On rational interpolation with positivity constraints (Q1179989)

This is motivated by the practical problem that in constructing a filter with given amplitude response \(| G(i\omega)|\) one actually has to construct the rational function \(G(z)\) with real coefficients. Thus \(| G(i\omega)|^ 2=G(i\omega)G(-i\omega)\).Conversely, given a nonnegative rational function \(F(\omega^ 2)\), it can be factored as above and is the squared modulus of a stable minimum phase \(G\). The paper treats the problem of interpolating given data with an \(F\) that is nonnegative.