返回题库

30 面骰一次 vs 20 面骰两次取最大(平局 Bob

30 Die Split II

专题
Probability / 概率
难度
L4

题目详情

Alice 有一枚公平 30 面骰,Bob 有一枚公平 20 面骰。

Alice 掷一次。Bob 可以掷两次并取两次中的最大值。

点数更大者获胜;若平局,则 Bob 获胜。

求 Alice 获胜的概率。

Alice and Bob have fair 3030-sided and 2020-sided dice, respectively. The goal is to obtain the largest possible value. Alice rolls her die one time. However, Bob can roll his die twice and keep the maximum of the two values. In the event of a tie, Bob is the winner. Find the probability Alice is the winner.

解析

记 Alice 的点数为 AA,Bob 两次为 B1,B2B_1,B_2,Bob 的有效点数为 max{B1,B2}\max\{B_1,B_2\}

A>20A>20A20A\le 20 分情况:

P(A>20)=13,P(A20)=23.P(A>20)=\frac{1}{3},\quad P(A\le 20)=\frac{2}{3}.
  • A>20A>20,则 Alice 必胜。
  • A20A\le 20,条件在 A=aA=a 时 Alice 获胜需要 max{B1,B2}<a\max\{B_1,B_2\}<a,即 B1<aB_1<aB2<aB_2<a,概率为 (a120)2\left(\frac{a-1}{20}\right)^2

并且在 A20A\le 20 条件下,AA 在 1..20 上均匀,所以

P(A20)=120a=120(a120)2=247800.P(\text{赢}\mid A\le 20)=\frac{1}{20}\sum_{a=1}^{20}\left(\frac{a-1}{20}\right)^2=\frac{247}{800}.

合并得

P()=13+23247800=6471200.P(\text{赢})=\frac{1}{3}+\frac{2}{3}\cdot\frac{247}{800}=\frac{647}{1200}.

Original Explanation

Let WW be the event Alice wins. Then we can use the Law of Total Probability to condition on whether or not Alice's roll is larger than 2020. Let AA be the value Alice rolls. Furthermore, let B1B_1 and B2B_2 be the values that Bob rolls on the first and second roll. We have that P[W]=P[WA>20]P[A>20]+P[WA20]P[A20]\mathbb{P}[W] = \mathbb{P}[W \mid A > 20] \mathbb{P}[A > 20] + \mathbb{P}[W \mid A \leq 20] \mathbb{P}[A \leq 20] It is clear to see that P[A>20]=13\mathbb{P}[A > 20] = \frac{1}{3} as this accounts for 1010 of 3030 values. Furthermore, P[WA>20]=1\mathbb{P}[W \mid A > 20] = 1 as this is strictly larger than any value Bob can roll.

For the other case, we compute P[WA20]=a=120P[max{B1,B2}<a]P[A=aA20]=a=120P[B1<a]2P[A=aA20]\mathbb{P}[W \mid A \leq 20] = \displaystyle \sum_{a=1}^{20} \mathbb{P}[\text{max}\{B_1,B_2\} < a] \mathbb{P}[A = a \mid A \leq 20] = \sum_{a=1}^{20} \mathbb{P}[B_1 < a]^2 \mathbb{P}[A = a \mid A \leq 20] This is the event of interest as we want Alice to have a strictly higher value than Bob. We simplify the maximum statement by noting that B1B_1 and B2B_2 are IID and that max{B1,B2}<a\text{max}\{B_1,B_2\} < a is equivalent to saying that B1<a,B2<aB_1 < a, B_2 < a.

We know P[A=aA20]=120\mathbb{P}[A = a \mid A \leq 20] = \frac{1}{20} as we know A20A \leq 20, so there are 2020 equally-likely values. Furthermore, we have that P[B1<a]=a120\mathbb{P}[B_1 < a] = \frac{a-1}{20} as we want B1B_1 to take some value between 11 and a1a - 1, inclusive of both. Therefore, P[WA20]=120a=120(a120)2=247800\mathbb{P}[W \mid A \leq 20] = \frac{1}{20} \sum_{a=1}^{20} \left(\frac{a-1}{20}\right)^2 = \frac{247}{800} Combining these, we have that P[W]=13+23247800=6471200\mathbb{P}[W] = \frac{1}{3} + \frac{2}{3} \cdot \frac{247}{800} = \frac{647}{1200}