51³Ô¹ÏÍø

The Best Experience Paper award at ASE 2017 was awarded to a paper from authors including 51³Ô¹ÏÍø researchers Liew, Cadar and Donaldson.

The was held last week at the University of Illinois at Urbana-Champaign, Illinois. The conference's Programme Committee selected a paper from 51³Ô¹ÏÍø researchers (PhD student) and and (Daniel's PhD supervisors), co-authored with collaborators at RWTH Aachen University, for the Best Experience Paper Award.

The paper, , tells a story whereby the teams of researchers at 51³Ô¹ÏÍø and Aachen discovered that each other had independently implemented extensions to the KLEE symbolic execution engine to support floating-point reasoning. The teams used this coincidence, an undesirable situation in some respects, to undertake a controlled study in N-version programming, with N = 2, undertaking a controlled experiment in which the two extensions to KLEE were systematically improved, driven by sets of benchmarks written independently produced by opposite teams, and then rigorously compared, both quantitatively with respect to reliability and performance, and qualitatively to investigate notable design similarities and differences.

The paper describes in detail the methodology that was followed, and the lessons learned, both about floating-point symbolic execuction and N-version programming, via the study.