Benjamin Ralph | University of Bath (original) (raw)
Uploads
Papers by Benjamin Ralph
The reduction of undecidable first-order logic to decidable propositional logic via Herbrand's th... more The reduction of undecidable first-order logic to decidable propositional logic via Herbrand's theorem has long been of interest to theoretical computer science, with the notion of a Herbrand proof motivating the definition of expansion proofs. The problem of building a natural proof system around expansion proofs, with composition of proofs and cut-free completeness, has been approached from a variety of different angles. In this paper we construct a simple deep inference system for first-order logic, KSh2, based around the notion of expansion proofs, as a starting point to developing a rich proof theory around this foundation. Translations between proofs in this system and expansion proofs are given, retaining much of the structure in each direction.
If we track atom occurrences in classical propositional proofs in deep inference, we see that the... more If we track atom occurrences in classical propositional proofs in deep inference, we see that they can form cyclic structures between cuts and identity steps. These cycles are an obstacle to a very natural form of normalisation, that simply unfolds all the contractions in a proof. This mechanism, which we call 'decomposition', has many points of contact with explicit substitutions in lambda calculi. In the presence of cycles, decomposition does not terminate, and this is an obvious drawback if we want to interpret proofs computationally. One way of eliminating cycles is eliminating cuts. However, we could ask ourselves whether it is possible to eliminate cycles independently of (general) cut elimination. This paper shows an efficient way to do so, therefore establishing the independence of decomposition from cut elimination. In other words, cut elimination in propositional logic can be separated into three separate procedures: 1) cycle elimination, 2) unfolding of contractions, 3) elimination of cuts in the linear fragment. 1998 ACM Subject Classification F.4.1 Mathematical Logic
We show that, in deep inference, there is a natural and confluent cut elimination procedure that ... more We show that, in deep inference, there is a natural and confluent cut elimination procedure that has a strikingly simple semantic justification. We proceed in two phases: we first tackle the propositional case with a construction that we call the 'experiments method'. Here we build a proof made of as many derivations as there are models of the given tautology. Then we lift the experiment method to the predicate calculus, by tracing all the existential witnesses, and in so doing we reconstruct the Herbrand theorem. The confluence of the procedure is essentially taken care of by the commutativity and associativity of disjunction.
The reduction of undecidable first-order logic to decidable propositional logic via Herbrand's th... more The reduction of undecidable first-order logic to decidable propositional logic via Herbrand's theorem has long been of interest to theoretical computer science, with the notion of a Herbrand proof motivating the definition of expansion proofs. The problem of building a natural proof system around expansion proofs, with composition of proofs and cut-free completeness, has been approached from a variety of different angles. In this paper we construct a simple deep inference system for first-order logic, KSh2, based around the notion of expansion proofs, as a starting point to developing a rich proof theory around this foundation. Translations between proofs in this system and expansion proofs are given, retaining much of the structure in each direction.
If we track atom occurrences in classical propositional proofs in deep inference, we see that the... more If we track atom occurrences in classical propositional proofs in deep inference, we see that they can form cyclic structures between cuts and identity steps. These cycles are an obstacle to a very natural form of normalisation, that simply unfolds all the contractions in a proof. This mechanism, which we call 'decomposition', has many points of contact with explicit substitutions in lambda calculi. In the presence of cycles, decomposition does not terminate, and this is an obvious drawback if we want to interpret proofs computationally. One way of eliminating cycles is eliminating cuts. However, we could ask ourselves whether it is possible to eliminate cycles independently of (general) cut elimination. This paper shows an efficient way to do so, therefore establishing the independence of decomposition from cut elimination. In other words, cut elimination in propositional logic can be separated into three separate procedures: 1) cycle elimination, 2) unfolding of contractions, 3) elimination of cuts in the linear fragment. 1998 ACM Subject Classification F.4.1 Mathematical Logic
We show that, in deep inference, there is a natural and confluent cut elimination procedure that ... more We show that, in deep inference, there is a natural and confluent cut elimination procedure that has a strikingly simple semantic justification. We proceed in two phases: we first tackle the propositional case with a construction that we call the 'experiments method'. Here we build a proof made of as many derivations as there are models of the given tautology. Then we lift the experiment method to the predicate calculus, by tracing all the existential witnesses, and in so doing we reconstruct the Herbrand theorem. The confluence of the procedure is essentially taken care of by the commutativity and associativity of disjunction.