Weak type inequality for logarithmic means of Walsh-Kaczmarz-Fourier series (Q541394): Difference between revisions

The Walsh functions \(w_0(x), w_1(x),\dots\) are defined on \(I=[0,1]\) by \(w_0=1\) and \(w_k(x)=r_{n_1}(x)\cdots r_{n_s} (x)\) where \(k=2^{n_1}+\cdots + 2^{n_s}\) with \(n_1> \cdots >n_s\geq 0\) where \(r_0(x)=1\) if \(0\leq x<1/2\), \(r_0(x)=-1\) if \(1/2\leq x<1\), and \(r_0 (x+1)=r_0(x)\), and \(r_n(x)=r_0(2^n x)\). The Walsh-Kaczmarz functions are defined by \(\kappa_0(x)=1\) and, for \(n\geq 1\), \(\kappa_n(x) =r_{n_1}(x) \prod_{k=0}^{n_1-1} (r_{n_1-1-k}(x))^{n_k}\). One defines Dirichlet kernels \(D_n^\alpha (x) = \sum_{k=0}^{n-1} \alpha_k (x)\) where \(\alpha_k=w_k\) or \(\kappa_k\). The Walsh(-Kaczmarz) coefficients are defined by \(\hat{f}^\alpha(n)=\int_I f\, \alpha_n\) and the \(n\)th partial sum of the Walsh(-Kaczmarz) series is \(S_n^\alpha f(x) = \sum_{k=0}^{n-1} \hat{f}^\alpha (k) \alpha_k (x)\). The Nörlund logarithmic means and kernels of the Walsh(-Kaczmarz) series are defined via \[ t_n^\alpha (f,x) = \frac{1}{\ell_n} \sum_{k=1}^{n-1} \frac{S_k^\alpha (f,x)}{n-k};\qquad F_n^\alpha (t) = \frac{1}{\ell_n} \sum_{k=1}^{n-1} \frac{D^\alpha (f,x)}{n-k}, \] where \(\ell_n=\sum_{k=1}^{n-1} 1/k\). The rectangular partial sums of double series with respect to Walsh(-Kaczmarz) systems on \(I^2\) are defined by \[ S_{mn}^\alpha (f,x,y) = \sum_{i=0}^{m-1}\sum_{j=0}^{n-1} \hat{f}^\alpha (i,j) \alpha_i(x) \alpha_j (y) \] with corresponding suitably defined logarithmic means, also denoted \(t_{mn}^\alpha(f,x,y)\). With \(|S|\) denoting the Lebesgue measure of a subset \(S\subset I\), one has that \(t_n^\alpha \) is weak-type \((1,1)\), that is, for any \(\lambda>0\) and \(f\in L^1(I)\) one has \[ \lambda |x\in I: |t_n^\kappa (f,x)|>\lambda|\leq c \|f\|_1. \] As a corollary, if \(0<p<1\) and \(f\in L\ln^{+}L (I^2)\) then \[ \biggl(\int_{I^2} |t_{mn}^\kappa (f,x,y)|^p\, dx dy\biggr)^{1/p} \leq c_p \int_{I^2} |f(x,y)| \ln^+ |f(x,y)|\, dx \, dy+c_p \] and \[ \int_{I^2} |t_{mn}^\kappa (f,x,y)-f(x,y)|^p\to 0\quad (m,n\to\infty). \] Similarly, it is shown that if \(f\in L\ln^+ L(I^2)\) then \[ |(x,y)\in I^2: |t_{mn}^\kappa (f,x,y)|>\lambda|\leq \frac{c}{\lambda} \int_{I^2} |f(x,y)|\ln^+ |f(x,y)|\, dx \, dy +c \] and that \(t_{mn}^\kappa (f,x,y)\) converges to \(f(x,y)\) in measure on \(I^2\) as \(m,n\to\infty\). On the other hand, it is shown that if \(L_\Phi(I^2)\) is an Orlicz space that is not contained in \(L \ln^+ L (I^2)\) then the set of functions \(f\) from \(L_\Phi(I^2)\) having quadratic logarithmic means in the Walsh-Kaczmarz system converging to \(f\) in measure on \(I^2\) is of first Baire category in \(L_\Phi (I^2)\). In particular, if \(\varphi:[0,\infty)\to [0,\infty)\) is a nondecreasing function satisfying \(\varphi(x)=o(x\log x)\) as \(x\to\infty\) then there exists a function \(f\in L^1(I^2)\) and \(\int_{I^2} \varphi(|f(x,y)|)\, dx dy<\infty\) but having quadratic logarithmic means of the Walsh-Kaczmarz series of \(f\) diverging in measure on \(I^2\).