Algebras of higher dimension for displacement decompositions and computations with Toeplitz plus Hankel matrices (Q1904050)

A new decomposition formula for the representation of a matrix in the form of a sum of products of matrices belonging to two particular matrix algebras having dimension about \(2n\) and being noncommutative. These algebras are \({\mathcal C}_e = {\mathcal C} + K {\mathcal C}\) and \({\mathcal S}_e = {\mathcal S} + K {\mathcal C}\), where \({\mathcal C}\) and \({\mathcal S}\) are, respectively, the algebras of circulant and skew circulant matrices, \(K = (\delta_{i,n - 1 - j})\) with \(i,j = 0, \dots, n - 1\) is the reversion matrix, and + denotes the sum of linear spaces. The algebras \({\mathcal C}_e\) and \({\mathcal S}_e\) retain the main computational properties of \({\mathcal C}\) and \({\mathcal S}\), although they have dimension about \(2n\) and are noncommutative, while \({\mathcal C}\) and \({\mathcal S}\), as well as all the algebras mentioned above, have dimension \(n\) and are commutative. The main result is the following theorem: Let \(n\) be even, \(A \in \mathbb{C}^{n \times n}\) and \[ {\mathcal D}_{S_e} (A) = \sum^\alpha_{m = 1} x_m y^T_m = \sum^{\alpha/2}_{m = 1} (x_{2m - 1} - y_{2m}) \left( \begin{matrix} y^T_{2m - 1} \\ y^T_{2m} \end{matrix} \right), \quad x_m, y_m \in \mathbb{C}^n, \] with \(e^T(y_{2m - 1} - y_{2m}) = 0\), \(\widetilde e^T (y_{2m - 1} + y_{2m}) = 0\), where \(e = \sum^{n - 1}_{k = 0} e_k\) and \(\widetilde e = \sum^{n - 1}_{k = 0} (-1)^ke_k\) and \(e_k (k = 0, \dots, n - 1)\) is the canonical basis. Then \[ A = {1 \over 2} \sum^{\alpha/2}_{m = 1} {\mathcal S}^T_e (x_{2m - 1}, x_{2m}) {\mathcal C}_e (y_{2m}, y_{2m - 1}) + {\mathcal S}_e^T (Ae_0, Ae_{n - 1}). \] Finally, applications of this result are exploited, e.g., for the computation of the product of the matrix \(A\) by a vector \(b\), given through an ``\(\alpha\)-length'' \({\mathcal D}_{S_e}\)-generator \(X = (x_1, \dots, x_\alpha)\), \(Y = (y_1, \dots, y_\alpha)\), i.e., \({\mathcal D}_{S_e} (A) = \sum^\alpha_{m = 1} x_m y_m^T\), \(A \in \mathbb{C}^{n \times n}\).