Strong convergence theorems for two-parameter Vilenkin-Fourier series (Q1586362)

Let \(m:= (m_k , k\in {\mathbb N})\) \(({\mathbb N}:= \{0,1,\dots \})\) be a sequence of integers each of them not less than \(2\). Let \(Z_{m_k}\) denote the \(m_k\)-th discrete cyclic group. \(Z_{m_k}\) can be represented by the set \(\{0,1,\dots,m_k-1\}\) , where the group operation is the \(\mod m_k\) addition and every subset is open. Let \[ G_m := \overset{\infty} {\underset{k=0} {\times}} Z_{m_k}. \] Consequently, \(G_m\) is a compact Abelian group. Let \(M_0 :=1 , M_{n+1} := m_nM_n (n\in {\mathbb N}).\) Then each natural number \(n\) can be uniquely expressed as \[ n = \sum _{i=0}^{\infty} n_iM_i \quad (n_i \in \{0,1,\dots,m_i-1\},\quad i\in {\mathbb N}). \] The generalized Rademacher functions are defined as \[ r_n(x) := \exp \left(2\pi \imath \frac{x_n}{m_n}\right)\quad (x\in G_m , n\in {\mathbb N}, \imath := \sqrt{-1}). \] The \(n\)th Vilenkin function is \[ \psi _n := \prod _{j=0}^{\infty} r_j^{n_j} \quad (n \in {\mathbb N}) . \] The system \({\mathbb\psi} := (\psi _n : n\in {\mathbb N})\) is called a Vilenkin system. Define \(G:= G_m \times G_m\) and \(\psi_{k,l}:= \psi_k\psi_l\) the two-dimensional Vilenkin group and functions, respectively. If \(f\in L^1(G)\) then \[ \widehat f(k,l) := \int _{G}f\overline \psi_ {k,l}\quad (k,l\in \mathbb N), \] the (two-dimensional) Fourier coefficients, \[ S_{n,l}f:=\sum_{k=0}^{n-1}\sum_{j=0}^{l-1}\widehat f(k,j)\psi_{k,j}, \] the \((n,l)\)-th partial sum of the Fourier series of \(f\). It is well known that the Walsh-Paley system (\(m_k=2\) for each \(k\in\mathbb N\)) does not form a basis in \(L^1\). However, \textit{P. Simon} proved [Acta Math. Hung. 49, 425-431 (1987; Zbl 0643.42020)] the following strong convergence theorem for all functions \(f\) in the Hardy space \[ \lim_{n\to \infty}\frac{1}{\log n}\sum_{k=1}^{n}\frac{\|S_kf-f\|_1}{k} =0. \] This result was generalized for the Vilenkin systems by \textit{G. Gát} [Acta Math. Hung. 61, No. 1-2, 131-149 (1993; Zbl 0805.42019)] and for the two-parameter Walsh-Fourier series by \textit{F. Weisz} [Stud. Math., 117, No. 2, 173-194 (1996; Zbl 0839.42009)]. In the present paper -- among others -- the authors prove the following \[ \frac{1}{\log N \log M}\sum _{N,M}^{\alpha, \beta}\frac{\|S_{k,l}f-f\|_1}{kl} \to 0 \quad (N,M\to \infty). \]