DescriptionEnsuring the correctness of high-performance computing (HPC) applications is one of the fundamental challenges that developers and users of these applications face today. An application is correct when it performs what a user expects with respect to a specification. Given today's complex HPC software stack, correctness is very difficult to achieve---the use of combined parallel programing models (e.g., MPI+OpenMP), complex compiler optimizations/transformations, floating-point precision issues, and unanticipated scale-dependent behavior, are some of the challenges to achieve correctness. As emerging task-based programing models and heterogeneous architectures become more predominant in HPC, the level of non-determinism in applications increase, which makes the isolation of errors much harder. The aim of this workshop is to bring together researchers and developers to present and discuss novel ideas to address the problem of correctness in HPC.