3
$\begingroup$

I know how to calculate the probability of throwing the same number in every throw of a dice (die, if you prefer!), and I know how to calculate the probability of throwing a different number in each throw of a dice. But how do I calculate the probability of throwing n different numbers in m throws of a dice?

Just to be clear, it doesn't matter what the numbers are or how many times each number is thrown – so two ones and three twos in five throws will be equivalent to four threes and one four in the same number of throws, for example.

$\endgroup$
2
$\begingroup$

Let:

  • $T_{n, m}$ be the event "exactly $n$ different numbers in $m$ throws of a dice".
  • $A$ be the event "in the $m^{th}$ throw, a number that has been seen before appears".
  • $D$ be the number of sides on your dice.

We assume that the dice is fair. The objective is to find $P (T_{n, m })$.

Then by the law of total probability, we have:

\begin{align} P (T_{n, m}) =& P (T_{n, m}|A)P(A) + P (T_{n, m}|\overline{A})P(\overline{A})\\ =& P (T_{n, m -1}|A)P(A) + P (T_{n-1, m-1}|\overline{A})P(\overline{A}) \\ =& P (A| T_{n, m -1})P (T_{n, m -1}) +P(\overline{A}|T_{n-1, m -1})P (T_{n-1, m -1}) \\ =& \frac{n}{D}P (T_{n, m -1}) + \frac{D - (n - 1)}{D}P (T_{n-1, m -1}) \end{align}

Base case: $P (T_{n, n })=\frac{n!}{n^n} $ and $ n>m\Rightarrow P (T_{n, m }) =0 $.

$\endgroup$
  • 1
    $\begingroup$ I can't quite see this. The probability of throwing six different numbers in six throws of a six-sided dice comes out as 0.665, but I thought it was 0.015 (6/6*5/6*4/6*3/6*2/6*1/6). Perhaps I'm misunderstanding the notation. $\endgroup$ – Remster Feb 19 '17 at 23:43
  • $\begingroup$ @Remster Thanks, good point, the answer was incorrect. Another counter example can be found taking m=1. I have edited the answer. Maybe you can get a neat closed form by induction, I haven't tried. $\endgroup$ – Franck Dernoncourt Feb 20 '17 at 4:22
  • $\begingroup$ @Remster I am not a dice expert so you may want to double check :-) My only recent dice experience are job interviews (don't do a PhD in machine learning, just play dice if you want to get hired in some places…) $\endgroup$ – Franck Dernoncourt Feb 20 '17 at 15:15

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.