Rolling Right Along

Suppose we roll 3 3 fair, standard six-sided dice. Any dice that show a 6 6 are removed, after which the remaining dice, (if any), are rolled again. As before, any dice showing a 6 6 are removed, and any remaining dice are rolled again. We repeat this process until all 3 3 dice have been removed.

The expected number of rolls needed until all 3 3 dice have been removed is a b \dfrac{a}{b} , where a a and b b are positive coprime integers. Find a b . a - b.

Image Credit: Wikimedia Walter J. Pilsak


The answer is 9565.

This section requires Javascript.
You are seeing this because something didn't load right. We suggest you, (a) try refreshing the page, (b) enabling javascript if it is disabled on your browser and, finally, (c) loading the non-javascript version of this page . We're sorry about the hassle.

2 solutions

D G
Mar 31, 2015

The process is a Markov chain with transition matrix:

P = [ 125 216 25 72 5 72 1 216 0 25 36 5 18 1 36 0 0 5 6 1 6 0 0 0 1 ] P = \begin{bmatrix} \frac{125}{216} & \frac{25}{72} & \frac{5}{72} & \frac{1}{216} \\[0.3em] 0 & \frac{25}{36} & \frac{5}{18} & \frac{1}{36} \\[0.3em] 0 & 0 & \frac{5}{6} & \frac{1}{6} \\[0.3em] 0 & 0 & 0 & 1 \\[0.3em] \end{bmatrix}

Finding the fundamental matrix:

N = ( [ 1 0 0 0 1 0 0 0 1 ] [ 125 216 25 72 5 72 0 25 36 5 18 0 0 5 6 ] ) 1 = [ 216 91 2700 1001 5490 1001 0 36 11 60 11 0 0 6 ] N = \left( \begin{bmatrix} 1 & 0 & 0 \\[0.3em] 0 & 1 & 0 \\[0.3em] 0 & 0 & 1 \\[0.3em] \end{bmatrix} - \begin{bmatrix} \frac{125}{216} & \frac{25}{72} & \frac{5}{72} \\[0.3em] 0 & \frac{25}{36} & \frac{5}{18} \\[0.3em] 0 & 0 & \frac{5}{6} \\[0.3em] \end{bmatrix} \right)^{-1} = \begin{bmatrix} \frac{216}{91} & \frac{2700}{1001} & \frac{5490}{1001} \\[0.3em] 0 & \frac{36}{11} & \frac{60}{11} \\[0.3em] 0 & 0 & 6 \\[0.3em] \end{bmatrix}

Finally the expected number of steps to reach the absorbing state is:

[ 216 91 2700 1001 5490 1001 0 36 11 60 11 0 0 6 ] [ 1 1 1 ] = [ 10566 1001 96 11 6 ] \begin{bmatrix} \frac{216}{91} & \frac{2700}{1001} & \frac{5490}{1001} \\[0.3em] 0 & \frac{36}{11} & \frac{60}{11} \\[0.3em] 0 & 0 & 6 \\[0.3em] \end{bmatrix} \begin{bmatrix} 1 \\[0.3em] 1 \\[0.3em] 1 \\[0.3em] \end{bmatrix} = \begin{bmatrix} \frac{10566}{1001} \\[0.3em] \frac{96}{11} \\[0.3em] 6 \\[0.3em] \end{bmatrix}

and since we started from the first state our expected number of rolls is 10566 1001 \frac{10566}{1001} .

Let E ( n ) E(n) be the expected number of rolls of n n dice needed before each of the dice has shown a 6 6 and been removed according to the description in the text of the question.

It is well-known that E ( 1 ) = 6. E(1) = 6. Now for two dice, there is a probability of 1 36 \frac{1}{36} that both dice show 6 6 on the first roll, a probability of ( 5 6 ) 2 = 25 36 (\frac{5}{6})^{2} = \frac{25}{36} that neither do and a probability of 10 36 \frac{10}{36} that precisely one of the dice shows a 6. 6. Thus

E ( 2 ) = 1 36 1 + 10 36 ( 1 + E ( 1 ) ) + 25 36 ( 1 + E ( 2 ) ) E(2) = \dfrac{1}{36}*1 + \dfrac{10}{36}*(1 + E(1)) + \dfrac{25}{36}*(1 + E(2))

11 36 E ( 2 ) = 96 36 E ( 2 ) = 96 11 . \Longrightarrow \dfrac{11}{36}*E(2) = \dfrac{96}{36} \Longrightarrow E(2) = \dfrac{96}{11}.

Next, for 3 3 dice, there is a probability of 1 216 \frac{1}{216} that all 3 3 dice show 6 6 on the first throw, 3 ( 5 6 ) 2 1 6 = 75 216 3*(\frac{5}{6})^{2}*\frac{1}{6} = \frac{75}{216} that precisely one dice does, 3 ( 1 6 ) 2 5 6 = 15 216 3*(\frac{1}{6})^{2}*\frac{5}{6} = \frac{15}{216} that precisely 2 2 do and ( 5 6 ) 3 = 125 216 (\frac{5}{6})^{3} = \frac{125}{216} that none do. We then have that

E ( 3 ) = 1 216 1 + 75 216 ( E ( 2 ) + 1 ) + 15 216 ( E ( 1 ) + 1 ) + 125 216 ( E ( 3 ) + 1 ) E(3) = \dfrac{1}{216}*1 + \dfrac{75}{216}(E(2) + 1) + \dfrac{15}{216}(E(1) + 1) + \dfrac{125}{216}(E(3) + 1)

91 216 E ( 3 ) = 306 216 + ( 75 216 ) ( 96 11 ) E ( 3 ) = 10566 1001 . \Longrightarrow \dfrac{91}{216}E(3) = \dfrac{306}{216} + (\dfrac{75}{216})(\dfrac{96}{11}) \Longrightarrow E(3) = \dfrac{10566}{1001}.

Thus a b = 10566 1001 = 9565 . a - b = 10566 - 1001 = \boxed{9565}.

Solved using the same method. I wasn't much familiar with Expectations, read about it on the net and afterwards found it rather straight forward to reach the solution.

Pawan Kumar - 6 years, 2 months ago

Log in to reply

Yes, once you have the general concept it is pretty straightforward. Glad this question led you to learn more about Expectations. :)

Brian Charlesworth - 6 years, 2 months ago

Why are you permuting the dices aren't they identical??

Harshita Arya - 6 years, 2 months ago

Log in to reply

They are, but for the purpose of creating the sample space we consider them to be distinct. So, for example, with two dice the sample space consists of 6 6 = 36 6*6 = 36 outcomes, each having the same probability of occurring. 1 1 of these outcomes results in two 6 6 's, 10 10 result in precisely one 6 6 and 25 25 of which result in no 6 6 's. Have a look at this link for further explanation.

Brian Charlesworth - 6 years, 2 months ago

0 pending reports

×

Problem Loading...

Note Loading...

Set Loading...