A decrease in growth of Lebesgue constants (Q1364820)

Let \(S_N\) denote the cubic partial sums of the Fourier series of \(n\) variables, i.e., \[ S_N(f)(x)= \sum_{k:|k_j|\leq N,j=1,\dots,n}\widehat f(k)e^{ikx}, \] where \(f\in L(\mathbb{T}^n)\) and \(\widehat f(k)\) denotes its \(k\)th Fourier coefficient. It is well known that the norms of \(S_N\) from \(C(\mathbb{T}^n)\) to \(C(\mathbb{T}^n)\) or \(L(\mathbb{T}^n)\) to \(L(\mathbb{T}^n)\) which are usually called Lebesgue constants have asymptotically the order \(\log^nN\) as \(N\to\infty\). \textit{A. A. Kivinukk} [Mat. Zametki 20, 597-604 (1976; Zbl 0351.42002)] investigated the linear means \[ L^\lambda_N(f)(x):= \sum_{k:|k_j|\leq N,j=1,\dots,n}\lambda\Biggl({k\over N}\Biggr) \widehat f(k)e^{ikx}, \] where \[ \lambda(x)=\begin{cases} 1-|x_1\cdots x_n|^\gamma, & |x_j|\leq 1,\;j=1,\dots,n;\\ 0 &\text{otherwise},\end{cases} \] with \(\gamma>0\). And he proved that the asymptotic order of the norms of \(L^\lambda_N\) from \(C(\mathbb{T}^n)\) to \(C(\mathbb{T}^n)\) or \(L(\mathbb{T}^n)\) to \(L(\mathbb{T}^n)\) decreases to \(\log^{n-1}N\). In the present paper, the author generalized the result of Kivinukk by setting \[ \lambda(x)=\begin{cases} (1-|x_1|^{\gamma_1}\cdots|x_n|^{\gamma_n})^\alpha, & |x_j|\leq 1,\;j=1,\dots,n;\\ 0 &\text{otherwise}\end{cases} \] with \(\gamma_j>0\), \(j=1,\dots,n\), \(\alpha>0\).