Genetic algorithms categorized as global search heuristics are search technique of seeking approximate or exact solutions for optimizing problems. Our generative tool named notchSpace starts from a population of randomly generated individuals which rationally subdivide architectural spaces with a low “fitness”. During generationsi evolution, the population is recombined, modified and mutated at a low rate to form new modes of building spatial subdivision, a new population is used in the next iteration. NotchSpace works up a research platform to evolve architectural spaces, and it explores the thinking of translation from computer algorithms to the field of architecture.