Local Completeness for Program Correctness and Incorrectness (Invited Talk)

Program correctness techniques aim to prove the absence of bugs, but can yield false alarms because they tend to over-approximate program semantics. Vice versa, program incorrectness methods are aimed to detect true bugs, without false alarms, but cannot be used to prove correctness, because they under-approximate program semantics. In this invited talk we will overview our ongoing research on the use of the abstract interpretation framework to combine under- and over-approximation in the same analysis and distill a logic for program correctness and incorrectness.