Optimal sequential search: A Bayesian approach (Q1074496)

n objects are hidden in m boxes in such a way that each of the objects is hidden in box i with probability \(\pi_ i\), independently of other objects. The aim is to search box by box until at least one object is found. The cost of inspection of a box is positive and the probability of discovering any given object hidden in a box when searching this box is smaller than one. Optimal search procedures for the case that \(\pi =(\pi_ 1,...,\pi_ m)\) is fixed and known are given in the literature; in the paper the case that an a priori distribution for \(\pi\) is given is considered. It appears that even small uncertainty about \(\pi\) may lead to a solution totally different from that for \(\pi\) fixed and known.