English

A note on the Erd\H{o}s-Szekeres theorem in two dimensions

Combinatorics 2021-04-20 v3

Abstract

Burkill and Mirsky, and Kalmanson, prove independently that, for every r2,n1r\ge 2, n\ge 1, there is a sequence of r2nr^{2^n} vectors in Rn\mathbb R^n, which does not contain a subsequence of r+1r+1 vectors v1,v2,,vr+1v^1, v^2,\dots,v^{r+1} such that, for every ii between 1 and nn, (vij)1jr+1(v^{j}_i)_{1\le j\le r+1} forms a monotone sequence. Moreover, r2nr^{2^n} is the largest integer with this property. In this short note, for two vectors u=(u1,u2,,un)u = (u_1, u_2,\dots, u_n) and v=(v1,v2,,vn)v = (v_1, v_2, \dots, v_n) in Rn\mathbb R^n, we say that uvu\le v if, for every ii between 1 and nn, uiviu_i\le v_i. Just like Burkill and Mirsky, and Kalmanson, for every k,1,d2k, \ell\ge 1, d\ge 2 we find the maximal N1,N2N_1, N_2 (which turn out to be equal) such that there are numerical two-dimensional arrays of size (k+1)×N1(k+\ell-1)\times N_1 and (k+)×N2(k+\ell)\times N_2, which neither contain a subarray of size k×dk\times d, whose columns form a non-decreasing sequence of dd vectors in Rk\mathbb R^k, nor contain a subarray of size ×d\ell\times d, whose columns form a non-increasing sequence of dd vectors in R\mathbb R^{\ell}. In a consequent discussion, we consider a generalisation of this setting and make a connection with a famous problem in coding theory.

Keywords

Cite

@article{arxiv.2009.08164,
  title  = {A note on the Erd\H{o}s-Szekeres theorem in two dimensions},
  author = {Lyuben Lichev},
  journal= {arXiv preprint arXiv:2009.08164},
  year   = {2021}
}

Comments

6 pages, 1 figure. In the second version, the upper bound on M_2(n) in the second remark after Theorem 2.1 was corrected. In the third version, minor corrections were introduced