Logic and computation as combinatorics (2306.07806v1)

Abstract: The syntactic nature of logic and computation separates them from other fields of mathematics. Nevertheless, syntax has been the only way to adequately capture the dynamics of proofs and programs such as cut-elimination, and the finiteness and the atomicity of syntax are preferable for foundational aims as seen in Hilbert's program. Another issue is that a uniform basis for logic and computation has been missing, and this problem hampers a coherent view on them. For instance, formal proofs in proof theory are far from (ordinary) proofs of the validity of a formula in model theory. Our goal is to solve these fundamental problems by rebuilding central concepts in logic and computation such as formal systems, validity (in such a way that it coincides with the existence of proofs), cut-elimination and computability uniformly in terms of finite graphs based on game semantics. Unlike game semantics, however, we do not rely on anything infinite or extrinsic to the graphs. A key idea that enables our finitary, autonomous approach is the shift from graphs in game semantics to dynamic ones. The resulting combinatorics establishes a single, syntax-free, finitary framework that recasts formal systems admitting proofs with cuts, validity, the finest computational steps of cut-elimination and higher-order computability. This subsumes fully complete semantics of intuitionistic linear logic, which solves a problem open for thirty years, and even extends the full completeness to proofs with cuts. As a byproduct, our dynamic graphs give rise to Hopf algebras, which opens up new applications of algebras to logic and computation.