Papers
Topics
Authors
Recent
Search
2000 character limit reached

Refinements of Alon-Babai-Suzuki-type intersection theorems via non-shadows and binomial support

Published 11 Mar 2026 in math.CO | (2603.10309v1)

Abstract: We prove a multilevel non-shadow refinement of the Alon--Babai--Suzuki (ABS) nonuniform restricted-intersection theorem. Let K=k1,,krK={k_1,\dots,k_r} and let LL be a set with L=s|L|=s. If FkK([n]k)\mathcal{F}\subseteq \bigcup_{k\in K}\binom{[n]}{k} is LL-intersecting and $k_i&gt;s-r$ for every ii, then F+j=sr+1<sup>s</sup>N<em>j(F)N(n,s,r),|\mathcal{F}| + \sum_{j=s-r+1}<sup>{s}</sup> |\mathcal{N}<em>j(\mathcal{F})| \le N(n,s,r), equivalently F</em>j=sr+1<sup>s</sup>jF.|\mathcal{F}| \le \sum</em>{j=s-r+1}<sup>{s}</sup> |\partial_j\mathcal{F}|. Thus the ABS bound is sharpened by the total non-shadow deficit on the top rr levels. In the modular setting, we take a coefficient-sensitive viewpoint: the polynomial method depends not just on the degree of the annihilator polynomial PL(t)=L(t)F<em>p[t]P_L(t)=\prod_{\ell\in L}(t-\ell)\in\mathbb{F}<em>p[t], but on which binomial terms actually appear in it. This yields a gap-free modular bound depending only on the active support levels of PLP_L. For almost-initial residue patterns L=0,1,,sm1R(modp)L={0,1,\dots,s-m-1}\cup R \pmod p we obtain the collapse F</em>i=0<sup>m(nsi).|\mathcal{F}|\le \sum</em>{i=0}<sup>{m}\binom{n}{s-i}. In particular, for consecutive residues L=0,1,,s1(modp)L={0,1,\dots,s-1}\pmod p we get the sharp bound F(ns)|\mathcal{F}|\le \binom{n}{s}, giving a partial negative answer to a question of Alon--Babai--Suzuki: the modular ABS bound N(n,s,r)N(n,s,r) is not attainable in the consecutive-residue regime whenever r2r\ge 2.

Authors (2)

Summary

No one has generated a summary of this paper yet.

Paper to Video (Beta)

No one has generated a video about this paper yet.

Whiteboard

No one has generated a whiteboard explanation for this paper yet.

Open Problems

We haven't generated a list of open problems mentioned in this paper yet.

Continue Learning

We haven't generated follow-up questions for this paper yet.