Characterizations of orthonormal scale functions: A probabilistic approach (Q5950234)

The authors discuss the specification of scaling functions, which constitute the first step in multiresolution analysis. In the Fourier domain these scaling functions can be expressed as an infinite product of `low pass filters'. The prime interest of the paper is the question of finding necessary and sufficient conditions on the `filters', when they are not necessarily \(C^1(R)\). The authors propose a novel and more general approach by interpreting the squared filters as conditional probabilities. The authors are able to demonstrate that previous results in the field can be seen as special cases of the present work.