RISC-Linz RISC-Linz Research Institute for Symbolic Computation  
  • @article{RISC5429,
    author = {David Cerna and Alexander Leitsch and Giselle Reis and Simon Wolfsteiner},
    title = {{Ceres in Intuitionistic Logic}},
    language = {english},
    abstract = {Abstract In this paper we present a procedure allowing the extension of a CERES-based cut-elimination method to intuitionistic logic. Previous results concerning this problem manage to capture fragments of intuitionistic logic, but in many essential cases structural constraints were violated during normal form construction resulting in a classical proof. The heart of the \{CERES\} method is the resolution calculus, which ignores the structural constraints of the well known intuitionistic sequent calculi. We propose, as a method of avoiding the structural violations, the generalization of resolution from the resolving of clauses to the resolving of cut-free proofs, in other words, what we call proof resolution. The result of proof resolution is a cut-free proof rather than a clause. Note that resolution on ground clauses is essentially atomic cut, thus using proof resolution to construct cut-free proofs one would need to join the two proofs together and remove the atoms which where resolved. To efficiently perform this joining (and guarantee that the resulting cut-free proof is intuitionistic) we develop the concept of proof subsumption (similar to clause subsumption) and indexed resolution, a refinement indexing atoms by their corresponding positions in the cut formula. Proof subsumption serves as a tool to prove the completeness of the new method CERES-i, and indexed resolution provides an efficient strategy for the joining of two proofs which is in general a nondeterministic search. Such a refinement is essential for any attempt to implement this method. Finally we compare the complexity of CERES-i with that of Gentzen-based methods.},
    journal = {Annals of Pure and Applied Logic},
    pages = {1783--1836},
    publisher = {Elsevier},
    isbn_issn = { ISSN 0168-0072},
    year = {2017},
    month = {October},
    refereed = {yes},
    length = {66},
    url = {http://www.sciencedirect.com/science/article/pii/S0168007217300490}