Competitive search in symmetric trees

Abstract

We consider the problem of searching for one of possibly many goals situated at unknown nodes in an unknown tree T. We formulate a universal search strategy and analyse the competitiveness of its average (over all presentations of T) total search cost with respect to strategies that are informed concerning the number and location of goals in T. Our results generalize earlier work on the multi-list traversal problem, which itself generalizes the well-studied m-lane cow-path problem. Like these earlier works our results have applications in areas beyond geometric search problems, including the design of hybrid algorithms and the minimization of expected completion time for Las Vegas algorithms. © 2011 Springer-Verlag.