On high dimensional maximal functions associated to Gaussians, balls, and spheres

Abstract: We prove that for each $p\in (1,\infty),$ the norms on $L^p(\mathbb{R}^d)$ of the maximal functions associated to Gaussians (heat semigroup), balls (Hardy-Littlewood averages), and spheres (spherical averages) converge, as the dimension $d\to \infty,$ to the same quantity $\lambda(p)$. This is derived from the fact that the norms on $L^2(\mathbb{R}^d)$ of the maximal functions corresponding to the differences of Gaussian, ball, and spherical averages converge to zero with the dimension $d.$ The fact is proved with the aid of estimates for Fourier multiplier symbols corresponding to these averages, a general principle that allows us to control the norm of a maximal function corresponding to a Fourier multiplier operator by the norm of the multiplier operator itself, and concentration properties of high dimensional Gaussian random vectors. Moreover, relying on the properties of the $d$-dimensional maximal function for the heat semigroup $\mathcal{G}\ast^d$, we show that $\lambda(p)$ satisfies $$ \frac25\frac{p}{p-1}\le|\mathcal{G}\ast^1|_{L^p(\mathbb{R})\rightarrow L^p(\mathbb{R})}\le \lambda(p)\le \frac{p}{p-1}. $$ In particular, to obtain the middle inequality we show that the norms on $L^p(\mathbb{R}^d)$ of the maximal function for the heat semigroup are non-decreasing in $d.$