Buggy Pinball: A Novel Single-point Meta-heuristic for Global Continuous Optimization

In this work, we propose a fundamentally novel single-point meta-heuristic designed for continuous optimization. Our algorithm continuously improves on a solution via a trajectory-based search inspired by the pinball arcade game in an anytime optimization manner. We evaluate our algorithm against widely employed meta-heuristics on several standard test-bed functions and various dimensions. Our algorithm exhibits high precision, and superior accuracy compared to the benchmark, especially when complex configuration spaces are considered.