English

Sparse Euclidean Spanners with Tiny Diameter: A Tight Lower Bound

Data Structures and Algorithms 2022-01-03 v2

Abstract

In STOC'95 [ADMSS'95] Arya et al. showed that any set of nn points in Rd\mathbb R^d admits a (1+ϵ)(1+\epsilon)-spanner with hop-diameter at most 2 (respectively, 3) and O(nlogn)O(n \log n) edges (resp., O(nloglogn)O(n \log \log n) edges). They also gave a general upper bound tradeoff of hop-diameter at most kk and O(nαk(n))O(n \alpha_k(n)) edges, for any k2k \ge 2. The function αk\alpha_k is the inverse of a certain Ackermann-style function at the k/2\lfloor k/2 \rfloorth level of the primitive recursive hierarchy, where α0(n)=n/2\alpha_0(n) = \lceil n/2 \rceil, α1(n)=n\alpha_1(n) = \left\lceil \sqrt{n} \right\rceil, α2(n)=logn\alpha_2(n) = \lceil \log{n} \rceil, α3(n)=loglogn\alpha_3(n) = \lceil \log\log{n} \rceil, α4(n)=logn\alpha_4(n) = \log^* n, α5(n)=12logn\alpha_5(n) = \lfloor \frac{1}{2} \log^*n \rfloor, \ldots. Roughly speaking, for k2k \ge 2 the function αk\alpha_{k} is close to k22\lfloor \frac{k-2}{2} \rfloor-iterated log-star function, i.e., log\log with k22\lfloor \frac{k-2}{2} \rfloor stars. Also, α2α(n)+4(n)4\alpha_{2\alpha(n)+4}(n) \le 4, where α(n)\alpha(n) is the one-parameter inverse Ackermann function, which is an extremely slowly growing function. Whether or not this tradeoff is tight has remained open, even for the cases k=2k = 2 and k=3k = 3. Two lower bounds are known: The first applies only to spanners with stretch 1 and the second is sub-optimal and applies only to sufficiently large (constant) values of kk. In this paper we prove a tight lower bound for any constant kk: For any fixed ϵ>0\epsilon > 0, any (1+ϵ)(1+\epsilon)-spanner for the uniform line metric with hop-diameter at most kk must have at least Ω(nαk(n))\Omega(n \alpha_k(n)) edges.

Cite

@article{arxiv.2112.09124,
  title  = {Sparse Euclidean Spanners with Tiny Diameter: A Tight Lower Bound},
  author = {Hung Le and Lazar Milenkovic and Shay Solomon},
  journal= {arXiv preprint arXiv:2112.09124},
  year   = {2022}
}
R2 v1 2026-06-24T08:20:59.109Z