BenchmarkFcns

https://github.com/mazhar-ansari-ardeh/BenchmarkFcns/actions/workflows/build.yaml/badge.svg

BenchmarkFcns is a high-performance library providing a comprehensive suite of over 310 mathematical benchmark functions designed for the rigorous testing and evaluation of optimization algorithms.

By combining a core engine implemented in C++ with Eigen and multi-core parallelization via OpenMP, the library offers extremely fast evaluation speeds, often providing 10x-30x speedups over traditional pure-Python implementations. This makes it an ideal choice for large-scale tasks, Reinforcement Learning training, and high-fidelity surrogate modeling.

Core Benefits

  • High Performance: Optimized C++ core leveraging modern SIMD instruction sets.

  • AI & RL Ready: Standardized Gymnasium (OpenAI Gym) environments for training agents on complex landscapes.

  • Massive Batch Support: Native vectorization allows evaluating millions of data points in parallel.

  • Comprehensive Coverage: Includes Classic (100+), Multi-Objective (50), Multi-Fidelity (49), and official CEC Competition suites (2005-2022).

  • Cross-Language Consistency: Mathematically verified implementations ensuring consistent results across Python, C++, and MATLAB.

Indices and tables