Table of Contents
Fetching ...

A Correlation Inequality on Three Functions

Kada Williams

TL;DR

This work investigates whether the Harris-Kleitman two-set correlation inequality extends to three upward closed families $X$, $Y$, and $Z$ on the Boolean lattice when they have equal densities. It derives bounds on the density of the set $S_1$ of points contained in exactly one of the three families, using Harris-Kleitman inequalities, a multi-slice decomposition, and measure-preserving mappings on weighted cubes. The key results include a concrete counterexample showing that the equal-density conjecture fails (even in small dimensions) and an upper bound of about $0.515$ for $|S_1|/2^n$, along with a larger counterexample to Kahn's conjecture in a high-dimensional weighted cube, highlighting a gap between known bounds and extremal configurations. The work also discusses a weighted-poset framework and opens questions about finer triple-density thresholds in the Boolean lattice.

Abstract

Let $X$ and $Y$ be upward closed set systems in the lattice of $\{0,1\}^n$. The celebrated Harris-Kleitman inequality implies that if $|X|=α2^n$, $|Y|=β2^n$, the density of the set of points in exactly one of $X$ and $Y$ is maximal when $X$ and $Y$ are independent, meaning $|X\cap Y|=αβ2^n$. Is the same true of three upward closed systems, $X$, $Y$, and $Z$? Suppose $|X|=|Y|=|Z|$. Kahn asked whether the set of points in exactly one of $X$, $Y$, $Z$ has density at most $\frac49$. We answer this question in the negative.

A Correlation Inequality on Three Functions

TL;DR

This work investigates whether the Harris-Kleitman two-set correlation inequality extends to three upward closed families , , and on the Boolean lattice when they have equal densities. It derives bounds on the density of the set of points contained in exactly one of the three families, using Harris-Kleitman inequalities, a multi-slice decomposition, and measure-preserving mappings on weighted cubes. The key results include a concrete counterexample showing that the equal-density conjecture fails (even in small dimensions) and an upper bound of about for , along with a larger counterexample to Kahn's conjecture in a high-dimensional weighted cube, highlighting a gap between known bounds and extremal configurations. The work also discusses a weighted-poset framework and opens questions about finer triple-density thresholds in the Boolean lattice.

Abstract

Let and be upward closed set systems in the lattice of . The celebrated Harris-Kleitman inequality implies that if , , the density of the set of points in exactly one of and is maximal when and are independent, meaning . Is the same true of three upward closed systems, , , and ? Suppose . Kahn asked whether the set of points in exactly one of , , has density at most . We answer this question in the negative.

Paper Structure

This paper contains 6 sections, 2 theorems, 11 equations.

Key Result

Theorem 1

Let $X$ and $Y$ be upward closed set systems on a ground set of $n$ elements. Then the following inequality holds Har, Kle:

Theorems & Definitions (4)

  • Theorem 1: Harris-Kleitman
  • Conjecture 2
  • Theorem 3
  • Conjecture 4: Kahn