Test Problems in Optimization

Xin-She Yang

DOI: https://doi.org/10.48550/arXiv.1008.0549

2010-08-03

Abstract:Test functions are important to validate new optimization algorithms and to compare the performance of various algorithms. There are many test functions in the literature, but there is no standard list or set of test functions one has to follow. New optimization algorithms should be tested using at least a subset of functions with diverse properties so as to make sure whether or not the tested algorithm can solve certain type of optimization efficiently. Here we provide a selected list of test problems for unconstrained optimization.

Optimization and Control

What problem does this paper attempt to address?

The paper primarily focuses on addressing the issues of optimization algorithm verification and performance comparison. Specifically, the author Xin-She Yang emphasizes the importance of test functions in the paper. These test functions are used to verify the effectiveness of new optimization algorithms and to compare their performance with existing ones. Due to the presence of various test functions in the literature but the lack of a unified standard list, newly developed optimization algorithms need to be tested with a set of test functions with different characteristics to determine whether they can effectively solve specific types of optimization problems.

The paper provides several classic examples of optimization test functions, including but not limited to the Ackley function, the De Jong function series (such as the Sphere function, the Hyper-Ellipsoid function, etc.), the Easom function, the Griewank function, the Rosenbrock function (also known as the "Banana function"), and some functions designed by the author himself. These test functions cover a range of scenarios from unimodal to multimodal, and from convex to non-convex functions, aiming to comprehensively evaluate the capabilities of optimization algorithms.

In summary, the core purpose of this paper is to provide a selected set of test problems for unconstrained optimization issues, to help researchers better develop and evaluate new optimization algorithms.