1

How would I go about proving this statement: $\mathbb{P}\left(A_{1} \cup A_{2} \cup A_{3}\right)=\left(\sum_{i=1}^{3} \mathbb{P}\left(A_{i}\right)\right)-\mathbb{P}\left(A_{1} \cap A_{2}\right)-\mathbb{P}\left(A_{1} \cap A_{3}\right)-\mathbb{P}\left(A_{2} \cap A_{3}\right)+\mathbb{P}\left(A_{1} \cap A_{2} \cap A_{3}\right)$

The probability space is $(\Omega,\mathbb{P})$ and for the upper equation applies $A_{i} \subseteq \Omega, i=1,2,3$

My idea is to reshape the LHS:

So we have

(i) $\left(\sum_{i=1}^{3} \mathbb{P}\left(A_{i}\right)\right)-\mathbb{P}\left(A_{1} \cap A_{2}\right)-\mathbb{P}\left(A_{1} \cap A_{3}\right)-\mathbb{P}\left(A_{2} \cap A_{3}\right)+\mathbb{P}\left(A_{1} \cap A_{2} \cap A_{3}\right)$

with

(ii) $\left(\sum_{i=1}^{3} \mathbb{P}\left(A_{i}\right)\right) = \mathbb{P}(A_{1}) \cup \mathbb{P}(A_{2}) \cup \mathbb{P}(A_{3}) $

(iii) $ \mathbb{P}(A_{1}) = \mathbb{P}\left(A_{1} \setminus A_{2}\right) + \mathbb{P}\left(A_{1} \cap A_{2}\right) <=> -\mathbb{P}\left(A_{1} \cap A_{2}\right) = \mathbb{P}\left(A_{1} \setminus A_{2}\right) - \mathbb{P}(A_{1})$

same for $\mathbb{P}(A_{2})$ and $\mathbb{P}(A_{3})$

Now we combine (i), (ii) and we get:

(iV) $\mathbb{P}(A_{1}) \cup \mathbb{P}(A_{2}) \cup \mathbb{P}(A_{3})-\mathbb{P}\left(A_{1} \cap A_{2}\right)-\mathbb{P}\left(A_{1} \cap A_{3}\right)-\mathbb{P}\left(A_{2} \cap A_{3}\right)+\mathbb{P}\left(A_{1} \cap A_{2} \cap A_{3}\right)$

Now we combine (iV), (iii) and we get:

(V) $\mathbb{P}(A_{1}) \cup \mathbb{P}(A_{2}) \cup \mathbb{P}(A_{3})+ \mathbb{P}\left(A_{1} \setminus A_{2}\right) - \mathbb{P}(A_{1})+ \mathbb{P}\left(A_{3} \setminus A_{1}\right) - \mathbb{P}(A_{3})+ \mathbb{P}\left(A_{2} \setminus A_{3}\right) - \mathbb{P}(A_{2})+\mathbb{P}\left(A_{1} \cap A_{2} \cap A_{3}\right) = \mathbb{P}\left(A_{1} \setminus A_{2}\right) + \mathbb{P}\left(A_{3} \setminus A_{1}\right) + \mathbb{P}\left(A_{2} \setminus A_{3}\right) +\mathbb{P}\left(A_{1} \cap A_{2} \cap A_{3}\right)$

Now we use this equation:

(Vi) $\mathbb{P}(A \cap B)=\mathbb{P}(A \backslash B)+\mathbb{P}(B \backslash A)+\mathbb{P}(A \cap B)$

And finally we use (VI) on (V)

$\mathbb{P}\left(A_{1} \setminus A_{2}\right) + \mathbb{P}\left(A_{3} \setminus A_{1}\right) + \mathbb{P}\left(A_{2} \setminus A_{3}\right) +\mathbb{P}\left(A_{1} \cap A_{2} \cap A_{3}\right) = \mathbb{P}\left(A_{1} \cup A_{2} \cup A_{3}\right)$

Could this work? If yes i will get drunk today :)

Asaf Karagila
  • 393,674
germinator
  • 13
  • 4
  • 3
    Is $\mathbb P$ the probability measure function? How can you put the union sign for a number? – Fareed Abi Farraj Apr 12 '19 at 22:19
  • 2
    I think this works (I didn't check any of the details but the idea seems fine) but it's way more complicated than necessary. You ought to use the inclusion-exclusion principle, namely that $\mathbb{P}(X \cup Y) = \mathbb{P}(X) + \mathbb{P}(Y) - \mathbb{P}(X \cap Y)$. For your problem, define $B = A_1 \cup A_2$. Then $\mathbb{P}(A_1 \cup A_2 \cup A_3) = \mathbb{P}(B \cup A_3) = \mathbb{P}(B) + \mathbb{P}(A_3) - \mathbb{P}(B \cap A_3)$. Use inclusion-exclusion again to rewrite $\mathbb{P}(B)$ and $\mathbb{P}(B \cap A_3)$ in terms of the $\mathbb{P}(A_i)$'s. – diracdeltafunk Apr 12 '19 at 22:21
  • @ Fareed AF Yes $\mathbb{P}$ stands for the probability measure function. Because of my bad English, i do not understand the second question :) – germinator Apr 12 '19 at 22:25
  • @ diracdeltafunk Well, I've never heard of it but thanks for the tip, i try to use it :) – germinator Apr 12 '19 at 22:27
  • 2
    This is the "inclusion-exclusion principle" for three sets. In fact, there is a general version for $n$ sets, see here for example: https://math.stackexchange.com/questions/1427473/inclusion-exclusion-probability-proof-using-a-partition-of-the-space. – Minus One-Twelfth Apr 12 '19 at 22:33
  • @SK19 Mmh could be, but I think my problem is a little bit different ? – germinator Apr 12 '19 at 22:44
  • Have you looked at the first answer? I found the explanation for three events quite sufficient. – SK19 Apr 12 '19 at 22:46
  • @germinator, I can't see how your problem is different from what SK19 posted.

    At any rate, I would recommend first proving a special case: that $$ \mathbb{P}(A_1 \cup A_2) = \mathbb{P}(A_1) + \mathbb{P}(A_2) - \mathbb{P}(A_1 \cap A_2). $$ And even before this, how about first proving the first equality in this Wikipedia article. In general, all the terms that are subtracted are intended to correct for an "overcount".

    – avs Apr 12 '19 at 22:48
  • @SK19 Yes i saw it. Sry, my math skills aren't good, I'm still in school 11 grade (Germany).

    My intention is only to find out if my way works

    – germinator Apr 12 '19 at 22:49
  • You may add the proof-verification tag. – SK19 Apr 12 '19 at 22:56
  • @SK19 done:) and sry for the problems – germinator Apr 12 '19 at 22:58

1 Answers1

1

As pointed out in the comments, this identity is the probabilistic version of the Inclusion-Exclusion Principle in the special case of three sets.

Your calculation looks good to me. I'd consider proof by diagram rigorous in this setting. For example, one can read off your equation (V), namely, $$\Bbb P(A_1 \cup A_2 \cup A_3) = \color{#ff0000}{\Bbb P(A_2 \setminus A_3)} + \color{#009f00}{\Bbb P(A_3 \setminus A_1)} + \color{#3f7fff}{\Bbb P(A_1 \setminus A_2)} + \color{#7f00ff}{\Bbb P(A_1 \cap A_2 \cap A_3)} ,$$ directly from the below Venn diagram, appropriately labeled.

enter image description here

Travis Willse
  • 99,363
  • "Your calculation looks good to me" This is exactly what I wanted to hear :) I will study Inclusion-Exclusion Principle could be helpful for future problems. To solve this problem on my way has taken me way to long. – germinator Apr 12 '19 at 23:21
  • You're welcome, glad to help! Another, less symmetric way of proving this is applying the two-set Inclusion-Exclusion Principle twice. A good exercise for checking that you understand that principle well is to prove the $n$-set version by induction. – Travis Willse Apr 12 '19 at 23:35