Alessandra Carbone,S. Semmes
Abstract:This paper is intended to provide an introduction to cut elimination which is accessible to a broad mathematical audience. Gentzen's cut elimination theorem is not as well known as it deserves to be, and it is tied to a lot of interesting mathematical structure. In particular we try to indicate some dynamical and combinatorial aspects of cut elimination, as well as its connections to complexity theory. We discuss two concrete examples where one can see the structure of short proofs with cuts, one concerning feasible numbers and the other concerning "bounded mean oscillation" from real analysis.
What problem does this paper attempt to address?
The problems that this paper attempts to explore and solve mainly focus on the concept of "cut elimination" and its complexity in logical proofs. Specifically, the paper is concerned with how to construct proofs without using "Modus Ponens", especially the impact of removing the "cut rule" in logical systems. The paper points out that although the "cut rule" can compress the length of proofs, it introduces a dynamic structure in the proof. After removing this rule, although the dynamic part of the proof is simplified, it will lead to a significant increase in the length of the proof, sometimes even an exponential increase.
The core problems of the paper can be summarized as the following aspects:
1. **Cut - Elimination Theorem**: The paper introduces Gentzen's cut - elimination theorem, that is, any proof in the LK system can be effectively converted into a proof without using the cut rule. This theorem applies not only to propositional logic but also to predicate logic. For arithmetic systems, it also holds if infinite proofs are allowed.
2. **Complexity of Proofs**: The paper discusses the relationship between the length and complexity of proofs. In particular, for some propositional tautologies, after removing the cut rule, the length of their proofs will increase from a polynomial level to an exponential level. For example, the pigeonhole principle has a polynomial - level proof when the cut rule is allowed, but requires an exponential - level proof when the cut rule is removed.
3. **Dynamic Structure**: The paper emphasizes the importance of the dynamic structure in proofs. The cut rule allows repeated substitution and calculation during the proof process, and this dynamic property helps to shorten the length of the proof. After removing the cut rule, these dynamic parts are eliminated, resulting in more lengthy proofs.
4. **Combinatorial and Geometric Perspectives**: The paper also analyzes the structure of proofs from the perspectives of combinatorics and geometry. For example, a proof can be encoded as a tree - like structure, and the proof tree - like structure after removing the cut rule is simpler but may be very large.
5. **Algorithmic Complexity**: The paper explores the relationship between the length of proofs and algorithmic complexity. In particular, the condition that all propositional tautologies have short proofs is equivalent to \( \text{NP}=\text{co - NP} \), which is an unsolved computational complexity problem.
Through the exploration of these problems, the paper aims to reveal some basic properties and challenges in logical proofs and provide a new perspective for understanding the complexity of proofs.