A characterization of multidimensional multi-knot piecewise linear spectral sequence and its applications (Q383618)

Let \(\mathbb{Z}\) and \(\mathbb{N}\) be the sets of all the integers and all the positive integers, respectively. Let \(N_n:=\{1,\,\ldots,\,n-1\}\) for each \(n\in\mathbb{N}\). For each fixed vector \[ \vec{p}:=(p_0,\ldots,\,p_{d-1})\in\mathbb{N}^d, \] let \(\nu_{\vec{p}}:=\prod_{k\in N_d}p_k:=p_0\cdots p_{d-1}\) and \[ N_{\vec{p}}:=\prod_{k\in N_d}N_{p_k}:=N_{p_0}\times\cdots\times N_{p_{d-1}}. \] Suppose that \(\pi_{\vec{p}}\) is a mapping from \(N_{\vec{p}}\) to \(N_{\nu_{\vec{p}}}\) defined by \[ \pi_{\vec{p}}(\vec{j}):=\sum_{k\in N_d}\left(\prod_{s\in N_k}p_s\right)j_k \] for any \(\vec{j}:=(j_0,\,\ldots,\,j_{d-1})\in N_{\vec{p}}\), where \(N_0:=\emptyset\) and \(\prod_{s\in\emptyset}p_s:=1\). The authors obtain a partition of \(I^d:=[0,1)^d\) as follows: \(I^d=\cup_{j\in N_{\nu_{\vec{p}}}} I^{\vec{p}}_j\), where, for each \(j\in N_{\nu_{\vec{p}}}\) and \(\vec{j}:=\pi^{-1}_{\vec{p}}(j) =(j_0,\,\ldots,\,j_{d-1})\in N_{\vec{p}}\), \(I^{\vec{p}}_j :=\prod_{k\in N_d} I^{p_k}_{j_k}\) and, for each \(k\in N_d\) and \(j_k\in N_{p_k}\), \(I^{p_k}_{j_k}:=[\frac{j_k}{p_k},\frac{j_k+1}{p_k})\). The authors further introduce the piecewise linear spectral sequences with \(\vec{p}\)-knots on \(I^d\). Precisely, for each \(\vec m=(m_0,\,\ldots,\,m_{d-1})\in\mathbb{Z}^d\) and \(k\in N_d\), \(m_k\) is presented as \(m_k:=p_k\widetilde m_k+l_k\), where \(\widetilde m_k\in\mathbb{Z}\) and \(l_k\in N_{p_k}\). Let \(\widetilde{\vec{m}} :=(\widetilde{m}_0,\,\ldots,\,\widetilde{m}_{d-1})\in \mathbb{Z}^d\), \(\vec{l}:=(l_0,\,\ldots,\,l_{d-1})\in N_{\vec{p}}\) and \(l:=\pi_{\vec{p}}(\vec{l})\). For any vectors \[ \vec{x}:=(x_0,\,\ldots,\,x_{d-1}),\quad \vec{y}:=(y_0,\,\ldots,\,y_{d-1})\in\mathbb{R}^d, \] the authors define \(\vec{x}\cdot\vec{y}:=\sum_{k=0}^{d-1}x_ky_k\), \(\vec{x}\vec{y}:=(x_0y_0,\,\ldots,\,x_{d-1}y_{d-1})\) and \[ \frac{\vec{x}}{\vec{y}}:=\left(\frac{x_0}{y_0},\,\ldots,\, \frac{x_{d-1}}{y_{d-1}}\right), \] where, for any \(i\in\{0,\,\ldots,\,d-1\}\), \(y_i\neq 0\). By these notation, the authors further consider the sequence of multi-knots piecewise linear functions \(g_{\vec{m}}\) (\(\vec{m}\in\mathbb{Z}^d\)), which have the form \[ g_{\vec{m}}(\vec{x}):=g_{\widetilde{\vec{m}},\,l}(\vec{x}):= (\vec{p}\widetilde{\vec{m}}+\vec{c}_l)\cdot\vec{x} +b^j_{\widetilde{\vec m},\,l}, \quad \vec{x}\in I^{\vec{p}}_j \;\text{and} \;j\in\{0,\ldots,\nu_{\vec{p}}-1\}, \] where \(b^j_{\widetilde{\vec m},\,l}\in\mathbb{R}\), \(\vec{c}_{l}\in\mathbb{R}^d\), \(\widetilde{\vec{m}}\in\mathbb{Z}^d\), \(l,\,j\in N_{\nu_{\vec{p}}}\), satisfy the following conditions. (i) For any \(l\neq s\) and \(k\in N_d\), the \(k\)-th coordinate of the vector \(\vec{c}_l-\vec{c}_s\), \((\vec{c}_l-\vec{c}_s)_k\), is not in \(p_k\mathbb{Z}\backslash\{0\}\). (ii) There exists a sequence \(\{\alpha_{l,\,j}\}_{l,\,j\in N_{\nu_{\vec p}}}\subset \mathbb{R}\) such that \[ b^0_{\widetilde{\vec m},\,l}-b^j_{\widetilde{\vec m},\,l} \in\mathbb{Z}+\vec{c}_l\cdot\frac{\vec{j}}{\vec{p}}+\alpha_{l,\,j} \] holds true for any \(\widetilde{\vec m}\in\mathbb{Z}^d\) and \(l,\,j\in N_{\nu_{\vec{p}}}\), where \(\vec{j}:=\pi^{-1}_{\vec{p}}(j)\). The authors further define functions \(f_{\vec{m}}:=e^{2\pi i g_{\vec{{m}}}}\), \(\vec{m}\in\mathbb{Z}^d\), \(\mathcal{F}:=\{f_{\vec{m}}\}_{\vec{m}\in\mathbb{Z}^d}\) and \(\mathcal{G}:=\{g_{\vec{m}}\}_{\vec{m}\in\mathbb{Z}^d}\). \(\mathcal{G}\) is called a spectral sequence of \(I^d\) if \(\mathcal{F}\) forms an orthonormal basis for \(L^2(I^d)\). Suppose that the \(g_{\vec{m}}\), \(\vec m\in\mathbb{Z}^d\), satisfy the conditions (i) and (ii) as above. The authors prove that \(\mathcal{G}\) is a spectral sequence on \(I^d\) if and only if the matrix \[ \mathcal{A} :=\frac1{\sqrt{\nu}_{\vec{p}}}\left(e^{2\pi i\alpha_{l,\,j}}\right)_{l,\,j \in N_{\nu_{\vec p}}} \] is a unitary matrix. As applications, the authors investigate the convergence both in the pointwise sense and in the Lebesgue spaces of the generalized rectangular partial summation and the generalized Bochner-Riesz means of the generalized Fourier series.