There is no doubt that both determining theoretical properties and characterizing the observed behavior of an evolutionary algorithm allow us to understand when to use such an algorithm in solving a class of optimization problems. One of those evolutionary algorithms is the Hybrid Adaptive Evolutionary Algorithm (haea). The general scheme followed by a haea algorithm is to evolve every individual of the population by selecting genetic operators according to a kind of chaotic competition mechanism. This paper proposes and studies, from both theoretical and experimental points of view, the class of hybrid adaptive evolutionary algorithms (called chavela), i.e., the class of evolutionary algorithms that follow such a general scheme. In this way, this paper presents a formal characterization of the chavela class in terms of Markov kernels; establishes convergence properties; proves that (parallel) hill-climbing algorithms belong to the chavela class; develops generational, steady-state, and classic versions; and analyzes the running behavior of chavela on well-known optimization functions.

毫无疑问，确定理论性质和表征演化算法的观测行为都使我们能够了解何时使用此类算法来解决一类优化问题。那些进化算法之一是混合自适应进化算法（haea）。遵循haea算法的通用方案是根据一种混沌竞争机制，通过选择遗传算子来进化种群的每个个体。本文从理论和实验的角度提出并研究了混合自适应进化算法（称为chavela）一类。），即遵循这种一般方案的进化算法的类别。这样，本文就用Markov核表示了chavela类的形式化特征。建立收敛特性；证明（并行）爬山算法属于chavela类；开发世代，稳态和经典版本；并根据著名的优化函数分析了chavela的运行行为。