Diagonalization and spectral decomposition of factor block circulant matrices (Q1100534)

A square matrix is called circulant if each row after the first is obtained from its predecessor by a cyclic shift. The basic fact concerning circulant matrices with scalar components is that they are simultaneouly diagonalizable by the Fourier matrix. This property is extremely useful for deriving innumerable properties (for example, solutions of differential equations) in ``closed form''. For block circulants, i.e. circulant matrices whose entries are square matrices, there is still a diagonalization by blocks in terms of Kronecker products of appropriate Fourier matrices; however this representation is more difficult to work with. The authors introduce factor circulant matrices: matrices with the structure of circulants, but with entries below the diagonal multiplied by a fixed factor. For this class they derive results and applications analogous to those discussed above for circulants.