setminus\{0\} \right\}$$ where $f:\mathbb N\setminus\{0\}\to\mathbb R$ is a sequence such that $\sum_{j=1}^\infty f(j)<\infty$. It was shown in my previous question, where $f(j) = \frac1{j^2}$, that the set $A$ was the disjoint union of compact intervals. Is it ever the case where $A$ is connected?

The motivation is that $X_n\stackrel{\mathrm{i.i.d.}}\sim\mathrm{Ber}(p)$, $Y(n) = Xf(n)$ and $S = \sum_{n=0}^\infty Y_n<\infty$. I would like to see if it is possible to define $f(j)$ such that $S$ is a continuous random variable. My intuition is that this is not possible due to the nature of $X_1$ taking values in $\{0,1\}$ - there is no way to "smooth out" this distribution so that the limiting distribution $S$ is continuous. How can I make this argument rigorous?

Math1000
  • 36,983
  • 2
    Yes. If $f(j)=\frac1{2^j}$ this just gives all binary fractions, so the set is $[0,1]$. More generally, if $f$ is positive and the condition in the accepted answer is valid starting with the first term then it is $[0,s]$, where $s$ is the sum of the series, as follows from the proof given there. – Conifold Dec 20 '19 at 11:40
  • I was thinking of $f(j) = \frac1{2^j}$ but wasn't sure that it worked. Is this a continuous distribution on $[0,1]$? My guess is that it is the uniform distribution. – Math1000 Dec 20 '19 at 11:54
  • As the accepted answer says, it isn't uniform, or even continuous, except when $p=1/2$. – Conifold Dec 20 '19 at 11:58
  • So when $p=1/2$, it is uniformly distributed over $[0,1]$? I can see how $p\ne\frac12$ would skew the distribution, and I suppose make it discontinuous. – Math1000 Dec 20 '19 at 12:20
  • See Random Variables with Independent Binary Digits:"Let $X = . b_1b_2b_3 ...$ be a random variable with independent binary digits $b_n$ taking values $0$ or $1$ with probability $p_n$ and $q_n = 1 -p_n$. When does X have a density? A continuous density? A singular distribution? This note gives necessary and sufficient conditions for the distribution of $X$ to be: discrete: $\sum\min (p_n, q_n) < \infty$; singular: $\sum_m^\infty[\log (p_n/q_n)]^2=\infty$ for every $m$; absolutely continuous: $\sum_m^\infty[\log (p_n/q_n)]^2 < \infty$ for some $m$" – Conifold Dec 21 '19 at 06:23
  • @Conifold Thatnks for the reference. – Math1000 Dec 21 '19 at 19:34

0 Answers0