White-Spotted Dogfish Optimization Algorithm

To address the common problems in high-dimensional complex optimization problems, such as multimodal structures, scale inhomogeneities, local extremum traps, and convergence instability, this paper proposes a swarm intelligence optimization algorithm—the White-spotted Shark Optimization Algorithm—based on information curvature fields and asymmetric dynamics. The algorithm uses continuous-time dynamical systems as its theoretical prototype. It constructs a nonlinear iterative framework through discretization and introduces an information density function, a curvature-driven electric field, an anisotropic dynamic metric tensor, and an irreversible energy release mechanism into the search space, thus forming a metaheuristic model with directional adaptability and temporal irreversibility. The algorithm estimates the information curvature by constructing a local second-order difference approximation and embedding the curvature field into a time-evolving non-Euclidean metric space, allowing the search behavior to dynamically deform with the population's statistical structure. Simultaneously, it establishes asymmetric time dynamics through latent energy accumulation and a cooling-freezing mechanism to enhance the ability to escape local extrema. Furthermore, a risk-information coupling potential function is constructed to regulate the balance between exploration and exploitation. Theoretical analysis shows that the algorithm constitutes a nonlinear stochastic dynamical system with dissipative terms, satisfying probabilistic convergence under weak conditions. This model provides a unified modeling framework for swarm intelligence algorithms based on information geometry and irreversible dynamics.