Hybrid Compositional Reasoning for Reactive Synthesis from Finite-Horizon Specifications
This work addresses a scalability problem in automated reactive system synthesis for finite-horizon specifications, offering an incremental improvement over existing conversion techniques.
The paper tackles the bottleneck in LTLf synthesis, where converting formulas to deterministic finite-state automata (DFAs) is slow and memory-intensive, by proposing a hybrid representation approach that combines explicit and symbolic methods to improve performance and prevent state-space explosion, achieving faster conversion and synthesis in benchmarks.
LTLf synthesis is the automated construction of a reactive system from a high-level description, expressed in LTLf, of its finite-horizon behavior. So far, the conversion of LTLf formulas to deterministic finite-state automata (DFAs) has been identified as the primary bottleneck to the scalabity of synthesis. Recent investigations have also shown that the size of the DFA state space plays a critical role in synthesis as well. Therefore, effective resolution of the bottleneck for synthesis requires the conversion to be time and memory performant, and prevent state-space explosion. Current conversion approaches, however, which are based either on explicit-state representation or symbolic-state representation, fail to address these necessities adequately at scale: Explicit-state approaches generate minimal DFA but are slow due to expensive DFA minimization. Symbolic-state representations can be succinct, but due to the lack of DFA minimization they generate such large state spaces that even their symbolic representations cannot compensate for the blow-up. This work proposes a hybrid representation approach for the conversion. Our approach utilizes both explicit and symbolic representations of the state-space, and effectively leverages their complementary strengths. In doing so, we offer an LTLf to DFA conversion technique that addresses all three necessities, hence resolving the bottleneck. A comprehensive empirical evaluation on conversion and synthesis benchmarks supports the merits of our hybrid approach.