A Note on the Gaussian Minimum Conjecture

TL;DR

This paper proves that the Gaussian minimum conjecture, which compares the expected minimum absolute value of a Gaussian vector to independent Gaussian variables with the same variances, holds only when the dimension n equals 2.

Contribution

The paper establishes that the Gaussian minimum conjecture is valid exclusively for the case n=2, providing a complete characterization of its applicability.

Findings

The conjecture holds if and only if n=2.

For n>2, the conjecture does not hold.

The result clarifies the limitations of the Gaussian minimum conjecture.

Abstract

Let $n\geq 2$ and $(X_i,1\leq i\leq n)$ be a centered Gaussian random vector. The Gaussian minimum conjecture says that $E\left(\min_{1\leq i\leq n}|X_i|\right)\geq E\left(\min_{1\leq i\leq n}|Y_i|\right)$, where $Y_1,\ldots,Y_n$ are independent centered Gaussian random variables with $E(X_i^2)=E(Y_i^2)$ for any $i=1,\ldots,n$. In this note, we will show that this conjecture holds if and only if $n=2$.