Bigeye Tuna Optimization Algorithm

To address the shortcomings of traditional swarm intelligence optimization algorithms in high-dimensional complex spaces, such as insufficient information representation, dimensional convergence imbalance, premature convergence, and search path oscillation, this paper proposes a novel swarm intelligence optimization method—the Bigeye Tuna Optimization Algorithm. Inspired by the deep-sea perception and group hunting behavior of bigeye tuna in marine ecosystems, this algorithm constructs an information depth field model to characterize the density of historical search trajectories, introduces an anisotropic visual tensor to achieve a dimensional adaptive convergence mechanism, establishes a curvature-driven subspace transition model to enhance adaptability in high-dimensional spaces, and designs an energy-curvature coupling scheduling mechanism, a self-organized prey phantom perturbation mechanism, and an irreversible risk compression term to suppress path oscillation and premature convergence. This paper continuously models the algorithm from a dynamic perspective and analyzes its stability and complexity. Theoretical analysis shows that the algorithm can achieve multi-scale balanced convergence while maintaining global exploration capabilities. This method provides a new information field-driven swarm dynamics framework for complex non-convex optimization problems.