Let be a positive integer, let the sets be non-empty subsets of the set prove that there exist two disjoint non-empty subsets of the set : and such that .
Problem 1698
Official solution
1. Base Case:
For , we have the sets and which are non-empty subsets of . Since there are only two sets, they must either be equal or one is a subset of the other. In either case, we can find two disjoint non-empty subsets of such that their unions are equal. This is trivial and serves as our base case.
2. Induction Hypothesis:
Assume the statement is true for . That is, for any non-empty subsets of , there exist two disjoint non-empty subsets of such that the union of the subsets is equal.
3. Induction Step:
We need to show that the statement holds for . Consider the sets which are non-empty subsets of .
4. Definitions:
For a set of indices, define:
Note that is the union of the members of .
5. Classification of Sets:
Call a set *bad* if it contains and *good* otherwise. If there exists at most one *bad* set, then by the induction hypothesis, we are done. So assume there are at least two *bad* sets.
6. Assumption:
Assume that there do not exist disjoint subsets of indices such that are *good* for all and . Otherwise, we are trivially done.
7. Choosing Families:
Choose two families of sets and with elements each such that each of them contains at least one bad set. Remove the element from all the *bad* sets. By the induction hypothesis, we find two disjoint subsets of indices in and in satisfying:
8. Ensuring Disjointness:
By our assumption, at least one of or must contain a bad set, say . Similarly, assume that contains a bad set.
9. Combining Sets:
Hence, we find such that:
We can add back the to all the bad sets which exist both in the LHS and RHS, and so we are done.
10. Handling Overlap:
If there is a set on both sides, then are not disjoint. We can remove from both sides and still have:
This is because all elements of are in and . Hence, even if we delete the set from both sides, the elements will still remain on both sides, and we will still have .
Thus, we have shown that there exist two disjoint non-empty subsets of such that their unions are equal.