Creating discrete joint densities from continuous ones: the moment matching-maximum entropy approach

Abstract An approach for converting continuous densities into discrete ones with a pre-assigned set of support points is developed. The algorithm performs this conversion in a most skep1tical, least-informative way, by finding the discrete distribution with the maximum entropy that satisfies a set of moment constraints derived from the stated continuous distribution, such as means and variances. Individualized drug therapies based on multiple model control rely on the availability of discrete distributions to generate the underlying model set. The methods developed herein are especially compatible with individualized drug therapies based on multiple model control that rely on the availability of discrete densities to generate the model set.