The sets of abstract trigonometric series induced by \(L^p\) (Q1578468): Difference between revisions

Let \(\lambda\) be a rate and \(T=(t_{nk})\) be a summability method. A series \(\sum_{k\in \mathbb{Z}}u_k\), where \(\mathbb{Z}\) is the additive abelian group of integers, is called \(T^\lambda\)-summable if \[ \lim_n\sum_{|k |\leq n}t_{n|k|}u_k =t\text{ exists} \] and \[ \lambda_n \left|\sum_{|k|\leq n}t_{n|k|}u_k- t\right|=O(1), \] and is said to be \(T^\lambda_0\)-summable if \[ \lambda_n\left( \sum_{|k|\leq n} t_{n |k|}u_k\right) =O(1). \] The sequence \(\varepsilon= (\varepsilon_k)_{k \in\mathbb{Z}}\) is called a summability factor of class \((T^\lambda, U^\mu)\) if for every \(T^\lambda\)-summable series \(\sum_{k\in \mathbb{Z}}u_k\), the series \(\sum_{ k\in \mathbb{Z}}\varepsilon_k u_k\) is \(U^\mu\)-summable, written \(\varepsilon\in (T^\lambda, U^\mu)\). The classes \((T^\lambda, U_0^\mu)\), \((T_0^\lambda, U^\mu)\) and \((T^\lambda_0, U_0^\mu)\) are defined similarly. Earlier the author [Acta Comment. Univ. Tartu. Math. 2, 61-65 and 67-73 (1998; Zbl 0951.42003 and Zbl 0951.42004)] studied the \(T^\lambda\)-constructive spaces \(L^p_{T \lambda}\) and \({\mathcal L}^p_{T\lambda}\). He continues his studies in the present paper to prove the following: Theorem 1. Let \(1/p+1/q=1\) for \(p,q>1\) and \(g \in L^q\) with \(g^0(x)= \sum_{k\in\mathbb{Z}} d_ke^{ikx}\). Then \(f^0\in {\mathcal L}^p_{T \lambda}\), where \(f^0(x)= \sum_{k\in \mathbb{Z}}c_k e^{ikx}\) is the Fourier series of \(f\in L^p\), if and only if for every \(g\in L^q\) \[ \sum_{k\in \mathbb{Z}}c_k d_{-k}\tag{1} \] is \(T^\lambda_0\)-summable. Theorem 2. Let \(\lambda\) be a monotonic rate, \(1/p+1/q=1\) for \(p>1\), and let \(T=(t_{nk})\) be the series-to-sequence matrix which, for all \(f\in L^p\), satisfies \[ \lim_n \bigl\|t_n(f)-f\bigr \|_p =0,\tag{2} \] where \[ t_n(f;x)=\sum_{|k|\leq n}t_{n|k|} e^{ikx}. \] Then \(f\in L^p_{T\lambda}\) if and only if, for every \(g\in L^q\), the series (1) is \(T^\lambda\)-summable. Further, the author investigates norms for the spaces \(L^p_{T\lambda}\) with monotonic rate and \({\mathcal L}^p_{T\lambda}\) to be Banach and normed spaces respectively. Let \(\widehat X\) be the space of Fourier coefficients of \(X\). Then a sequence \(\varepsilon= (\varepsilon_k)_{k\in\mathbb{Z}}\) is called a multiplier of class \((X,Y)\) if for every \((c_k)_{k\in\mathbb{Z}}\) in \(X\), the sequence \((\varepsilon_k c_k)_{k\in \mathbb{Z}} \in\widehat Y\), written \(\varepsilon\in (X,Y)\). The author has also characterised the \((L^p,L^p)\)-multipliers and multipliers of the classes \(({\mathcal L}^p_{T\lambda}, {\mathcal L}^p_{U\mu})\), \((L^p_{T\lambda}, {\mathcal L}^p_{U\mu})\), \((L^p_{T\lambda}, L^p_{U\mu})\), and \(({\mathcal L}^p_{T\lambda}, L^p_{U\mu})\). One of his results reads as follows: Theorem 3. Let \(p,p_1\in (1,\infty)\), \(1/p+ 1/q=1\), \(1/p_1+ 1/q_1=1\). If \(\varepsilon\in (L^q,L^{q_1})\) then \(\varepsilon \in({\mathcal L}^{p_1}_{T\lambda}, {\mathcal L}^p_{T\lambda})\). Further, if \(\lambda\) is monotonic and \(T\) satisfies (2) then also \(\varepsilon\in (L_{T\lambda}^{p_1}, L_{T\lambda}^p)\). Remark. Theorem 3 may be compared with Theorem 1 of the author's paper [Acta Comment. Univ. Tartu. Math. 2, 67-73 (1998; Zbl 0951.42004)].