Can we generate event with probabilities (a) $6.75\space p^2q$, (b) $20\space p^3q^2$, (c) $3.9\space pq$? If yes, then how? If no, then why?

Here, in a coin toss, let P(occurring head)$=p=1-q$

I can easily generate events for $3\space pq$(with three trials which is just $3\space p^2q+3\space pq^2$), $10\space p^3q^2$ (with five trials) and $3\space p^2q$ (with three trials), but could not find anything in these cases. Thanks for any help!


For (b), for $0\le k\le4$ choose $20\binom4k$ patterns with $3+k$ heads and $2+(4-k)$ tails, and thus each with probability $p^{3+k}q^{2+(4-k)}$. This is possible because $\binom9{3+k}\ge20\binom4k$ for all $k$. The sum of the probabilities is

$$ \sum_{k=0}^420\binom4kp^{3+k}q^{2+(4-k)}=20p^3q^2(p+q)^4=20p^3q^2\;. $$

You can't do the other two for general $p$. Consider $p$ the reciprocal of any odd prime. Then the denominator of any probability in any finite number of trials is always a power of that odd prime and can never contain a factor of $2$, whereas the denominators of the two probabilities in $(a)$ and $(c)$ do.

This of course assumes that you were implying that the problem should be solved with a finite number of trials. It's easy to achieve any desired probability with infinitely many trials.

  • $\begingroup$ but the first calculation gives $10p^3q^2$ $\endgroup$ – Stat_prob_001 Jul 16 '18 at 17:47
  • $\begingroup$ @Stat_prob_001: Sorry! I replaced it by a slightly more involved solution that works (unless I've made another mistake...). $\endgroup$ – joriki Jul 16 '18 at 19:41
  • $\begingroup$ @Stat_prob_001: I posted and answered a more general form of this question here: math.stackexchange.com/questions/2853884. $\endgroup$ – joriki Jul 16 '18 at 21:39

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.