Genetic-Algorithmic Optimisation for School-Allocation Mechanisms: A Study of Amsterdam's Student to High-School Allocation Problem Philip W.B. Michgelsen Abstract: Genetic-Algorithmic Optimisation for School-Allocation Mechanisms A Study of Amsterdam's Student to High-School Allocation Problem Philip W.B. Michgelsen Abstract: In the municipality of Amsterdam, students who graduated from primary school, are allocated to high schools through a school-allocation mechanism. In recent years, different school-allocation mechanism have been used and all have been criticised by the public. This thesis investigates the school-allocation mechanisms that have been used in Amsterdam, from a mathematical point of view, and aims to make it mathematically clear and formal what properties a school-allocation mechanism must have, according to the municipality of Amsterdam's policy makers, and what properties the used allocations mechanisms actually have. Furthermore, this thesis aims to explore if a genetic-algorithmic optimisation method can help in the design of a new schoolallocation mechanism, which can outperform the school-allocation mechanisms used in Amsterdam. The two most important formal properties desired of school-allocation mechanisms by policy makers, are Pareto consistency and strategy proofness. A school-allocation mechanism is Pareto consistent if it only produces Pareto-optimal allocations. In addition, other conditions are posed by Amsterdam's policy makers, which determine, given two Pareto-optimal allocations, which is considered to be better. With these additional conditions, an objective fitness function can be given, which determines which Pareto-optimal allocation is fittest. The existence of a particular school-allocation mechanism, the Permutation Allocation Mechanism, which is able to produce all Pareto-optimal allocations for some given school market, makes it then possible to find the fittest Pareto-optimal school allocation by using a genetic-algorithmic optimisation method. This thesis shows that the so obtained genetic-algorithmic optimised school-allocation mechanism is expected to statistically outperform its not-optimised counterpart.