Striped Bass Optimization Algorithm

To address the problems of premature convergence, insufficient utilization of directional information, and weak scale adaptation in traditional swarm intelligence algorithms for complex, non-convex, multimodal, and high-dimensional optimization problems, this paper proposes an adaptive manifold migration optimization algorithm for striped bass based on the migration behavior of a striped bass population. This method constructs a unified dynamic system from five levels: geometric structure, adaptive metric, anisotropic damping, non-equilibrium energy regulation, and stable distributed transitions. First, a local covariance structure matrix is constructed using elite individuals to define a dynamic metric space, realizing a manifold compression convergence mechanism. Second, an information streamline tensor is constructed based on differential gradients to achieve anisotropic directional regulation. Third, a dual-field system of energy and entropy is introduced to couple and control population diversity and kinetic energy. Subsequently, a stable distributed transition operator enhances global ergodicity. Finally, step size modulation is achieved by combining local second-order differential curvature. The algorithm can be interpreted as a dissipative non-equilibrium dynamic system on a random manifold. This paper systematically presents a complete mathematical modeling, unified update equation, dynamic interpretation, and convergence analysis framework, providing a new approach to the geometric modeling of swarm intelligence algorithms.