Show that $\lim_{k\to\infty} P(A_k)=P(A)$ for $A_1\supseteq A_2\supseteq…$ and $A=\bigcap_{k=1}^\infty A_k$

probability

Show that $\lim_{k\to\infty} P(A_k)=P(A)$ for $A_1\supseteq A_2\supseteq…$ and $A=\bigcap_{k=1}^\infty A_k$

Using basic properties of probability:

  1. $P\left(\bigcup_{i=1}^\infty A_i \right)=\sum_{i=1}^\infty P(A_i)$ Where each $A_i$ is disjoint.

  2. $P(S)=1$

  3. $0 \le P(A)\le P(S)$

I also have $P(A\cup B)=P(A)+P(B)-P(A\cap B)$ derived from these which I think should be used here.

I believe the point I need here is a choice of disjoint sets $B_i$ so I can make use of the properties but whose intersection is still $\bigcap_{i=k}^\infty A_k$.

Best Answer

I denote the complement of a set $E$ by $E^c$.

Because $P(A_k)+P(A_k^c)=1$ and $P(A) + P(A^c) = 1$, it suffices to show that $\lim_{k \to \infty} P(A_k^c) = P(A^c)$.

Note that $A_1^c \subseteq A_2^c \subseteq A_3^c \subseteq \cdots$. Let $B_k = A_k^c \setminus A_{k-1}^c$. Then the $B_k$ are disjoint and $\bigcup_{k=1}^\infty B_k = \bigcup_{k=1}^\infty A_k^c$. Using axiom (1), you have $$P(A^c) = P(\bigcup_{k=1}^\infty A^c_k) = P(\bigcup_{k=1}^\infty B_k) = \sum_{k=1}^\infty P(B_k) = \sum_{k=1}^\infty (P(A_k^c) - P(A_{k-1}^c)) = \lim_{k \to \infty} P(A_k^c). $$