Table of Contents
Fetching ...

Translation of Temporal Logic for Efficient Infinite-State Reactive Synthesis (Full Version)

Philippe Heim, Rayna Dimitrova

TL;DR

This work constructs a monitor incorporating first-order and temporal reasoning at the formula level, enriching the constructed game with semantic information that leads to more efficient solving, and demonstrates that this method outperforms the state-of-the-art techniques across a range of benchmarks.

Abstract

Infinite-state reactive synthesis has attracted significant attention in recent years, which has led to the emergence of novel symbolic techniques for solving infinite-state games. Temporal logics featuring variables over infinite domains offer an expressive high-level specification language for infinite-state reactive systems. Currently, the only way to translate these temporal logics into symbolic games is by naively encoding the specification to use techniques designed for the Boolean case. An inherent limitation of this approach is that it results in games in which the semantic structure of the temporal and first-order constraints present in the formula is lost. There is a clear need for techniques that leverage this information in the translation process to speed up solving the generated games. In this work, we propose the first approach that addresses this gap. Our technique constructs a monitor incorporating first-order and temporal reasoning at the formula level, enriching the constructed game with semantic information that leads to more efficient solving. We demonstrate that thanks to this, our method outperforms the state-of-the-art techniques across a range of benchmarks.

Translation of Temporal Logic for Efficient Infinite-State Reactive Synthesis (Full Version)

TL;DR

This work constructs a monitor incorporating first-order and temporal reasoning at the formula level, enriching the constructed game with semantic information that leads to more efficient solving, and demonstrates that this method outperforms the state-of-the-art techniques across a range of benchmarks.

Abstract

Infinite-state reactive synthesis has attracted significant attention in recent years, which has led to the emergence of novel symbolic techniques for solving infinite-state games. Temporal logics featuring variables over infinite domains offer an expressive high-level specification language for infinite-state reactive systems. Currently, the only way to translate these temporal logics into symbolic games is by naively encoding the specification to use techniques designed for the Boolean case. An inherent limitation of this approach is that it results in games in which the semantic structure of the temporal and first-order constraints present in the formula is lost. There is a clear need for techniques that leverage this information in the translation process to speed up solving the generated games. In this work, we propose the first approach that addresses this gap. Our technique constructs a monitor incorporating first-order and temporal reasoning at the formula level, enriching the constructed game with semantic information that leads to more efficient solving. We demonstrate that thanks to this, our method outperforms the state-of-the-art techniques across a range of benchmarks.

Paper Structure

This paper contains 69 sections, 17 theorems, 38 equations, 4 figures, 1 table.

Key Result

theorem 1

Let $\varphi \in \mathit{RP{-}LTL}(\mathbb{X} \cup \mathbb{I} \cup \mathbb{X}')$ be a formula and $(\mathcal{G},\Lambda)$ be a symbolic game for $\varphi$ with $\mathcal{G} = (L, l_{\mathit{init}},\mathbb{I}, \mathbb{X}, \mathit{dom}, \delta)$. The formula $\varphi$ is realizable if and only if $(l_

Figures (4)

  • Figure 1: Definition of $\mathit{Closure} : \mathit{RP{-}LTL}(\mathbb{X} \cup \mathbb{I} \cup \mathbb{X}') \to 2^{\mathit{RP{-}LTL}(\mathbb{X} \cup \mathbb{I} \cup \mathbb{X}')}$.
  • Figure 2: Formula simplification rules. Each rule is given by its set of premisses and its effect on the monitor state, where $D \in \{\mathsf{A},\mathsf{G}\}$. Components of the state not assigned in the effect are not modified. by the rule.
  • Figure 3: Rules for saturating the sets $\mathit{Imp}_D$ for $D \in \{\mathsf{A},\mathsf{G}\}$. These rules modify only $\mathit{Imp}_D$.
  • Figure 4: Rules for deducing implied formulas. These rules modify only the set $\mathit{Imp}_D$ for some $D \in \{\mathsf{A},\mathsf{G}\}$.

Theorems & Definitions (27)

  • definition 1: $\mathit{RP{-}LTL}$ Syntax
  • definition 2: $\mathit{RP{-}LTL}$ Semantics
  • definition 3: $\mathit{RP{-}LTL}$ Realizability
  • definition 4: Symbolic Game Structure
  • definition 5: Semantics of Symbolic Game Structures
  • definition 6: Location-Based Winning Condition
  • theorem 1: Symbolic Game Correctness
  • definition 7: Monitor
  • definition 8: Monitor for an $\mathit{RP{-}LTL}$ Formula
  • definition 9: Game-Monitor Product
  • ...and 17 more