Neurotoxin-Field–Driven Black Mamba Optimization with Reversible Dimensional Collapse

In high-dimensional continuous optimization and black-box optimization problems, traditional swarm intelligence algorithms generally rely on random exploration and collective statistical behavior, and their search efficiency decreases significantly with increasing dimensionality. This paper proposes a novel continuous optimization method, Neuro-Venom Black Mamba Optimization (NV-BMO), which uses neurotoxin propagation and dimensional phase transitions as its core mechanisms to model the optimization process as an adaptive search space evolution induced by a toxin field. Unlike existing methods, NV-BMO does not treat the optimal solution as a single attractor point, but instead constructs a toxin skeleton to characterize the geometric relationships between multiple structurally feasible solutions. Simultaneously, the algorithm introduces a reversible dimensional freezing and thawing mechanism during iteration, causing the search space to spontaneously collapse to a low-dimensional manifold during operation, and then switching to geometry-driven optimization dynamics after collapse. This paper systematically presents the mathematical modeling, update rules, and search collapse conditions of NV-BMO, and analyzes its dimensional collapse behavior and convergence characteristics from a theoretical perspective. This method provides a unified perspective for high-dimensional optimization that differs from traditional swarm intelligence and gradient methods.