A generalization of Pisier homogeneous Banach algebra (Q6108381): Difference between revisions

Let \((Z_n)_{n\in \mathbb Z}\) be a sequence of independent standard Gaussian random variables. \textit{G.~Pisier} [Isr. J. Math. 34, 38--44 (1979; Zbl 0428.46035)] proved that the set \(\mathscr P\) of continuous functions on the unit circle \(\mathbb T\) for which the process \(X_t= \sum_n \widehat f(n) Z_n e^{2\pi i nt}\) has a.s. bounded sample paths (equivalently, the series converges a.s. uniformly) is a homogeneous Banach algebra lying strictly between the Wiener algebra and \(C(\mathbb T)\) and on which Lipschitz functions operate. The idea of the present paper is to suggest a similar construction for a complex stationary Gaussian process \((\xi_n)\) with spectral measure \(\mu\) in place of \((Z_n)\), and the author proves by ``a replication of Pisier's original argument'' (p.~246) that one then obtains a homogeneous Banach algebra \(\mathscr P(\mu)\). (The main point here is that the \(\xi_n\) no longer need to be independent.) He goes on to give sufficient conditions for the inclusion \(\mathscr P(\mu) \subset \mathscr P\), and he discusses the example of fractional Gaussian noise in detail, solving a problem from his paper [\textit{S.~Mukeru}, Lith. Math. J. 61, No.~1, 60--68 (2021; Zbl 1479.46064)]. In the final section, a sufficient condition for the a.s. boundedness of the random function \(t\mapsto \sum_n \widehat f(n) \xi_n e^{2\pi i nt}\) is proved.