Greedy Randomised Adaptive Solution Builder using Priority Rules

Priority rules are widely utilised in combinatorial optimisation because they effectively balance solution quality and computational cost. This makes them particularly suitable for addressing problems with extreme time constraints, such as real-time or dynamic scenarios. In contrast, other methods such as metaheuristics or exact algorithms are less frequently applied in these contexts due to their higher computational complexity. However, designing effective rules is a challenging task. To address this, genetic programming has been widely used for the automated generation of rules. Despite their advantages, individual rules can sometimes yield suboptimal results. However, numerous studies demonstrated that by using a set of rules instead of individual rule significantly improves their effectiveness. This paper examines various strategies of how a set of rules generated by genetic programming can be utilised to solve optimisation problems and introduces a novel algorithm, named the greedy randomised adaptive solution builder, for solving the travelling salesman problem. The experimental results demonstrate that the proposed method is capable of obtaining reasonably good solutions under highly time-constrained scenarios.