Combining a spatiotemporal, multi-agent based model of a foraging ecosystem with linear, genetically programmed rules for the agents' behaviors results in implicit, endogenous, objective functions and selection algorithms based on "natural selection". Use of this implicit optimization of genetic programs for study of biological systems is tested by application to an artificial foraging ecosystem, and compared with established biological, ecological, and stochastic gene diffusion models. Limited program memory and execution time constraints emulate real-time and concurrent properties of physical and biological systems, and stress test the optimization algorithms. Relative fitness of the agents' programs and efficiency of the resultant populations as functions of these constraints gauge optimization effectiveness and efficiency. Novel solutions confirm the creativity of the optimization process and provide an unique opportunity to experimentally test the neutral code bloating hypotheses. Use of this implicit, endogenous, evolutionary optimization of spatially interacting, genetically programmed agents is thus shown to be novel, consistent with biological systems, and effective and efficient in discovering fit and novel solutions.

翻译：结合基于时空多智能体的觅食生态系统模型与智能体行为的线性遗传编程规则，产生了基于“自然选择”的内隐、内源目标函数与选择算法。通过将其应用于人工觅食生态系统，并对照已有的生物学、生态学及随机基因扩散模型进行检验，验证了这种基于基因程序隐式优化在生物系统研究中的有效性。有限的程序内存与执行时间约束模拟了物理及生物系统的实时性与并发特性，并对优化算法进行了压力测试。智能体程序的相对适应度及由此产生的种群效率作为这些约束条件的函数，衡量了优化的有效性与效率。新颖的解证明了优化过程的创造性，并为实验检验中性代码膨胀假说提供了独特机会。因此，这种对空间交互的遗传编程智能体进行内隐、内源性进化优化的方法，被证明具有新颖性、与生物系统的一致性，并且在发现适应性与创新性解方面高效且有力。