Matrix Padé-type approximant and directional matrix Padé approximant in the inner product space. (Q1426783)

Earlier the author defined a generalized inverse matrix Padé approximant (GMPA) on the basis of the scalar product of matrices. As compared to the existing matrix Padé approximants, GMPA does not need multiplication of matrices in the construction process and it may be useful in the noncommutativity problems of the matrix multiplication. However, it possesses the degree constraint and the divisibility constraint, which is caused by its construction process. The constraints imply that the method does not construct matrix Padé approximants of type \([m/n]\) when \(n\) is an odd number. The aim of this paper is to fill up this gap. The author defines a new matrix Padé-type approximant (MPTA) by introducing a generalized linear functional in the inner product space, which is different from the Clifford algebra approach in the vector case. At the same time, it is also different from GMPA approach in matrix case since it does not use the generalized inverse of matrices in its computation. The expressions of MPTA are provided with the generating function form, which is derived from the definition, and the determinant form, which is connected to the orthogonal polynomials. Directional vector Padé approximant of Graves-Morris is naturally extended to the matrix case by giving a set of linearly independent matrices. It is pointed out that the method of MPTA can be applied to the reduction problems of the high degree multivariable linear system.