Unofficial archive — problems, solutions & results © IMC, reproduced with permission.

IMC / 1997 / Problems / Day 1, P6

IMC 1997 · Day 1 · P6

Suppose that FF is a family of finite subsets of N\mathbb{N} and for any two sets A,BFA, B \in F we have ABA \cap B \ne \emptyset.

a) Is it true that there is a finite subset YY of N\mathbb{N} such that for any A,BFA, B \in F we have ABYA \cap B \cap Y \ne \emptyset?

b) Is the statement a) true if we suppose in addition that all of the members of FF have the same size?

Justify your answers.

Solution (official)

a) No. Consider F={A1,B1,,An,Bn,}F = \{ A_1, B_1, \dots, A_n, B_n, \dots \}, where An={1,3,5,,2n1,2n}A_n = \{ 1, 3, 5, \dots, 2n-1, 2n \}, Bn={2,4,6,,2n,2n+1}B_n = \{ 2, 4, 6, \dots, 2n, 2n+1 \}.

b) Yes. We will prove inductively a stronger statement:

Suppose FF, GG are two families of finite subsets of N\mathbb{N} such that:

1) For every AFA \in F and BGB \in G we have ABA \cap B \ne \emptyset;

2) All the elements of FF have the same size rr, and elements of GG – size ss. (we shall write #(F)=r\#(F) = r, #(G)=s\#(G) = s).

Then there is a finite set YY such that ABYA \cup B \cup Y \ne \emptyset for every AFA \in F and BGB \in G.

The problem b) follows if we take F=GF = G.

Proof of the statement: The statement is obvious for r=s=1r = s = 1. Fix the numbers rr, ss and suppose the statement is proved for all pairs FF', GG' with #(F)<r\#(F') < r, #(G)<s\#(G') < s. Fix A0FA_0 \in F, B0GB_0 \in G. For any subset CA0B0C \subset A_0 \cup B_0, denote F(C)={AF:A(A0B0)=C}.F(C) = \{ A \in F : A \cap (A_0 \cup B_0) = C \}. Then F=CA0B0F(C)F = \bigcup\limits_{\emptyset \ne C \subset A_0 \cup B_0} F(C). It is enough to prove that for any pair of non-empty sets C,DA0B0C, D \subset A_0 \cup B_0 the families F(C)F(C) and G(D)G(D) satisfy the statement.

Indeed, if we denote by YC,DY_{C,D} the corresponding finite set, then the finite set C,DA0B0YC,D\bigcup\limits_{C, D \subset A_0 \cup B_0} Y_{C,D} will satisfy the statement for FF and GG. The proof for F(C)F(C) and G(D)G(D).

If CDC \cap D \ne \emptyset, it is trivial.

If CD=C \cap D = \emptyset, then any two sets AF(C)A \in F(C), BG(D)B \in G(D) must meet outside A0B0A_0 \cup B_0. Then if we denote F~(C)={AC:AF(C)}\widetilde{F}(C) = \{ A \setminus C : A \in F(C) \}, G~(D)={BD:BG(D)}\widetilde{G}(D) = \{ B \setminus D : B \in G(D) \}, then F~(C)\widetilde{F}(C) and G~(D)\widetilde{G}(D) satisfy the conditions 1) and 2) above, with #(F~(C))=#(F)#C<r\#(\widetilde{F}(C)) = \#(F) - \#C < r, #(G~(D))=#(G)#D<s\#(\widetilde{G}(D)) = \#(G) - \#D < s, and the inductive assumption works.

Similar problems