On wavelet polynomials and Weyl multipliers (Q6081432)

The paper under review provides new growth estimates for \(1<p<\infty\) and for wavelet type orthonormal systems \(\Phi:=\{\phi_n\in L^2([0,1]): n=1,2,\dots\}\) of the form \[ \left\| \max_{1\leq m\leq n}\Big | \sum_{j\in G_m}\langle f,\phi_j\rangle\phi_j\Big |\,\right\|_p \lesssim_p \sqrt{\log (n+1)}\, \|f\|_p,\tag{1} \] where \(G_m\subset \mathbb{N}\) are arbitrary finite sets of indexes. These estimates are the key to showing that the sequence \(\{\log n: N=1,2,\dots\}\) is an almost everywhere convergence Weyl multiplier (optimal in this context) for any orthonormal system of non-overlapping wavelet polynomials. Recall that given \(\Phi\) as above, a sequence of positive numbers \(\{\omega(n): n=1,2,\dots\}\) increasing to infinity is an a.e. convergence Weyl multiplier if every series \(\sum_{n=1}^{\infty} a_n\phi_n(x)\) with coefficients satisfying the growth condition \(\sum_{n=1}^{\infty} a_n^2\omega (n)<\infty\), is a.e. convergent. If the property persists after any rearrangement of the terms, the sequence is an a.e. unconditional convergence Weyl multiplier for \(\Phi\). The main result (1) parallels the classic Menshov-Rademacher theorem ( references \textit{D. Menchoff} [Fundam. Math. 4, 82--105 (1923; JFM 49.0293.01)], \textit{H. Rademacher} [Math. Ann. 87, 112--138 (1922; JFM 48.0485.05)] in the paper) where a similar estimate in \(L^2([0,1])\) is shown to hold for all orthogonal systems with \(\log{(n+1)}\) instead of \(\sqrt{\log{(n+1)}}\) and a counterexample proving optimality of the growth factor is presented [Menchoff, loc. cit.]. The question then is can one obtain better growth order for specialized classes of orthonormal systems? Here the complete answer is presented for an orthonormal system of wavelet type whose size and smoothness conditions are governed by the function \(\xi(x)=(1+|x|)^{-1-\delta}\) for \(0<\delta<1\). Examples of such wavelet-type systems are classical wavelets, and orthonormal spline wavelets (in particular the Franklin system). Some results can be established for periodic wavelet systems and biorthogonal systems. Although the Haar system does not satisfy the smoothness condition, nevertheless the estimate can be shown to hold bypassing a lemma and instead using the Haar square function. Prior results for the Haar and trigonometric systems were studied by the second author [Ark. Mat. 58, No. 1, 121--131 (2020; Zbl 1439.42037); Sb. Math. 211, No. 12, 1704--1736 (2020; Zbl 1458.42008); translation from Mat. Sb. 211, No. 12, 49--82 (2020); Russ. Math. Surv. 75, No. 3, 569--571 (2020; Zbl 1447.42027); translation from Usp. Mat. Nauk 75, No. 3, 183--184 (2020)].