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

IMC / 2021 / Problems / Day 2, P8

IMC 2021 · Day 2 · P8

killer

Let nn be a positive integer. At most how many distinct unit vectors can be selected in Rn\mathbb{R}^n such that from any three of them, at least two are orthogonal?

(proposed by Alexander Polyanskii, Moscow Institute of Physics and Technology; based on results of Paul Erdős and Moshe Rosenfeld)

Solution 1 of 2 (official)

Hint: Play with the Gram matrix of these vectors.

2n2n is the maximal number.

An example of 2n2n vectors in the set is given by a basis and its opposite vectors. In the rest of the text we prove that it is impossible to have 2n+12n + 1 vectors in the set.

Consider the Gram matrix AA with entries aij=eieja_{ij} = e_i \cdot e_j. Its rank is at most nn, its eigenvalues are real and non-negative. Put B=AI2n+1B = A - I_{2n+1}, this is the same matrix, but with zeros on the diagonal. The eigenvalues of BB are real, greater or equal to 1-1, and the multiplicity of 1-1 is at least n+1n + 1.

The matrix C=B3C = B^3 has the following diagonal entries cii=ijkiaijajkaki.c_{ii} = \sum_{i \ne j \ne k \ne i} a_{ij} a_{jk} a_{ki}. The problem statement implies that in every summand of this expression at least one factor is zero. Hence trC=0\operatorname{tr} C = 0. Let x1,,xmx_1, \dots, x_m be the positive eigenvalues of BB, their number is mnm \le n as noted above. From trB=trC\operatorname{tr} B = \operatorname{tr} C we deduce (taking into account that the eigenvalues between 1-1 and 0 satisfy λ3λ\lambda^3 \ge \lambda): x1++xmx13++xm3x_1 + \dots + x_m \ge x_1^3 + \dots + x_m^3 Applying trC=0\operatorname{tr} C = 0 once again and noting that CC has eigenvalue 1-1 of multiplicity at least n+1n + 1, we obtain x13++xm3n+1.x_1^3 + \dots + x_m^3 \ge n + 1. It also follows that (x1++xm)3(x13++xm3)(n+1)2.(x_1 + \dots + x_m)^3 \ge (x_1^3 + \dots + x_m^3)(n+1)^2. By Hölder's inequality, we obtain (x13++xm3)m2(x1++xm)3,(x_1^3 + \dots + x_m^3) m^2 \ge (x_1 + \dots + x_m)^3, which is a contradiction with mnm \le n.

Solution 2 of 2 (official)

Let PiP_i denote the projection onto ii-th vector, i=1,,Ni = 1, \dots, N. Then our relation reads as tr(PiPjPk)=0\operatorname{tr}(P_i P_j P_k) = 0 for distinct i,j,ki, j, k. Consider the operator Q=i=1NPiQ = \sum_{i=1}^{N} P_i, it is non-negative definite, let t1,,tnt_1, \dots, t_n be its eigenvalues, ti=trQ=N\sum t_i = \operatorname{tr} Q = N. We get ti3=trQ3=N+6i<jtrPiPj=N+3(trQ2N)=3ti22N\sum t_i^3 = \operatorname{tr} Q^3 = N + 6 \sum_{i<j} \operatorname{tr} P_i P_j = N + 3 (\operatorname{tr} Q^2 - N) = 3 \sum t_i^2 - 2N (we used the obvious identities like trPiPjPi=trPi2Pj=trPiPj\operatorname{tr} P_i P_j P_i = \operatorname{tr} P_i^2 P_j = \operatorname{tr} P_i P_j). But (ti2)2(ti+1)=ti33ti2+40(t_i - 2)^2 (t_i + 1) = t_i^3 - 3 t_i^2 + 4 \geqslant 0, thus 2N=ti33ti24n-2N = \sum t_i^3 - 3 t_i^2 \geqslant -4n and N2nN \leqslant 2n.

How the field did

contestants scored
514
average (of 10)
0.05
solved (≥ 80%)
0.2%
near-0 (≤ 10%)
99.4%
discrimination
0.08

Score distribution (field cohort)

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

Similar problems

IMC 2004 · Day 2 · P12killeravg 0.2/10 · solved 1% · near-0 97% · disc 0.13
IMC 2010 · Day 2 · P9killeravg 0.5/10 · solved 2% · near-0 93% · disc 0.42
IMC 2024 · Day 1 · P3mediumavg 5.3/10 · solved 40% · near-0 30% · disc 0.55