Software Correctness at Scale through Testing and Verification

Leonidas Lampropoulos
Victor Basili Postdoctoral Fellow, Department of Computer Science University of Maryland, University Of Pennsylvania
Talk Series: 
02.26.2020 11:15 to 12:15

IRB 4105

Software correctness is becoming an increasingly important concern as our society grows more and more reliant on computer systems. Even the simplest of software errors can have devastating financial or security consequences. How can we find errors in real-world applications? How can we guarantee that such errors don't exist? To even begin to answer these questions we need a specification: a description of what it means for a piece of code to be correct, stated either explicitly (e.g. my private data are never leaked to third parties) or implicitly (e.g. this code will not terminate with an uncaught exception). In this talk, I will discuss efficient ways to debug and reason about software and specifications, focusing on two techniques and their interplay: property-based testing and mechanized formal verification. Testing can quickly uncover errors in complex software, but it cannot guarantee their absence. Formal verification provides strong correctness guarantees, but is still very expensive in time and effort. Together, they are a match made in heaven.