Megamouth Shark Optimization Algorithm

Swarm intelligence optimization algorithms have wide applications in global optimization problems in continuous space, but traditional methods often face problems such as premature convergence, isotropic orientation, and an imbalance between exploration and exploitation in high-dimensional complex functions. This paper proposes a giant shark optimization algorithm based on information refraction field, manifold compression mechanism, and non-equilibrium free energy driven structure. The algorithm constructs an information potential function to characterize the local information structure of individuals, introduces an inverse covariance tensor to achieve anisotropic compression search, and designs a non-equilibrium dissipative free energy system to control reversible and irreversible dual-drive channels. The algorithm also employs a fractal aperture decay mechanism to adjust the ratio of global to local search and integrates quantum probability perturbation to enhance the ability to escape local extrema. This paper analyzes the stability and energy dissipation properties of the algorithm from the perspective of dynamical systems, provides a discussion of Lyapunov-type convergence, and analyzes the time complexity. Theoretical analysis shows that the proposed method maintains global exploration capability while exhibiting good local fine-grained search performance.