MF2: A Collection of Multi-Fidelity Benchmark Functions in Python

The Journal of Open Source Software, vol. 5, pp. 2049

Abstract

The field of (evolutionary) optimization algorithms often works with expensive black-box optimization problems.However, for the development of novel algorithms and approaches, real-world problems are not feasible due to their high computational cost.Instead, benchmark functions such as Sphere, Rastrigin, and Ackley are typically used.These functions are not only fast to compute, but also have known properties which are very helpful when examining the performance of new algorithms.

Authors 2

1. Sander van Rijn Aachen

   Leiden University

   Affiliation as printed

   Leiden University , The Netherlands

   Leiden University, The Netherlands

2. Sebastian Schmitt

   Honda (Germany)

   Affiliation as printed

   Honda Research Institute Europe , Germany

   Honda Research Institute Europe, Germany