Kahane-Salem-Zygmund polynomial inequalities via Rademacher processes (Q524252)

Given \(m,n \in \mathbb{N}\) and a family of independent Bernoulli random variables \(\varepsilon_\alpha, \alpha \in \mathbb{N}_0^n\) with \(|\alpha|=m\), on a probability space, let \(P(\omega,z) = \sum_{|\alpha|=m} \varepsilon_\alpha(\omega) c_\alpha z^\alpha\) be an \(m\)-homogeneous Bernoulli random polynomial in the variable \(z \in \mathbb{C}^n\). The famous Kahane-Salem-Zygmund inequality shows that the expectation \(\mathbb{E} \|P\|_\infty \leq C \big( n \log m \sum_\alpha |c_\alpha|^2\big)^{1/2}\), where \(C>0\) is a uniform constant and the supremum norm \(\|P\|_\infty\) of \(P\) is taken on the unit ball of \(\ell_\infty^n\). In recent years various extensions of the KSZ-inequality were obtained replacing \(\ell_\infty^n\) by other finite-dimensional Banach spaces, as e.g. finite-dimensional \(\ell_p\)'s or Orlicz norms. Originally initiated in the following three articles, so far basically three approaches were considered: An approach through Khinchin-type inequalities for Rademacher processes due to \textit{H. P. Boas} [J. Korean Math. Soc. 37, No. 2, 321--337 (2000; Zbl 0965.32001)], a Gaussian approach through Slepian's lemma by \textit{A. Defant} et al. [J. Reine Angew. Math. 557, 173--197 (2003; Zbl 1031.46014)], and the entropy approach of \textit{F. Bayart} [Q. J. Math. 63, No. 1, 21--39 (2012; Zbl 1248.46033)] which is in a sense the most successful one. The article of Mastyło and Szwedek extends Bayart's approach substantially in order to prove concrete upper estimates for \(\mathbb{E} \|P\|_\infty\) if the sup norm is taken on the unit ball of an arbitrary Banach space. Its tools are mainly based on methods from stochastic processes and interpolation theory in Banach spaces. The key point is to control the increments of Rademacher processes in an Orlicz space via entropy integrals. The derived estimates are used to get lower estimates for unconditional basis constants of spaces of \(m\)-homogeneous polynomials on finite-dimensional Banach spaces, and in turn upper estimates for Bohr radii of the unit balls of such spaces.