Answers are grounded in this paper and cite their sources.
Let X be an n -element point set in the k -dimensional unit cube [0,1]ᵏ [ 0 , 1 ] k where k ≥ 2 k ≥ 2 . According to an old result of Bollobás and Meir (Oper Res Lett 11:19–21, 1992) , there exists a cycle (tour) x₁, x₂, … , xₙ x 1 , x 2 , … , x n through the n points, such that ( ∑ ᵢ₌₁ⁿ |xᵢ - xᵢ₊₁|ᵏ ) 1/k ≤ cₖ ∑ i = 1 n | x i - x i + 1 | k 1 / k ≤ c k , where $$|x-y|$$ | x - y | is the Euclidean distance between x and y , and cₖ c k is an absolute constant that depends only on k , where xₙ₊₁ ≡ x₁ x n + 1 ≡ x 1 . From the other direction, for every k ≥ 2 k ≥ 2 and n ≥ 2 n ≥ 2 , there exist n points in [0,1]ᵏ [ 0 , 1 ] k , such that their shortest tour satisfies ( ∑ ᵢ₌₁ⁿ |xᵢ - xᵢ₊₁|ᵏ ) 1/k = 21/k · √k ∑ i = 1 n | x i - x i + 1 | k 1 / k = 2 1 / k · k . For the plane, the best constant is c₂=2 c 2 = 2 and this is the only exact value known. Bollobás and Meir showed that one can take cₖ = 9 ( 2/3 ) 1/k · √k c k = 9 2 3 1 / k · k for every k ≥ 3 k ≥ 3 and conjectured that the best constant is cₖ = 21/k · √k c k = 2 1 / k · k , for every k ≥ 2 k ≥ 2 . Here we significantly improve the upper bound and show that one can take cₖ = 3 √5 ( 2/3 ) 1/k · √k c k = 3 5 2 3 1 / k · k or cₖ = 2.91 √k \ (1+oₖ(1)) c k = 2.91 k ( 1 + o k ( 1 ) ) . Our bounds are constructive. We also show that c₃ ≥ 27/6 c 3 ≥ 2 7 / 6 , which disproves the conjecture for $$k=3$$ k = 3 . Connections to matching problems, power assignment problems, related problems, including algorithms, are discussed in this context. A slightly revised version of the Bollobás–Meir conjecture is proposed.
No takes yet. Share an insight, caveat, or question.
Balogh et al. (2024) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: