Some saturation theorems for classical orthogonal expansions. II (Q1187310)

A theorem of \textit{G. Alexits} [Acta Math. Acad. Sci. Hung. 3, 29--40 (1952; Zbl 0047.06901)] says that the saturation class in \(C\)-metric of Fejér means of single Fourier series is the set \(\{f\in C_{2\pi}:\tilde f\in \hbox{Lip} 1\}\), where \(\tilde f\) denotes the conjugate function of \(f\). The main purpose of the present paper is to extend this theorem to Laguerre expansions of nonnegative parameter. Let \(\alpha>-1\). Define \(u_\alpha(x)=x^\alpha e^{-x}\), \(x\in(0,\infty)\). The normalized Laguerre polynomials \(l_ n^{(\alpha)}\) of order \(\alpha\) are defined by the orthogonality condition: \(\int^\infty_0 l_n^{(\alpha)}l_k^{(\alpha)}u_\alpha=\delta_{n,k}\), \(n,k=0,1,2,\ldots\). The Laguerre-Fourier series of the function \(f\) is defined by \(f\sim\sum^ \infty_{k=\infty}a_k l_k^{(\alpha)}\), \(a_k=\int^\infty_0 fl_k^{(\alpha)}u_\alpha\). The Riesz means of parameter \(1/2\) of the expansion of \(f\) is the sum \(R_nf=\sum^n_{k=0}(1- \sqrt{k/(n+1)})a_k l_k^{(\alpha)}\). Following \textit{B. Muckenhoupt} [Trans. Am. Math. Soc. 147, 403--418 (1970; Zbl 0192.46202)], the conjugate Poisson integral of \(f\) is defined by \[ \tilde f(x,y)=\int^ \infty_ 0q(x,y,z)f(z)u_ \alpha(z)\,dz\quad (x,y>0), \] where \(q(x,y,z)=\sqrt y\sum^ \infty_{n=1}e^{-\sqrt n x}l_{n- 1}^{(\alpha+1)}(y) l_n^{(\alpha)}(z)\). Then the conjugate function \(\tilde f\) of \(f\) is defined by \(\tilde f(y)=\lim_{x\to 0^+}\tilde f(x,y)\;(y>0)\) if the limit exists. The author establishes several Alexits type theorems for Laguerre expansion. His Theorem 3 reads as follows: If \(\alpha>0\), \(1<p<\infty\) and \(\int^\infty_0(f\sqrt{u_\alpha})^p<\infty\) then the following statements are equivalent: \[ \| \sqrt{u_ \alpha}(f-R_nf)\|_p = O(1/\sqrt{n}), \] \[ \tilde f \text{ is locally absolutely continuous and } \int^ \infty_ 0\{(u_{\alpha+1/2}\tilde f)'u_{\alpha}^{- 1/2}\}^ p<\infty. \] [For Part I see the review Zbl 0749.42013 above.]