Smooth rigidity and Remez inequalities via Topology of level sets

Abstract

A smooth rigidity inequalitiy provides an explicit lower bound for the $(d+1)$-st derivatives of a smooth function $f$, which holds, if $f$ exhibits certain patterns, forbidden for polynomials of degree $d$. The main goal of the present paper is twofold: first, we provide an overview of some recent results and questions related to smooth rigidity, which recently were obtained in Singularity Theory, in Approximation Theory, and in Whitney smooth extensions. Second, we prove some new results, specifically, a new Remez-type inequality, and on this base we obtain a new rigidity inequality. In both parts of the paper we stress the topology of the level sets, as the input information. Here are the main new results of the paper: \smallskip Let $B^n$ be the unit $n$-dimensional ball. For a given integer $d$ let $Z\subset B^n$ be a smooth compact hypersurface with $N=(d-1)^n+1$ connected components $Z_j$. Let $μ_j$ be the $n$-volume of the interior of $Z_j$, and put $μ=\min μ_j, \ j=1,\ldots, N$. Then for each polynomial $P$ of degree $d$ on ${\mathbb R}^n$ we have $$ \frac{\max_{B^n}|P|}{\max_{Z}|P|}\le (\frac{4n}μ)^d. $$ As a consequence, we provide an explicit lower bound for the $(d+1)$-st derivatives of any smooth function $f$, which vanishes on $Z$, while being of order $1$ on $B^n$ (smooth rigidity)}: $$ ||f^{(d+1)}||\ge \frac{1}{(d+1)!}(\frac{4n}μ)^d. $$ We also provide an interpretation, in terms of smooth rigidity, of one of the simplest versions of the results in \cite{Ler.Ste}.