On the calculation of functions of matrices (Q382372)

It is pointed out that the representation of entire functions of matrices involving symmetric polynomials is remarkable, since formulas expressing these functions via the principal minors of matrices can be derived. It is shown that these formulas can serve as a basis for developing new methods for the numerical calculation of functions of matrices. These methods do not require the calculation of eigenvalues of the relevant matrices. Two examples of such methods are presented in this paper. It is pointed out that such methods are reliable and accurate, easy to program, and generally better than other presently available methods.