Ras Bodik

"Adventures in Program Synthesis"

(Vortrag im Rahmen der "MPI Distinguished Lecture Series" in Kooperation mit dem Fachbereich Informatik)

If you possessed the power to synthesize any program from just a high-level description of the desired behavior, how would you deploy this power? It seems that you could end programming as we know it and automate the creation of everything that can be viewed as a program, including biological and economic models. This dream is not fully achievable but over the last decade we have learnt how to rephrase the synthesis problem to make it solvable and practically applicable.

This talk will mine seven projects from my group for surprising lessons and future opportunities. I will first present two fresh answers to the question of "where do specifications come from?" and then argue for synergy between domain-specific languages and synthesis. I will also explain why synthesis may help rethink compiler construction, necessitated by the era of unusual hardware. Looking into the next decade, I will illustrate how synthesis may facilitate computational doing --- the development of tools for data science and the digital life.

Bio: Ras Bodik is a Professor of Computer Science at UC Berkeley. He works on a range of techniques for program synthesis, from programming by demonstration, to sketching, and solver-aided languages. His group has applied synthesis to high-performance computing, web browser construction, algorithm design, document layout, and biology. He has designed a course on programming languages where student learn hands-on small-language design by constructing a modern web browser.