On the partial sums of Vilenkin-Fourier series (Q736392): Difference between revisions

Let \(m=(m_0,m_1,\dots)\) be a sequence of positive integers such that \(m_k\geq 2\). By \(\mathbb Z_{m_k}=\{0,1,\dots,m_k-1\}\) we denote the cyclic discrete group of order \(m_k\) with measure \(\mu_k(\{j\})=1/m_k\). Then \(G_m=\prod^\infty_{k=0}\mathbb Z_{m_k}\) is defined as the direct product of \(Z_{m_k}\) with the corresponding measure \(\mu\) and topology. The elements of \(G_m\) are \(x=(x_0,x_1,\dots)\), \(x_j\in \mathbb Z_{m_j}\) and \(I_n(x)=\{y\in G_m:x_0=y_0,x_1=y_1, \dots,x_{n-1}=y_{n-1}\}\), \(n\in\mathbb N\). Let \(M_0=1\), \(M_{k+1}=m_kM_k\), \(k\in\mathbb Z_+\), and \(r_k(x)=\exp(2\pi ix_k/m_k)\). For \(n=\sum^\infty_{j=0}n_jM_j\), \(n_j\in\mathbb Z_{m_j}\), we set \(\psi_n(x)=\prod^\infty_{k=0}r_k^{n_k}(x)\). The \(\sigma\)-algebra generated by \(\{I_n(x):x\in G_m\}\) will be denoted by \(F_n\). If \(f=(f_n, n\in\mathbb Z_+)\) is a martingale with respect to \(\{F_n\}_{n=1}^\infty\), then we define the maximal function of \(f\) by \(f^*(x)=\sup_{n\in\mathbb Z_+}|f_n(x)|\) and Hardy spaces \(H_p(G_m)=\{f:\|f\|_{H_p}=\|f^*\|_{L_p(G_m)}<\infty\}\), \(0<p<\infty\). The weak \(L_p\) space \(L_{p,\infty}(G_m)\) consists of all measurable functions \(f\) for which \(\|f\|_{L_{p,\infty}}=\sup_{\lambda>0}\lambda(\mu(|f|>\lambda))^{1/p}<\infty\). If \(f=(f_n, n\in\mathbb Z_+)\) is a martingale, then \[ \hat{f}(i)=\lim\limits_{k\to\infty}\int_{G_m}f_k(x)\overline{\psi_i(x)}\,d\mu(x) \quad \text{and}\quad S_n(f)(x)=\sum^{n-1}_{k=0}\hat{f}(i)\psi_i(x). \] For \(0<p\leq 1\) let \(\widetilde{S}^*_p(f):=\sup_{n\in\mathbb Z_+}|S_n(f)|/[(n+1)^{1/p-1}\log^{[p]}(n+1)]\). The main results of paper are Theorem 3.1. a) Let \(0<p\leq 1\). Then the maximal operator \(\widetilde{S}^*_p\) is bounded from \(H_p(G_m)\) to \(L_p(G_m)\). b) Let \(0<p\leq 1\) and \(\varphi:\mathbb Z_+\to [1,\infty)\) be a nondecreasing function such that \[ \limsup\limits_{n\to\infty}(n+1)^{1/p-1}\log^{[p]}(n+1)/\varphi(n)=+\infty. \] Then \(\sup_{n\in\mathbb Z_+}\|s_n(f)/\varphi(n)\|_{L_{p,\infty}}=\infty\) for \(0<p<1\) and \(\sup_{n\in\mathbb Z_+}\|s_n(f)/\varphi(n)\|_{L_1}=\infty\) for \(p=1\). Theorem 3.3. a) Let \(0<p<1\), \(f\in H_p(G_m)\) and \(\|f-S_{M_n}(f)\|_{H_p}=o(M_n^{1-1/p})\), \(n\to+\infty\). Then \(\|S_k(f)-f\|_{L_{p,\infty}}\to 0\) as \(k\to\infty\). b) For every \(p\in (0,1)\) there exists a martigale \(f\in H_p(G_m)\) such that \(\|f-S_{M_{2n}}(f)\|_{H_p}=O(M_{2n}^{1-1/p})\) but \(\|S_k(f)-f\|_{L_{p,\infty}}\) does not converge to zero. Reviewer's remark. There are many misprints and incorrect formulations in the paper. In the definition of \(L_{p,\infty}\) norm (page 25, 11 line from below) \(|f|\) must be instead of \(f\), while in definition of \(f^*\) there must be \(f_n\) instead of \(f^{(n)}\) (page 25, 7 line from below). In the definition of the \(p\)-atom, the interval \(I\) has the form \(I_n(x)\) and it is not dyadic (page 26, 6 line from above). On page 28 one can see references to Lemmas 1 and 2 while in the text there are only Lemmas 4.1 and 4.2. In ref. [13] must be ``Guličev'', references [22] and [24] are the same and in both of them there is no volume number indicated. On page 29 there are references to (3.12), (3.13) and (3.14) but these formulas are absent in the text. The expression \(M_{2k}=(\mathrm{supp}\, a_k)^{-1/p}\) on page 30, 7 line from below, is wrong since \(\mathrm{supp}\, a_k\) is a set. The last line in (5.11) is also incorrect. The necessity of the condition of boundedness of \(\{m_k\}_{k=0}^\infty\) for the results is not clear. For example, inequality (5.5) is valid only under this condition. However, in the reviewer's opinion, the results of the paper are true if \(\{m_k\}_{k=0}^\infty\) is bounded.