UA MATH564 概率论 计算至少有一个发生的概率:容斥原理与庞加莱公式

    技术2022-07-10  137

    UA MATH564 概率论 计算至少有一个发生的概率:容斥原理与庞加莱公式

    事件的并的Poincare公式事件的交的Poincare公式

    上一讲介绍了 P ( ⋃ i = 1 n A i ) P(\bigcup_{i=1}^{n} A_i) P(i=1nAi)的上界(Gunias不等式)和下界(钟开莱-艾尔迪希不等式),这一讲介绍准确计算 P ( ⋃ i = 1 n A i ) P(\bigcup_{i=1}^{n} A_i) P(i=1nAi)的方法——Poincare公式。

    先考虑一个简单情况,当 n = 2 n=2 n=2时, A 1 ∪ A 2 A_1 \cup A_2 A1A2可以很容易写成不相交的三个子事件的并: A 1 ∪ A 2 = ( A 1 − A 2 ) ∪ ( A 1 ∩ A 2 ) ∪ ( A 2 − A 1 ) A_1 \cup A_2 = (A_1-A_2) \cup (A_1\cap A_2) \cup (A_2 - A_1) A1A2=(A1A2)(A1A2)(A2A1)

    根据概率的有限可加性, P ( A 1 ∪ A 2 ) = P ( A 1 − A 2 ) + P ( A 1 ∩ A 2 ) + P ( A 2 − A 1 ) = P ( A 1 ) + P ( A 2 ) − P ( A 1 ∩ A 2 ) P(A_1 \cup A_2) = P(A_1-A_2) + P(A_1\cap A_2) + P(A_2 - A_1) \\ = P(A_1) + P(A_2)- P(A_1\cap A_2) P(A1A2)=P(A1A2)+P(A1A2)+P(A2A1)=P(A1)+P(A2)P(A1A2)

    这个表达式类似两个集合的容斥原理。与容斥原理类似,我们可以尝试把这个结果推广到 n n n个事件,获得准确计算 P ( ⋃ i = 1 n A i ) P(\bigcup_{i=1}^{n} A_i) P(i=1nAi)的方法。

    事件的并的Poincare公式

    引入一个记号: S m = ∑ 1 ≤ i 1 < i 2 < ⋯ < i m ≤ n P ( A i 1 ∩ A i 2 ∩ ⋯ ∩ A i m ) S_m = \sum_{1 \le i_1 < i_2 < \cdots < i_m \le n} P(A_{i_1} \cap A_{i_2} \cap \cdots \cap A_{i_m}) Sm=1i1<i2<<imnP(Ai1Ai2Aim)

    下面给出Poincare公式的形式: P ( ⋃ i = 1 n A i ) = ∑ m = 1 n ( − 1 ) m + 1 S m P(\bigcup_{i=1}^{n} A_i) = \sum_{m=1}^n (-1)^{m+1}S_m P(i=1nAi)=m=1n(1)m+1Sm

    证明 其实早在Boole不等式的证明中,我们就已经窥得Poincare公式的端倪。在Boole不等式的证明中,我们构造了一个集列 { B i } i = 1 n \{B_i\}_{i=1}^n {Bi}i=1n,其中 B 1 = A 1 , B 2 = A 2 − A 1 , ⋯   , B n = A n − ⋃ i = 1 n − 1 A n B_1 = A_1,B_2 = A_2-A_1,\cdots,B_n = A_n - \bigcup_{i=1}^{n-1}A_n B1=A1,B2=A2A1,,Bn=Ani=1n1An,显然 B i ∩ B j = ϕ , ∀ i ≠ j B_i \cap B_j = \phi, \forall i \ne j BiBj=ϕ,i=j,并且 ⋃ i = 1 n B i = ⋃ i = 1 n A i \bigcup_{i=1}^n B_i = \bigcup_{i=1}^{n} A_i i=1nBi=i=1nAi

    现在我们利用这个结构做计算 P ( ⋃ i = 1 n A i ) = P ( ⋃ i = 1 n B i ) = P ( ⋃ i = 1 n − 1 B i ) + P ( B n ) = P ( ⋃ i = 1 n − 1 A i ) + P ( A n − ⋃ i = 1 n − 1 A n ) = P ( ⋃ i = 1 n − 1 A i ) + P ( A n ) − P ( A n ∩ ⋃ i = 1 n − 1 A i ) P(\bigcup_{i=1}^{n} A_i) = P(\bigcup_{i=1}^n B_i) = P(\bigcup_{i=1}^{n-1} B_i)+P(B_n) \\= P(\bigcup_{i=1}^{n-1} A_i) + P(A_n - \bigcup_{i=1}^{n-1}A_n) = P(\bigcup_{i=1}^{n-1} A_i) + P(A_n) - P(A_n \cap \bigcup_{i=1}^{n-1} A_i) P(i=1nAi)=P(i=1nBi)=P(i=1n1Bi)+P(Bn)=P(i=1n1Ai)+P(Ani=1n1An)=P(i=1n1Ai)+P(An)P(Ani=1n1Ai)

    根据这个递推关系, P ( ⋃ i = 1 n A i ) = ∑ i = 1 n P ( A i ) − ∑ j = 1 n − 1 P ( A j + 1 ∩ ⋃ i = 1 j A i ) P(\bigcup_{i=1}^{n} A_i) = \sum_{i=1}^n P(A_i) - \sum_{j=1}^{n-1} P(A_{j+1} \cap \bigcup_{i=1}^{j} A_i) P(i=1nAi)=i=1nP(Ai)j=1n1P(Aj+1i=1jAi)

    下面考虑 P ( A j + 1 ∩ ⋃ i = 1 j A i ) P(A_{j+1} \cap \bigcup_{i=1}^{j} A_i) P(Aj+1i=1jAi),同样依据上面那个递推关系: P ( A j + 1 ∩ ⋃ i = 1 j A i ) = P ( ⋃ i = 1 j A i ∩ A j + 1 ) = P ( A j ∩ A j + 1 ) − P ( A j ∩ A j + 1 ∩ ⋃ k = 1 j − 1 A k ) P(A_{j+1} \cap \bigcup_{i=1}^{j} A_i) = P(\bigcup_{i=1}^{j} A_i\cap A_{j+1}) = P(A_j \cap A_{j+1}) - P(A_{j}\cap A_{j+1} \cap \bigcup_{k=1}^{j-1} A_k) P(Aj+1i=1jAi)=P(i=1jAiAj+1)=P(AjAj+1)P(AjAj+1k=1j1Ak)

    因此 P ( ⋃ i = 1 n A i ) = ∑ i = 1 n P ( A i ) − ∑ 1 ≤ i < j ≤ n P ( A i ∩ A j ) + ∑ j = 1 n − 1 P ( A j ∩ A j + 1 ∩ ⋃ k = 1 j − 1 A k ) P(\bigcup_{i=1}^{n} A_i) = \sum_{i=1}^n P(A_i) - \sum_{1 \le i < j \le n}P(A_i \cap A_j) + \sum_{j=1}^{n-1} P(A_{j}\cap A_{j+1} \cap \bigcup_{k=1}^{j-1} A_k) P(i=1nAi)=i=1nP(Ai)1i<jnP(AiAj)+j=1n1P(AjAj+1k=1j1Ak)

    接下来就是对 P ( A j ∩ A j + 1 ∩ ⋃ k = 1 j − 1 A k ) P(A_{j}\cap A_{j+1} \cap \bigcup_{k=1}^{j-1} A_k) P(AjAj+1k=1j1Ak)用那个递推公式,把所有三三相交的概率分离出来,一直做下去做到 n n nn nn相交的概率即可!

    证毕

    事件的交的Poincare公式

    基于事件的并的Poincare公式与de Morgan律,我们可以推出事件的交的Poincare公式。 P ( ⋂ i = 1 n A i ) = 1 − P ( ⋃ i = 1 n A ˉ i ) = 1 − ∑ m = 1 n ( − 1 ) m + 1 ∑ 1 ≤ i 1 < i 2 < ⋯ < i m ≤ n P ( A ˉ i 1 ∩ A ˉ i 2 ∩ ⋯ ∩ A ˉ i m ) = 1 − ∑ m = 1 n ( − 1 ) m + 1 ∑ 1 ≤ i 1 < i 2 < ⋯ < i m ≤ n [ 1 − P ( A i 1 ∪ A i 2 ∪ ⋯ ∪ A i m ) ] = 1 − ∑ m = 1 n ( − 1 ) m + 1 C n m + ∑ m = 1 n ( − 1 ) m + 1 ∑ 1 ≤ i 1 < i 2 < ⋯ < i m ≤ n P ( A i 1 ∪ A i 2 ∪ ⋯ ∪ A i m ) = ∑ m = 1 n ( − 1 ) m + 1 ∑ 1 ≤ i 1 < i 2 < ⋯ < i m ≤ n P ( A i 1 ∪ A i 2 ∪ ⋯ ∪ A i m ) P(\bigcap_{i=1}^{n} A_i) = 1 -P(\bigcup_{i=1}^{n} \bar{A}_i) = 1-\sum_{m=1}^n (-1)^{m+1}\sum_{1 \le i_1 < i_2 < \cdots < i_m \le n} P(\bar{A}_{i_1} \cap \bar{A}_{i_2} \cap \cdots \cap \bar{A}_{i_m}) \\ = 1-\sum_{m=1}^n (-1)^{m+1}\sum_{1 \le i_1 < i_2 < \cdots < i_m \le n} [1-P(A_{i_1} \cup A_{i_2} \cup \cdots \cup A_{i_m})] \\ = 1-\sum_{m=1}^n (-1)^{m+1}C_n^m + \sum_{m=1}^n (-1)^{m+1}\sum_{1 \le i_1 < i_2 < \cdots < i_m \le n}P(A_{i_1} \cup A_{i_2} \cup \cdots \cup A_{i_m}) \\ = \sum_{m=1}^n (-1)^{m+1}\sum_{1 \le i_1 < i_2 < \cdots < i_m \le n}P(A_{i_1} \cup A_{i_2} \cup \cdots \cup A_{i_m}) P(i=1nAi)=1P(i=1nAˉi)=1m=1n(1)m+11i1<i2<<imnP(Aˉi1Aˉi2Aˉim)=1m=1n(1)m+11i1<i2<<imn[1P(Ai1Ai2Aim)]=1m=1n(1)m+1Cnm+m=1n(1)m+11i1<i2<<imnP(Ai1Ai2Aim)=m=1n(1)m+11i1<i2<<imnP(Ai1Ai2Aim)

    定义记号 S ~ m = ∑ 1 ≤ i 1 < i 2 < ⋯ < i m ≤ n P ( A i 1 ∪ A i 2 ∪ ⋯ ∪ A i m ) \tilde{S}_m = \sum_{1 \le i_1 < i_2 < \cdots < i_m \le n}P(A_{i_1} \cup A_{i_2} \cup \cdots \cup A_{i_m}) S~m=1i1<i2<<imnP(Ai1Ai2Aim)

    事件的交的Poincare公式可以写成 P ( ⋂ i = 1 n A i ) = ∑ m = 1 n ( − 1 ) m + 1 S ~ m P(\bigcap_{i=1}^{n} A_i) = \sum_{m=1}^n (-1)^{m+1}\tilde{S}_m P(i=1nAi)=m=1n(1)m+1S~m

    Processed: 0.012, SQL: 9