The Boa Constrictor Optimization Algorithm: Irreversible Feasible-Region Dynamics for High-Dimensional Optimization

In recent years, swarm intelligence and metaheuristic optimization algorithms have been widely applied in black-box optimization, high-dimensional non-convex problems, and combinatorial optimization. However, most existing methods still focus on the "stochastic evolution of search points" as the core modeling object, lacking a systematic characterization of the geometric structure and information evolution mechanism of the search process. This point-centric perspective often leads to search oscillations, premature convergence, or efficiency degradation in high-dimensional, strongly coupled, or mixed discrete-continuous problems. This paper proposes a new optimization theoretical framework—Ball-Python Optimization (BPO-X). Its core idea is not to directly optimize a single candidate solution, but to formalize the optimization process as a "sequence of feasible regions that evolves over time and shrinks irreversibly." Under this framework, optimization is modeled as a feasible region dynamic system driven by information uncertainty, where the feasible region is progressively compressed under asymmetric metrics and achieves stable convergence through an irreversible dimensionality reduction mechanism triggered by information theory. This paper systematically presents the mathematical definition, dynamic update rules, and theoretical properties of BPO-X, and proves that the framework achieves several properties such as dimension freezing, monotonically shrinking feasible region, and near-global convergence within a finite number of steps. Furthermore, this paper shows that the method can be naturally generalized to continuous, discrete, and combinatorial optimization problems in a unified form, and forms an intrinsic connection with classic combinatorial structures such as Knapsack Polytope.