Using only the axioms of probability, derive the following result using induction:

\begin{align} \Pr\left(\bigcup_{i=1}^\infty A_i\right) = &\sum_{i=1}^\infty P(A_i) - \sum_{i<j}^\infty \Pr(A_i \cap A_j) - \\ &\sum_{i <j<k}^\infty \Pr(A_i \cap A_j \cap A_k) - \cdots - \\ &(-1)^{n+1}\Pr(A_1 \cap A_2 \cap \cdots \cap A_n) \end{align}

I have began the proof by showing that it holds for $n = 2$, and have assumed it holds for $n \le k$ for some arbitrary $k$. Now it just comes to showing that it holds for $k + 1$:

\begin{align} \Pr\left(\bigcup_{i=1}^{k+1} A_i\right) &= \Pr\left(A_{k+1} \cup \bigcup_{i=1}^kA_i\right) \\ &= \Pr(A_{k+1}) + \Pr\left(\bigcup_{i=1}^k A_i\right) - \Pr\left(A_{k+1} \cap \bigcup_{i=1}^kA_i\right) \end{align}

where the last term can be written as:

$$ \Pr\left(\bigcup_{i=1}^k(A_i \cap A_{k+1})\right) $$

This is where I am stuck and cannot simplify the expression to get the desired result. I am positive I am on the right track with the "proof", but do not know what result or axiom to use to continue

Any hints to how I can continue will be appreciated.

  • $\begingroup$ Remember that you want to prove it for $k+1$ assuming the $k$ case is true. Assuming the $k$ case is true, you can expand out $P(\bigcup_{i=1}^k X_i)$ for any collection of $k$ sets $X_1, X_2, \ldots, X_k$. The second and third terms on the RHS of your second equation are unions of $k$ sets, so you can now expand them. $\endgroup$
    – Paul
    Sep 27 '18 at 21:26
  • $\begingroup$ Please add the [self-study] tag & read its wiki. $\endgroup$ Oct 24 '18 at 14:53
  • $\begingroup$ Did @Sebastian's answer help you? If so, please consider upvoting or accepting the answer. This is done by clicking the upwards normal distribution & possibly clicking on the check mark below the vote total. If it isn't quite what you needed, consider leaving a comment below his answer. $\endgroup$ Oct 24 '18 at 14:55

You can use that you already proved the result for $n <k+1$.

Therefore $$P(\cup_{i=1}^k(A_{k+1}\cap A_i))=\sum_{i=1}^k P(A_{k+1}\cap A_i)-\sum_{i<j}^k P((A_{k+1}\cap A_i)\cap(A_{k+1}\cap A_j))- \sum_{i<j<l}^kP((A_{k+1}\cap A_i)\cap (A_{k+1} \cap A_j) \cap (A_{k+1} \cap A_l))- ... - (-1)^k((A_{k+1}\cap A_1)\cap...\cap(A_{k+1} \cap A_k))$$


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.