Overview
This research establishes that Propositional Dynamic Logic (PDL) exhibits the Craig Interpolation Property. The demonstration addresses a long-standing open question within the field, as previous attempts at proof had been criticized or retracted. The methodology builds upon concepts originally outlined by Borzechowski in 1988.
Research Context
The Craig Interpolation Property for Propositional Dynamic Logic (PDL) has been an open problem for an extended period. Prior to this work, three proof attempts were published. However, these attempts were subsequently met with criticism in the literature or withdrawn, indicating a need for a robust and validated proof.
Approach
The proof developed in this work is fundamentally based on key ideas derived from Borzechowski's 1988 master thesis. The core of the approach involves the definition of a cyclic tableau system tailored for PDL. This system incorporates a specific 'loading mechanism' designed to recognize successful repetitions within the tableau structure.
The soundness and completeness of this cyclic tableau system were established through a game-theoretic approach. To demonstrate the interpolation property itself, the researchers adapted Maehara's method. This adaptation was necessary to enable its application to tableaux that involve repeats. The modified method proceeds by first defining 'pre-interpolants' at each individual node within the tableau. Subsequently, a 'quasi-tableau' is employed to define interpolants for 'clusters', which are identified as strongly connected components within the tableau structure. In alternative terms, this method resolves the fixpoint equations that are characteristic of the desired interpolants. The method also ensures that the solutions obtained for these equations are expressible within PDL itself.
Findings
The central finding is that Propositional Dynamic Logic (PDL) possesses the Craig Interpolation Property. This conclusion is supported by a constructive proof, and the research outlines how to compute these interpolants. An implementation of the proof system, which provides interpolants, has been made available in Haskell.
Why This Matters
The resolution of the Craig Interpolation Property for PDL addresses a long-standing open question in logic, confirming a fundamental characteristic of this specific dynamic logic. The constructive nature of the proof means that the derived interpolants can be explicitly computed, moving beyond theoretical existence to practical computability. The availability of a Haskell implementation provides a concrete tool based on the proof system, facilitating further exploration or application. Furthermore, the mention of ongoing work to formally verify this proof in Lean, an interactive theorem prover, suggests a commitment to high standards of rigor and validation for the established result.
Potential Applications
While the source does not detail specific applications, the methodology developed is constructive, providing a mechanism to compute interpolants. An implementation in Haskell is available for the proof system, which generates interpolants. This suggests a basis for practical tools. Furthermore, formal verification efforts in interactive theorem provers like Lean could enhance the reliability and trust in systems built upon this logical property.
Key Limitations Mentioned by Researchers
The source material notes 'several questions for future work' without elaborating on their nature or implying current limitations of the presented proof.