The optFUMOLA package: A simulation-based black-box optimization library and interface

The optFUMOLA software package provides a dedicated interface and algorithm library for simulation-based blackbox optimization. Given a system model, it aims to calculate the optimal choice of model input parameters that minimize a given objective function by evaluating simulation results with the help of well-established optimization algorithms. To this end, optFUMOLA follows a simple and flexible design concept that serves as framework to utilize a wide spectrum of optimization algorithms, simulation tools and computing environments. As such it provides an easy-to-use handle for expert users to apply system optimization to a variety of simulation approaches, ranging from single-core domain-speciflc simulators to distributed co-simulation environments. The rationale behind optFUMOLA's design concept is explained and the most important features of the implementation are presented. Furthermore, the feasibility of the approach is demonstrated with the help of a case study. The results of the implemented optimization algorithms are compared, with a focus on the number of simulation runs needed to obtain competitive results and a discussion of implications for other applications.