Realization of the spline-wavelet decomposition of the first order (Q2405041)

This paper is on the decomposition and reconstruction of signals defined on a grid using discrete spline-wavelets. Let \((\alpha, \beta)\subset \mathbb{R}^1\) be an interval and \(\mathbb{L}\) be the linear function space defined on the interval \((\alpha,\beta)\). Let \(X:= \cdots < x_{-2} < x_{-1} < x_0 < x_1 < x_2 < \cdots\) be a grid inside the interval \((\alpha,\beta)\). Given \(m\in\mathbb{N}\), by considering a set \(\{g^{(s)}\}_{s\in\mathbb{Z}}\subset \mathbb{L}^*\) of functionals on \(\mathbb{L}\), the paper builds a basis \(\{\omega_s\}_{s\in\mathbb{Z}}\subset \mathbb{L}\) so that \[ \langle g^{(j)}, \omega_s\rangle = \delta_{j,s}, \quad j,s\in\mathbb{Z}, \] where \(\mathrm{supp} \omega_s \subset [x_s, x_{s+m+1}]\), \(s\in\mathbb{Z}\). A signal \(\mathbf{c} :=(\dots, c_{-2}, c_{-1}, c_0, c_1, c_2, \ldots)\) defined on the grid \(X\) can then be associated with as an element \(u\) in \(\mathbb{L}\) by \[ u(t):=\sum_j c_j \omega_j(t), \quad t\in (\alpha,\beta). \] Consider a coarse grid \(\tilde{X}:= \cdots < \tilde x_{-2} < \tilde x_{-1} <\tilde x_0 < \tilde x_1 < \tilde x_2 < \cdots\), which is a subset of the original grid \(X\). A similar approach can be done to obtain a basis (coarser) \(\{\tilde \omega_s\}_{s\in\mathbb{Z}}\subset \mathbb{L}\) so that \(u\) can be projected onto its spanned linear space that results in \[ \tilde u(t) =\sum_j a_j \tilde \omega_j(t), \quad t\in(\alpha, \beta). \] The difference \(w(t):=u(t)-\tilde u(t)\) can be represented under the original basis by \[ w(t) = \sum_j b_j \omega_j(t),\quad t\in(\alpha, \beta). \] In such a way, the original signal \(\mathbf{c}\) is decomposed to an approximation signal \(\mathbf{a}:=(\dots, a_{-2}, a_{-1}, a_0,a_1, a_2, \ldots)\) on the coarse grid \(\tilde X\) and a wavelet (detail) signal \(\mathbf{b}:=(\dots, b_{-2}, b_{-1}, b_0, b_1, b_2, \ldots)\) on the original grid \(X\). Reconstruction can be easily done by summing up the approximation signal and the detail signal. The remaining part of the paper is on the details of the above procedure.