A PROBLEM IN THE SEQUENTIAL DESIGN OF EXPERIMENTS

Abstract

Abstract : THE PROBLEM OF DETERMINING AN OPTIMAL TESTING POLICY WHERE ONE SIMULTANEOUSLY GAINS AND LEARNS FOR THE CASE WHERE THE OUTCOME OF ONE CHOICE IS KNOWN AND THE OTHER IS SUBJECT TO A KNOWN A PRIORI DISTRIBUTION IS CONSIDERED. Results of Johnson and Karlin, P-328, are obtained in a different way and extended. The methods used are applicable to more general processes.