Genesis: A Language for Generating Synthetic Programs

This thesis describes Genesis, a new language used for the expression and generation of synthetic programs. The language allows users to annotate a template program to identify code segments of the program they wish to vary. Users also describe how parameters in each code segment can vary across generated instance programs using statistical distributions. The Genesis preprocessor uses the annotations to generate a set of instance programs whose characteristics or features vary in a statistically controlled fashion. We describe Genesis, its functionalities, its language constructs, a prototype preprocessor for the language, and three case studies that demonstrate the ability of Genesis to express a range of programs in different domains. We evaluate the performance of the preprocessor during the generation of instance programs, and the statistical quality of the samples it generates. We believe that Genesis is a useful tool that eases the expression and creation of large and diverse program sets for a multitude of domains and applications, such as machine learning, image generation, and compiler testing.M.A.S