I need someone to assess whether the proof I attempt to provide for the following statement is correct.



$$ \left( A \cup B \right)\cap C \neq A \cup \left(B \cap C \right)$$

My approach: use indicator functions.

Two indicator functions $I_{S_1}$ are $I_{S_2}$ are identical if and only if the two sets, $S_1$ and $S_2$, are identical.

This means that $ \left( A \cup B \right)\cap C = A \cup \left(B \cap C \right)$ if and only if $I_{\left( A \cup B \right)\cap C} = I_{A \cup \left(B \cap C \right)}$.

$I_{\left( A \cup B \right)\cap C} = \left(I_A+I_B-I_A I_B\right)I_C = I_A I_C + I_B I_C - I_A I_B I_C$
$I_{A \cup \left(B \cap C \right)}= I_A + I_B I_C - I_A I_B I_C $

These 2 expressions are not always equal, therefore I conclude that these sets are not always identical.

I know, another approach I may try is to show that either $\left( A \cup B \right)\cap C \not\subset A \cup \left(B \cap C \right)$ or $\left( A \cup B \right)\cap C \not\supset A \cup \left(B \cap C \right)$. I am curious whether the first one above is rigorous enough?

I wrote wrongly the statement, which means some of the solutions provided below don't answer it precisely. Anyway I wasn't looking for a solution but rather for a proving method. Thanks anyone for suggestions!

  • 1
    $\begingroup$ If the question is that there exists some sets where $(A\cup B)\cap C\neq (A\cap B)\cup C$ then okay. If you've been taught indicator functions and it is an appropriate tool to use in your class, then fine. I personally would prefer just seeing an example. Many times for questions like these, examples and counterexamples can be formed using just $\emptyset$'s and $\{1\}$'s. I will point out also that in your final statement, we don't actually care that $(A\cup B)\cap C\subset (A\cap B)\cup C$ (although it is in fact a true statement), just that they aren't equal. $\endgroup$ – JMoravitz Aug 25 '17 at 15:09
  • 1
    $\begingroup$ If the question is that for all sets $A,B,C$ we have $(A\cup B)\cap C\neq (A\cap B)\cup C$, then think again and look for counterexamples. What happens if $A,B,C$ are all the same set? $\endgroup$ – JMoravitz Aug 25 '17 at 15:10
  • 1
    $\begingroup$ To show there exists at least one collection of sets $A,B,C$ where the statement is true, I'd show the example where $A=B=\emptyset$ and $C=\{1\}$. To show there exists at least one collection of sets $A,B,C$ where the statement is false (and therefore it isn't true that the statement is true for all sets, only for some), I'd show the counterexample where $A=B=C=\emptyset$. $\endgroup$ – JMoravitz Aug 25 '17 at 15:20
  • $\begingroup$ $(A\cup B)\cap C$ are elements that must be in $C$ and must also be in one of $A$ or $B$. $(A\cap B)\cup C$ are elements that might be in $C$ or could be in both $A$ and $B$. So if $a \in C$ but $a \not \in A\cup B$ that is a counter example. So Let $a \in C; a \not \in A; a\not \in C$ then $a \not \in (A\cup B)\cap C$ but $a \in (A\cap B)\cup C$. So that sets can not be equal. Far simpler and obvious. $\endgroup$ – fleablood Aug 25 '17 at 15:27
  • $\begingroup$ Let $A\ne \emptyset =C$. $\endgroup$ – DanielWainfleet Aug 25 '17 at 16:14

Just take $A = B = \{0\}$ and $C = \{1\}$. Then $A \cup B = A \cap B = \{0\}$ and $$\left( A \cup B \right)\cap C = \{0\} \cap \{1\} = \{\} = \emptyset$$ while $$\left( A \cap B \right)\cup C = \{0\} \cup \{1\} = \{ 0, 1 \}.$$

This simple examples shows that in general, $$\left( A \cup B \right)\cap C \neq \left( A \cap B \right)\cup C.$$

There are cases where equality holds. Just take $A = B = C.$


The elements of $(A \cup B) \cap C$ must be in $C$ and the elements of $(A \cap B) \cup C$ might be in $C$. Obviously these are different concepts and we can come up with a counter example but finding and element in the latter that is not in $C$.

Say $1 \not \in C$ but $1 \in (A\cap B) \cup C$ so $1 \in (A\cap B)$.

So let $1 \in A; 1 \in B; 1\not \in C$. Then $1 \not \in (A\cup B)\cap C \subset C$. But $1 \in (A\cap B) \subset (A\cap B) \cup C$. So $(A\cup B)\cap C\ne C \subset (A\cap B) \cup C$.

If you want to be even more specific: Let $A = B = \{1\}$ and $C = \emptyset$. Then $(A\cup B) \cap C = \emptyset$ while $(A \cap B) \cup C = \{1\}$.

  • $\begingroup$ Would you still use this approach if there were more caps, cups and more sets involved? $\endgroup$ – Sandu Ursu Aug 25 '17 at 15:43
  • 1
    $\begingroup$ Mmm, maybe. Actually, I'd probably break down $(A \cup B) \cap C = (A\cap C) \cup (B\cap C) = (A\cap B \cap C) \cup (A\cap B^c \cap C) \cup (A^c \cap B \cap C)$ whereas $(A\cap B)\cup C$ is $(A\cap B \cap C^c)\cup (A\cap B \cap C) \cup (A^c\cap B \cap C)\cup (A\cap B^c \cap C))\cup (A^c\cap B^c \cap C)$ which are different unions of disjoint sets. ... which I guess is the actual purpose of "indicator functions" which is a concept I have not seen taught much. $\endgroup$ – fleablood Aug 25 '17 at 15:56

You wrote down that

$$ \tag 1 I_A + I_B I_C - I_A I_B I_C \neq I_A I_C + I_B I_C - I_A I_B I_C $$

but did not explain why the LHS is not always equal to the RHS.

You need to finish off the argument with some simple observation. Well, if $C$ is the empty set, then $I_C = 0$ and (1) becomes

$$ \tag 2 I_A \ne 0$$

and I can think of some situations where (2) is true.


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.