A new algorithm for constructing optimal experimental designs

• In any experimental science we are sometimes confronted with new experimental situations, where underlying physical models are unknown. After a first attempt at identifying physical variables that could effect the experimental result, we need to perform what is called a screening experiment. The objective is to conduct a small number of experiments that span as much of the experimental space as possible, to merit further experiments in smaller, more promising test systems. In this report we present an algorithm that, given a set of possible experiments, selects a subset of these. The selection is designed to cover as much variation as possible in the experimental space, and to assure a minimized expected mean squared error in the model parameter estimation. We verified the algorithm with an example taken from organic synthesis.

Subject headings

TEKNIK OCH TEKNOLOGIER  -- Elektroteknik och elektronik -- Signalbehandling (hsv//swe)

ENGINEERING AND TECHNOLOGY  -- Electrical Engineering, Electronic Engineering, Information Engineering -- Signal Processing (hsv//eng)

Keyword

Signalbehandling

Signal Processing

Publication and Content Type

vet (subject category)

rap (subject category)