Venom as a Computational Resource: A Subspecies-Based King Cobra Optimization Framework

To address the problems of existing swarm intelligence optimization algorithms, such as reliance on static search space assumptions, overemphasis on cooperative information sharing, and susceptibility to premature convergence, this paper proposes a novel nature-inspired optimization method—the King Cobra Subspecies Optimization Algorithm. This method no longer views the optimization process as a point-by-point approximation of the optimal solution within a fixed search space, but rather models it as a dynamic ecosystem composed of heterogeneous individuals, actively reconstructing and shaping the feasible solution space during evolution. The algorithm actively intervenes in the search space structure by introducing venom-induced feasible region deformation, anti-cooperative information propagation, subspecies lineage dynamics, and failure-driven information amplification. Unlike traditional particle swarm optimization, gray wolf optimization, and their variants, the optimal solution in this method does not attract other individuals to cluster around it. Instead, it reduces the search density in local regions through information suppression and spatial repulsion, thus theoretically preventing population collapse. This paper systematically characterizes the evolutionary mechanism of the algorithm from a mathematical modeling perspective and analyzes its potential advantages in multi-peak continuous optimization problems.