Take the 2-minute tour ×
Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. It's 100% free, no registration required.

Given two random numbers $X$ and $Y$ between 0 and 1 define $Z$ to be the closest integer to $X/Y$. I know what the ratio of two uniform random variables (that is I know its PDF and CDF). However, working with this has gotten me nowhere for finding the above PMF. I have drawn various pictures but I cannot seem to generalize the result for any $X$ and $Y$. Any help is much appreciated. This was given as an in-class exercise (to be done before the next period and then discussed in class) and I wanted to be prepared for the discussion during the next class period. I do believe there are a few different cases such as when $Y \geq 1/2$ and when $Y<1/2$. but I can never get a concise value for differing intervals.

share|improve this question

2 Answers 2

A trick is to reduce everything to domains included in $[0,1]\times[0,1]$. Then one can use repeatedly the fact that, for every $0\leqslant a\leqslant b\leqslant1$, the events $U(a,b)=[aX\leqslant Y\leqslant bX]$ and $V(a,b)=[aY\leqslant X\leqslant bY]$ have probability $E[(b-a)Y]=E[(b-a)X]=\frac12(b-a)$ (note the condition that $a$ and $b$ are both in $[0,1]$, otherwise the result fails). Thus:

  • The event $[\color{green}{\mathbf{K=0}}]=[\frac{X}Y\lt\frac12]=[X\lt\frac12Y]=V(0,\frac12)$ has probability $\frac12(\frac12-0)=\color{red}{\mathbf{\frac14}}$.
  • For every $\color{green}{\mathbf{k\geqslant2}}$, the event $[\color{green}{\mathbf{K=k}}]=[k-\frac12\lt\frac{X}Y\lt k+\frac12]$ is $[a_kX\lt Y\lt b_kX]=U(a_k,b_k)$ with $a_k=\frac1{k+\frac12}$ and $b_k=\frac1{k-\frac12}$ hence it has probability $\frac12(b_k-a_k)=\color{red}{\mathbf{\frac2{4k^2-1}}}$.
  • To compute the probability of $[\color{green}{\mathbf{K=1}}]$, either one sums the other probabilities and one completes the sum to $1$, or one notes that $[K=1]=[\frac12\lt\frac{X}Y\lt\frac32]$ is the disjoint union of the events $[\frac12Y\lt X\lt Y]=V(\frac12,1)$ and $[\frac23X\lt Y\lt X]=U(\frac23,1)$ hence the probability of the event $[K=1]$ is $\frac12(1-\frac12)+\frac12(1-\frac23)=\color{red}{\mathbf{\frac5{12}}}$.

Sanity check: $$ \frac14+\frac5{12}+\sum_{k=2}^\infty\frac2{4k^2-1}=1. $$

share|improve this answer

Hint: $P\left\{ K=k\right\} =P\left\{ \left(k-\frac{1}{2}\right)Y<X<\left(k+\frac{1}{2}\right)Y\right\} =\underset{S}{\int\int}dxdy$ for $S=\left\{ \left(x,y\right)\in\left(0,1\right)^{2}\mid\left(k-\frac{1}{2}\right)y<x<\left(k+\frac{1}{2}\right)y\right\} $

share|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.