Community of Scientists Optimization: An autonomy oriented approach to distributed optimization

A novel optimization paradigm, Community of Scientists Optimization (CoSO), is presented in thispaper. The approach is inspired to the behaviour of a community of scientists interacting, pursuing for research results and foraging the funds needed to held their research activities. The CoSO metaphor can be applied to general optimization domains, where optimal solutions emerge from the collective behaviour of a distributed community of interacting autonomous entities.The CoSO framework presents analogies and remarkable differences with other evolutionary optimizationapproaches: swarm behaviour, foraging and selectionmechanism based on research funds competition, dynamically evolving multicapacity communication channels realized by journals and evolving population size regulated by research management strategies.Experiments and comparisons on benchmark problems show the effectiveness of the approach for numericaloptimization. CoSO, with the design of appropriate foraging and competition strategies, also represents a great potential as a general meta-heuristic for applications in non-numerical and agent-based domains.