Problem 6.2 Easy Free 3 variations

One Total Before Another

You roll a pair of fair six-sided dice repeatedly. Find the probability that a sum of 55 appears before a sum of 22 appears.

Work it out, then compare with the answer.

SolutionShow the worked solution

Key idea. Rolls showing neither total decide nothing, so only the first roll that shows a 55 or a 22 matters.

Of the 3636 equally likely ordered outcomes, four give a sum of 55: (1,4),(2,3),(3,2),(4,1)(1,4),(2,3),(3,2),(4,1). Only (1,1)(1,1) gives a sum of 22. So on each roll

P(sum 5)=436,P(sum 2)=136,P(neither)=3136.\mathbb{P}(\text{sum }5)=\tfrac{4}{36},\qquad \mathbb{P}(\text{sum }2)=\tfrac{1}{36},\qquad \mathbb{P}(\text{neither})=\tfrac{31}{36}.

The rolls are independent. A sum of 55 comes first exactly when the first n−1n-1 rolls show neither total and roll nn shows a 55, for some n≥1n\ge 1. Summing over nn,

P(5 before 2)=∑n≥1(3136)n−1436=4/361−31/36=4/364/36+1/36=45.\mathbb{P}(5\text{ before }2)=\sum_{n\ge1}\Big(\tfrac{31}{36}\Big)^{n-1}\tfrac{4}{36} =\frac{4/36}{1-31/36}=\frac{4/36}{4/36+1/36}=\frac45 .

The same computation shows that the probability of neither total ever appearing is lim⁡n(31/36)n=0\lim_n (31/36)^n=0, so the race is decided with probability 11.

In general, for repeated independent trials, P(A before B)=P(A)/(P(A)+P(B))\mathbb{P}(A\text{ before }B)=\mathbb{P}(A)/(\mathbb{P}(A)+\mathbb{P}(B)).

45\boxed{\tfrac45}

Variations

The same idea with a twist. Each solution gives only the changed step.

Variation 6.2a

Dave and Mike each roll a fair six-sided die simultaneously, repeating the following procedure: if the two values match, Mike wins outright; if Mike’s value is strictly greater than Dave’s, Dave wins outright; if Dave’s value is strictly greater than Mike’s, both dice are re-rolled and the procedure repeats. Find the probability that Mike eventually wins.

Work it out, then compare with the answer.

Solution

One round of two dice has three possible results:

  • the values match, with probability 636=16\tfrac{6}{36}=\tfrac16, and Mike wins;
  • Mike’s value is higher, with probability 1536=512\tfrac{15}{36}=\tfrac5{12}, and Dave wins;
  • Dave’s value is higher, with the remaining probability 512\tfrac5{12}, and the dice are re-rolled.

Only the round that ends the game matters. Applying the rule from the main problem,

P(Mike wins)=1/61/6+5/12=2/127/12=27.\mathbb{P}(\text{Mike wins})=\frac{1/6}{1/6+5/12}=\frac{2/12}{7/12}=\frac27 .
27\boxed{\tfrac27}
Variation 6.2b

Bill and Bob simultaneously and repeatedly flip coins that each land heads with probability 1/31/3, independently of each other and of past flips. Find the probability that Bill’s first head occurs on a strictly earlier flip than Bob’s first head (so a round in which both flip heads for the first time simultaneously does not count as a win for Bill).

Work it out, then compare with the answer.

Solution

Treat each pair of simultaneous flips as one round. Each coin lands heads with probability 13\tfrac13, independently, so a round has four possible results:

  • Bill heads and Bob tails, with probability 13⋅23=29\tfrac13\cdot\tfrac23=\tfrac29, and Bill wins;
  • Bob heads and Bill tails, with probability 29\tfrac29, and Bill loses;
  • both heads, with probability 19\tfrac19, and Bill loses;
  • both tails, with probability 49\tfrac49, and play continues.

Only the first round that is not both tails decides the outcome. Hence

P(Bill first)=2/92/9+2/9+1/9=25.\mathbb{P}(\text{Bill first})=\frac{2/9}{2/9+2/9+1/9}=\frac25 .
25\boxed{\tfrac25}
Variation 6.2c

You repeatedly roll two fair dice until you get either a double six or exactly one six (one die showing 66, the other not); any other outcome is rerolled. If the game stops on a double six you receive $100; if it stops on exactly one six you lose $x. Find the largest value of xx for which this game has non-negative expected value.

Work it out, then compare with the answer.

Solution

On each roll:

  • a double six has probability 136\tfrac1{36};
  • exactly one six has probability 1036\tfrac{10}{36}, since the six can be on either die and the other die shows one of 55 values.

Every other outcome is rerolled. So the game ends on a double six with probability

1/3611/36=111,\frac{1/36}{11/36}=\frac1{11},

and on exactly one six with probability 1011\tfrac{10}{11}.

The expected value of the game is

10011−10x11.\frac{100}{11}-\frac{10x}{11}.

This is non-negative exactly when x≤10x\le 10.

x=10\boxed{x=10}