English

Note on Existence and Non-Existence of Large Subsets of Binary Vectors with Similar Distances

Discrete Mathematics 2012-12-04 v2 Combinatorics

Abstract

We consider vectors from {0,1}n\{0,1\}^n. The weight of such a vector vv is the sum of the coordinates of vv. The distance ratio of a set LL of vectors is dr(L):=max{ρ(x,y): x,yL}/min{ρ(x,y): x,yL, xy},{\rm dr}(L):=\max \{\rho(x,y):\ x,y \in L\}/ \min \{\rho(x,y):\ x,y \in L,\ x\neq y\}, where ρ(x,y)\rho(x,y) is the Hamming distance between xx and yy. We prove that (a) for every constant λ>1\lambda>1 there are no positive constants α\alpha and CC such that every set KK of at least λp\lambda^p vectors with weight pp contains a subset KK' with KKα|K'|\ge |K|^{\alpha} and dr(K)C{\rm dr}(K')\le C, % even when Kλ|K|\ge \lambda, (b) For a set KK of vectors with weight pp, and a constant C>2C>2, there exists KKK'\subseteq K such that dr(K)C{\rm dr}(K')\le C and KKα|K'| \ge |K|^\alpha, where α=1/log(p/2)/log(C/2)\alpha = 1/ \lceil \log(p/2)/\log(C/2) \rceil.

Keywords

Cite

@article{arxiv.1202.6260,
  title  = {Note on Existence and Non-Existence of Large Subsets of Binary Vectors with Similar Distances},
  author = {Gregory Gutin and Mark Jones},
  journal= {arXiv preprint arXiv:1202.6260},
  year   = {2012}
}