I want to generalize a formula and I need your help with this. This is not my homework or assignment but I need to come up with a concise formula that fits my documentation.

Background for my problem:

Considering all events to be independent of each other,

Let the probability of $Event$ $0$ be $P_{0}$ and $Event$ $1$ be $P_{1}$ and so on ...

Then the probability that two events occur simultaneously is : $P_0P_1$ . This is nothing but the area of intersection of two circles $P_0$ and $P_1$.

Continuing the same way,

The total probability of simultaneous occurrence in case of three events is:

$P_0 P_1 + P_0P_2 + P_1P_2 - 2 P_{0}P_{1}P_{2}$.

Also can be visualized by drawing three intersecting circles.

One clarification here: This gives me the total probability of any two events plus all the three occurring at the same time right?

I cannot visualize the formula by drawing circles anymore. How can the above formula be generalized to get the probability of simultaneous occurrences when there are 4, 5, ... independent events.

I have seen that the inclusion-exclusion principle is the answer. But I am not able to get an intuition for it. The inclusion-exclusion principle gives the probability of 2,3,4 sets intersecting but isn't my question different?

I get this doubt because, probability of four independent events occurring simultaneous is: $P_0P_1P_2P_3$. But what I need is a general formula for the total probability of two or more independent events at the same time.

Can anyone of you please throw some light?

Yes indeed I meant that the probability of "two or more events". The answer you have given is very precise and the one I was looking for. Yes it is boring to try to visualize circles when the number exceeds three. Instead, in general if use 1 - ($w_0 + w_1$) then we should land up correctly given that the events are independent. Thank you so much.

  • $\begingroup$ It seems that by "The total probability of simultaneous occurrence " you actually mean "the ocurrence of more than two events" $\endgroup$ – leonbloy May 30 '12 at 16:12



It seems that by "The total probability of simultaneous occurrence " you actually mean "the ocurrence of two or more events" The inclusion-exclusion principle, in its most basic form, is usually apply to get the probability of "one or more" events (i.e.: an OR). It can certainly be applied here, also, but it gets a little more complicated. Your "elementary" events are now rather the pairs $A_{ij}=E_i \cap E_j$ (the intersections of two circles in your Venn diagram). Applying the inclusion-exclusion principle to the three events you'd get

$$P(A_{01} \cup A_{12} \cup A_{02}) =\\ P(A_{01})+P(A_{12})+P(A_{02})- P(A_{01} \cap A_{12})-P(A_{01} \cap A_{12})-P(A_{12} \cap A_{02}) + P(A_{12} \cap A_{02} \cap A_{02})=\\ $$

which, because $P(A_{12} \cap A_{02}) = P(E_0\cap E_1 \cap E_2)= P(A_{12} \cap A_{02} \cap A_{12})$ equals

$$P(E_0 \cap E_1) + P(E_1 \cap E_2) +P(E_0 \cap E_2) - 2 P(E_0\cap E_1 \cap E_2) $$

To extend this to 4 events is straighwforward, but borish and hard to visualize.

Of course, in your case, if the events are independent, it's more easy to compute the complement: $1-(w_0 + w_1)$ where $w_0=$ probability of no event and $w_1=$ probability of exactly one event.

$w_0 = (1-P_0)(1-P_1)(1-P_2)(1-P_3)$ $w_1 = P_0(1-P_1)(1-P_2)(1-P_3) + (1-P_0)P_1(1-P_2)(1-P_3) + (1-P_0)(1-P_1)P_2(1-P_3)+ (1-P_0)(1-P_1)(1-P_2)P_3 $


Inclusion-Exclusion formulas

Let $A_1, A_2,\dots, A_n$ be events. The probability that exactly $k$ of them occur $(k\geq 0)$ is

$$p_n(k)=\sum_{i=k}^n (-1)^{i-k}{i\choose k}S_i,$$

while the probability that $k$ or more of them occur $(k\geq 1)$ is

$$P_n(k)=\sum_{i=k}^n (-1)^{i-k}{i-1\choose k-1}S_i.$$

Here $S_0=1$, $S_1=\sum_i P(A_i)$, $S_2=\sum_{i< j} P(A_i A_j)$, $S_3=\sum_{i< j<k} P(A_i A_j A_k)$, etc.


Just use the opposite event : one or less.

For example for 3 events (none+only $P_1$+only $P_2$+only $P_3$): $$E_3=(1-P_1)(1-P_2)(1-P_3)+P_1(1-P_2)(1-P_3)+(1-P_1)P_2(1-P_3)+(1-P_1)(1-P_2)P_3$$

Then what you want is only $1-E_n$, that is easy to compute now.


$P[\bigcup_{k=1}^N P_k]=\sum_{k=1}^NP_k-\sum_{j<k}^NP_jP_k+\dots+(-1)^{N+1}P_1P_2\dots P_N$


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.