AI, Data Science & Cloud Certificates from Google, IBM & Meta — 40% Off
One plan covers every Professional Certificate on Coursera. 40% off Coursera Plus Annual.
Unlock All Certificates
Learn about a groundbreaking research presentation from POPL 2018 that introduces a generic approach for decomposing numerical abstract domain transformers in static program analysis. Explore how researchers from ETH Zurich developed a practical method to improve performance without compromising precision across domains like Polyhedra, Octagon, and Zones. Discover their "black box" solution that eliminates the need for manual transformer decomposition while achieving approximately 2x speedup compared to existing methods. Understand the implementation details, experimental results, and implications for abstract domain designers who can now benefit from decomposition without extensive rewrites of their transformers. The 21-minute talk demonstrates how this construction advances the field of abstract interpretation and numerical domains while providing practical performance optimization techniques.
