返回题库

掷 5 面骰直到和至少为 5 的期望次数

Sum Exceedance IV

专题
Probability / 概率
难度
L6

题目详情

反复掷一枚公平 5 面骰(点数为 1..5),直到累计点数和至少为 5。问:期望需要掷多少次?

A fair 55-sided die with the values 151-5 on the sides is rolled repeatedly until the sum of all upfaces is at least 55. Find the expected number of times the die needs to be rolled.

解析

结果为

(1+15)4=6454=12966252.0736.\left(1+\frac{1}{5}\right)^{4}=\frac{6^{4}}{5^{4}}=\frac{1296}{625}\approx 2.0736.

Original Explanation

We are going to solve the generalized version for nn-sided die with a sum of at least nn. Let's denote the expected number of rolls needed to obtain a sum of at least nn starting from a sum of kk by EkE_k. Clearly we have that En=0E_n = 0, as we already have a sum of nn. Further, we have that En1=1E_{n-1} = 1, as no matter what is obtained, we have a sum of at least nn.

Then, we have that En2=1+1nEn1=1+1nE_{n-2} = 1 + \dfrac{1}{n}E_{n-1} = 1 + \dfrac{1}{n}, as with probability 1n\dfrac{1}{n}, we obtain the value 11 and we have a total sum of n1n-1. Similarly, En3=1+1nEn1+1nEn2=1+2n+1n2E_{n-3} = 1+ \dfrac{1}{n} E_{n-1} + \dfrac{1}{n}E_{n-2} = 1 + \dfrac{2}{n} + \dfrac{1}{n^2}.

By continuing with this pattern, one can prove by induction that Enk=j=0k1(k1j)njE_{n-k} = \displaystyle \sum_{j=0}^{k-1} \dfrac{\binom{k-1}{j}}{n^j}.

Therefore, by the Binomial Theorem, E0=Enn=j=0n1(n1j)(1n)j=(1+1n)n1E_0 = E_{n-n} = \displaystyle \sum_{j=0}^{n-1} \binom{n-1}{j}\left(\dfrac{1}{n}\right)^j = \left(1 + \dfrac{1}{n}\right)^{n-1}.

We applied the Binomial Theorem with x=1nx = \dfrac{1}{n} and y=1y = 1. Therefore, for n=5n = 5, our answer is 6454=1296625\dfrac{6^4}{5^4} = \dfrac{1296}{625}.