Just had a bash at this question for my Intro to Maths Stats module...I got to the end with a probability density function rather than a probability mass function, namely $f_Y(y) = \lambda a e^{-\lambda a y}.$ Obviously I'm missing some subtleties with the floor function that makes the new r.v. into a probability mass function instead. Anyways, here it is...

Suppose $X$ is an $exponential(\lambda)$ r.v. given by 0 for $x$ < 0 and $\lambda e^{-\lambda x}$ for $x \geq 0$. Recall the function $\lfloor x\rfloor$ is defined as the largest integer $n \leq x.$

Let $Y$ be defined by $ Y = \lfloor \frac{X}{a} \rfloor$, where $a$ > 0. Find the probability mass function of $Y$ and hence deduce that $Y$ is a geometric r.v., stating its parameter.

Thanks in advance (should be a quick one!) Sam


Hint:* $Y$ is clearly discrete taking values in the set of non-negative integers, due to the the flooring. Then, for any integer $n\geq 0$ we have $$P(Y = n) = P\left(X\in [an, a(n+1))\right) $$ $$= \int_{an}^{a(n+1)} \lambda\mathrm e^{-\lambda x}dx= (1-p)^np$$ where $p = 1 - e^{-\lambda a} \in (0,1),$ as $\lambda>0$ and $a>0$.

  • $\begingroup$ This extra clarification complements Batman's work. I now have a great model solution. Thanks to you both! $\endgroup$ – user142671 May 22 '14 at 21:24
  • $\begingroup$ Ok. Please accept one of our answers (left hand side upper corner) so that this question appears as answered in the site's statistics. Good luck with your studies! $\endgroup$ – ir7 May 22 '14 at 21:34

$Y$ takes on value $k \in \{0,1,\ldots,\}$ if $k = \lfloor \frac{X}{a} \rfloor$. Thus, $P(Y=k) = P( \lfloor \frac{X}{a} \rfloor = k) = P(k \leq \frac{X}{a} < k+1) = P(ak \leq X < a(k+1)) = \int_{ak}^{a(k+1)} f(x) dx$ where $f(x)$ is the density of $X$.

  • $\begingroup$ Oh dear, quite obvious really then. Must be too late in the day... $\endgroup$ – user142671 May 22 '14 at 21:22

Your Answer

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

Not the answer you're looking for? Browse other questions tagged or ask your own question.