A Survey on Temporal Logics

This paper surveys main and recent studies on temporal logics in a broad sense by presenting various logic systems, dealing with various time structures, and discussing important features, such as decidability (or undecidability) results, expressiveness and proof systems.

💡 Research Summary

The surveyed paper provides a comprehensive overview of temporal logics, tracing their evolution from early linear‑time formalisms to the rich landscape of modern extensions that address real‑time, probabilistic, and hybrid system requirements. It begins by motivating the need for temporal reasoning in computer science, highlighting applications in model checking, planning, and natural‑language semantics. The authors then classify temporal logics according to the underlying time structure—discrete (ℕ), dense (ℚ), and continuous (ℝ)—and explain how each domain influences the semantics of the operators.

The core of the survey is a systematic presentation of the most influential families of logics. Linear‑time temporal logic (LTL) is described with its “next”, “until”, and “always” operators, together with the automata‑theoretic interpretation via ω‑automata that underpins efficient model‑checking algorithms. Branching‑time logics, notably CTL and CTL*, are examined for their path quantifiers (“A” for all paths, “E” for some path) and the resulting increase in expressive power, at the cost of higher computational complexity. Interval logics such as ITL and the Halpern‑Shoham (HS) logic treat time intervals as primitive, enabling direct expression of properties like “event A occurs during the first half of event B”. Metric temporal logic (MTL) and its real‑time variant (TTL) enrich the language with quantitative timing constraints, which are essential for embedded and cyber‑physical systems. The survey also covers probabilistic temporal logics (PCTL, CSL) that combine stochastic reasoning with temporal operators, and hybrid logics that integrate explicit time variables with state predicates.

A substantial portion of the paper is devoted to decidability and complexity results. The authors compile a table that juxtaposes the known decision problems for each logic: LTL model checking is PSPACE‑complete, CTL is EXPTIME‑complete, and CTL* reaches 2‑EXPTIME‑completeness. MTL is undecidable over the dense reals but becomes PSPACE‑complete when restricted to discrete time. Full HS is undecidable, yet several syntactic fragments regain decidability with complexities ranging from EXPSPACE to NEXPTIME. These results are linked to the underlying automata constructions (Büchi, Rabin, timed automata) and to reductions from classic decision problems.

The survey proceeds to discuss proof systems and tool support. Classical Hilbert‑style axiomatisations, tableau methods, and sequent calculi are presented for each logic, emphasizing their role in establishing completeness and soundness. On the practical side, the paper reviews leading model‑checking tools: SPIN for LTL, NuSMV for CTL/CTL*, UPPAAL for timed automata (hence MTL), and PRISM for probabilistic logics. The authors evaluate the scalability of these tools, noting that symbolic techniques (BDD‑based) and abstraction‑refinement loops have pushed the frontier of verifiable system sizes.

Recent research trends are highlighted, including the integration of real‑time and probabilistic aspects into a single logical framework, the development of hybrid interval‑branching logics for distributed microservice architectures, and the emergence of machine‑learning‑guided proof search that aims to automate the discovery of invariants. The paper also mentions exploratory work on quantum temporal logics, which seeks to model temporal properties of quantum circuits.

Finally, the authors identify open challenges: determining precise decidability boundaries for dense‑time interval logics, achieving complete axiomatizations for high‑order hybrid logics, and designing scalable verification algorithms that can handle the state explosion inherent in large‑scale cyber‑physical systems. They conclude that temporal logics remain a vibrant research area where theoretical advances and tool development must continue to inform each other to meet the growing demands of modern computing systems.