A revision of Litvak's conjecture on Gaussian minima and a volumetric zone conjecture

Abstract: Litvak (2018) conjectured that, for any $p &gt; 0$, the quantity $\mathbb{E}[\min_{i = 1}ⁿ |g_i|<sup>p]$ where $g \sim \mathcal{N}(0, Σ)$ is a centered Gaussian random vector is minimized among $n \times n$ correlation matrices $Σ$ by the Gram matrix of the regular simplex in $\mathbb{R}^{n - 1}$. We disprove this conjecture: the matrix with entries $Σ^{{\mathrm{cos}}_{ij}=\cos(π(i} - j) / n)$ already achieves a smaller moment for $p = 2$ and $n = 4$. We propose that $Σ<sup>{\mathrm{cos}}$ is in fact the correct minimizer of these moments for all $p &gt; 0$ and $n \geq 1$. Towards proving this, we conjecture a volumetric extension of Fejes Tóth's zone conjecture (1973), whose covering version was proved by Jiang and Polyanskii (2017). Conditional on this conjecture, we show the stronger result that $\min_{i = 1}ⁿ |g_i|$ for $g \sim \mathcal{N}(0, Σ<sup>{\mathrm{cos}})$ is stochastically dominated by $\min_{i = 1}ⁿ |h_i|$ for $h \sim \mathcal{N}(0, Σ)$ for any $n \times n$ correlation matrix $Σ$. Our counterexample $Σ<sup>{\mathrm{cos}}$ was found by the AlphaEvolve AI-assisted optimization system, and we also include a brief discussion of its application to such problems.