Probabilistic Program Analysis

This paper provides a survey of recent work on adapting techniques for program analysis to compute probabilistic characterizations of program behavior. We survey how the frameworks of data flow analysis and symbolic execution have incorporated information about input probability distributions to quantify the likelihood of properties of program states. We identify themes that relate and distinguish a variety of techniques that have been developed over the past 15 years in this area. In doing so, we point out opportunities for future research that builds on the strengths of different techniques.