Reactive Synthesis Empowered by Logical Reasoning
summary
Overview of SAT-based reactive synthesis with concrete encodings and results: bounded synthesis searches for finite-state implementations up to a given size via SMT/QBF/DQBF or explicit SAT encodings and is used to produce concrete arbiters and a “patient arbiter” prototype; the bounded maximum-realizability formulation reduces to partial weighted MaxSAT (instances enter MaxSAT competition sets); Strix (explicit method) outperforms many symbolic tools on SYNTCOMP benchmarks; quantitative goals and approximations require maximum model counting and other reasoning beyond SAT to get useful, graded implementations.
AI-written summary · about
chapters (12)
- 0:00Intro
- 3:08SAT-Based Model Checking
- 5:11Bounded Model Checking for Safety Properties
- 8:20Bounded Model Checking for Unbounded Verification
- 9:41Bounded Model Checking: Impact
- 10:59Beyond Verification: Synthesis
- 14:40A Classic Example: Arbiter
- 16:42Synthesis of Reactive Systems: Theoretical Developments
- 19:30Synthesis of Reactive Systems: Tools
- 22:42Bounded Synthesis for LTL Specifications
- 27:15The Challenge of Writing Specifications
- 32:10Example: A "Patient" Arbiter
