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

IMC / 2019 / Problems / Day 2, P8

IMC 2019 · Day 2 · P8

very hard

Let x1,,xnx_1, \dots, x_n be real numbers. For any set I{1,2,,n}I \subset \{1, 2, \dots, n\} let s(I)=iIxis(I) = \sum\limits_{i \in I} x_i. Assume that the function Is(I)I \mapsto s(I) takes on at least 1.8n1.8^n values where II runs over all 2n2^n subsets of {1,2,,n}\{1, 2, \dots, n\}. Prove that the number of sets I{1,2,,n}I \subset \{1, 2, \dots, n\} for which s(I)=2019s(I) = 2019 does not exceed 1.7n1.7^n.

Proposed by Fedor Part and Fedor Petrov, St. Petersburg State University

Solution (official)

Choose disctint sets I1,,IA{1,2,,n}I_1, \dots, I_A \subset \{1, 2, \dots, n\} where A1.8nA \ge 1.8^n, and let J1,,JB{1,2,,n}J_1, \dots, J_B \subset \{1, 2, \dots, n\} be all sets so that S(Ji)=2019S(J_i) = 2019; for the sake of contradiction, assume that B1.7nB \ge 1.7^n.

Every set I{1,2,,n}I \subset \{1, 2, \dots, n\} can be identified with a 010-1 vector of length nn: the kkth coordinate in the vector is 1 if kIk \in I. Then s(I)=I,Xs(I) = \langle I, X \rangle, where X=(x1,,xn)X = (x_1, \dots, x_n) and ,\langle \cdot, \cdot \rangle stands for the usual scalar product.

For all ordered pairs (a,b){1,,A}×{1,,B}(a, b) \in \{1, \dots, A\} \times \{1, \dots, B\} consider the vector IaJb{1,0,1}nI_a - J_b \in \{-1, 0, 1\}^n. By the pigeonhole principle, since AB(1.81.7)n>3nAB \ge (1.8 \cdot 1.7)^n > 3^n, there are two pairs (a,b)(a, b) and (c,d)(c, d) such that IaJb=IcJdI_a - J_b = I_c - J_d. Multiplying this by XX we get s(Ia)2019=s(Ic)2019s(I_a) - 2019 = s(I_c) - 2019; that implies a=ca = c. But then Jb=JdJ_b = J_d, that is, b=db = d, and our pairs coincide. Contradiction.

How the field did

contestants scored
360
average (of 10)
0.77
solved (≥ 80%)
7.2%
near-0 (≤ 10%)
91.4%
discrimination
0.38

Score distribution (field cohort)

Computed on contestants with a meaningful total (field cohort); discrimination is the corrected item–total correlation.

Similar problems

IMC 2018 · Day 2 · P8very hardavg 1.6/10 · solved 6% · near-0 64% · disc 0.40
IMC 2016 · Day 1 · P4very hardavg 1.3/10 · solved 5% · near-0 79% · disc 0.40
IMC 2008 · Day 1 · P4very hardavg 1.4/10 · solved 5% · near-0 72% · disc 0.45
IMC 2014 · Day 1 · P5very hardavg 1.0/10 · solved 5% · near-0 78% · disc 0.49