Summability of Fourier-Laplace series with the method of lacunary arithmetical means at Lebesgue points (Q5946548): Difference between revisions

The problem of how ``rare'' can be the sequence of integers \(\{n_k\}\) so that \({1\over N}\sum_{k=1}^N S_{n_k}(f;x)\to f(x)\) as \(N\to\infty\) was posed by Z. Zalcwasser in 1936. Here \(S_{n_k}(f;x)\) denotes the \(n_k\)th partial sum of the trigonometric Fourier series of a periodic function \(f\in L_p,\) \(p\geq 1.\) There exist several versions of this problem as regarded to the type of convergence. The problem is completely solved for uniform convergence, that is, for continuous functions. More precisely, for a convex sequence \(\{n_k\}\) the condition \(\sup_N N^{-1/2}\log n_N<\infty\) is necessary and sufficient. Much more problematic is the convergence at Lebesgue points (for integrable functions). The above-mentioned condition is also necessary in this case, but no criterion is known. Only recently \textit{E. S. Belinsky} [Proc. Am. Math. Soc. 125, No. 12, 3689-3693 (1997; Zbl 0907.42004)] gave a partial solution in this case, namely proved the existence of such ``rare'' sequence. The proof is based on a statement due to J. Bourgain, where the idea of random choice is exploited. In the paper under review this result is extended to the case of spherical harmonics in the even-dimensional case. The proof is similar; to apply it, some special properties of spherical harmonics expansions are used.