Fourier-Haar coefficients and Banach limits (Q735825)

Let us denote by \(ac\) the set of almost converging sequence: that is the set of real sequences \(x\in\ell^\infty\) such that there exists some \(a\in{\mathbb R}\) with \(B(x)=a\) for every Banach limit \(B\). Equivalently, \(\displaystyle\lim_{n\to\infty}\frac{1}{n}\sum_{k=m+1}^{m+n}x_k=a\), uniformly in \(m\in{\mathbb N}\). In this case, we write \(Lim(x)=a\). In the paper under review (which contains no proof), the authors are interested in the asymptotic behavior of Fourier coefficient of an integrable function with respect to the Haar system. Do these coefficient (almost) converge to \(0\)? They state a positive answer for functions \(f\) belonging to \(L^{p,\infty}\), which is the space of integrable \(f\) on \((0,1)\) such that \[ \sup_{\lambda(E)>0}\frac{1}{\lambda(E)^{1/p'}}\int_E|f|<\infty \] (endowed with the obvious norm) where \(\lambda\) is the Lebesgue measure and \(p'\) is the conjugate exponent of \(p\). Their main result is Theorem 2: For every \(1<p<\infty\) and every \(f\in L^{p,\infty}\), the Fourier coefficients \(c_m(f)\) of \(f\) with respect to the Haar system verifies: \[ m^{1/2-1/p}c_m(f)\in ac\, {\text{and}} \, Lim\Big(m^{1/2-1/p}c_m(f)\Big)=0. \] This result is optimal in the sense that when \(f\in L^{2,\infty}\), the coefficients are bounded but do not converge to \(0\) in general. Some examples are presented at the end of the paper.