Using Local Search with adaptive operator selection to solve the Progressive Party Problem

This paper investigates the use of adaptive operator selection in the context of Local Search to solve a constraint satisfaction problem, namely the Progressive Party Problem. Operators are selected according to a utility value which is computed, for each operator, from the solution quality and from the distance of the candidate solution to recently visited solutions in the search trajectory. We show that using several non-problem-specific operators gives comparable successful resolution rates to an algorithm customized for the problem, albeit with slower run times.