Abstract

Traditionally, reasoning about programs is a boolean problem that is solved using the discrete tools of logic. However, in many emergent applications, discrete reasoning about programs needs to be supplemented by quantitative analysis. This demand opens up a new playground for research: there is now the opportunity to combine the discrete tools of program analysis with continuous tools like numerical optimization, geometric algorithms, and statistical machine learning.

This talk will describe some recent and ongoing work on topic. Specifically, I will talk about smoothed proof search, a new approach to synthesis with respect to boolean and quantitative goals that couples discrete program analysis with continuous numerical optimization. I will also touch upon a few other ongoing projects on this topic in my group, including reasoning about programs under numerical uncertainty, a synthesis approach for robotics that combines SAT -based reasoning with continuous motion planning, and a statistical program analysis used to grade student submissions in a massive open online course.

Bio: Swarat Chaudhuri is an assistant professor of computer science at Rice University. He is an expert on program verification and synthesis. Swarat received a bachelor’s degree in computer science from the Indian Institute of Technology, Kharagpur, in 2001, and a doctoral degree in computer science from the University of Pennsylvania in 2007. He is a recipient of the National Science Foundation CAREER award, the ACM SIGPLAN Outstanding Doctoral Dissertation Award, and the Morris and Dorothy Rubinoff Dissertation Award from the University of Pennsylvania.