2D versus 3D Bloch Sphere Representations in Quantum-Inspired Evolutionary Algorithms: An Empirical Comparison on Benchmark Functions
Büşra Özcan, Abdullah Ateş
The Quantum-inspired Evolutionary Algorithm (QEA) describes each individual with a qubit-based structure that works with probabilities, not fixed sequences of bits. While the Bloch sphere is naturally three-dimensional, most QEA studies focus on just a two-dimensional part and ignore the phase part of the quantum state. This research looks at whether using the third dimension of the Bloch sphere, by adding the phase angle ϕ, actually helps the algorithm do better on continuous optimization problems. Two versions of QEA were tested. The first uses the usual two-dimensional approach, and the second adds a phase angle to make it three-dimensional. The testing procedure was identical for both models: each used three variables, assigned four bits to every variable, and carried out thirty independent runs for each problem. The paper presents ten problems, including some with one peak, some with many peaks, some with variables separated, some with variables not separated, and one engineering example. The results showed that the two-dimensional model had less variation in its results for eight of the ten problems, but the three-dimensional model was better only for the rotated ellipsoid and an engineering example, where the variables are connected to each other. The average CPU time was almost the same for both models. Rather than proposing a new optimiser, the study is designed as a controlled ablation in which the phase degree of freedom is the only manipulated factor, with encoding, population size, selection and computational budget held fixed. In summary, using phase information only helps when the problem has geometric relationships that match the way the three-dimensional update works.
International Artificial Intelligence and Data Processing Symposium (IDAP)•Sep 1, 2026•0 citations