proof complexity

Toward a Characterization of Simulation Between Arithmetic Theories

arXiv:2604.27787

summary

The paper investigates when a sound arithmetic theory can efficiently prove bounded consistency statements of its own extensions, providing constraints on such simulations and proposing a conjecture linking provability to the inability to simulate extensions, with Busy Beaver statements serving as canonical witnesses.

Abstract

We study when a sound arithmetic theory with polynomial-time decidable axioms efficiently proves the bounded consistency statements for a true sentence . Equivalently, we ask when , viewed as a proof system, simulates . The paper gives two unconditional constraints on possible characterizations. First, for finitely axiomatized sequential , if , then interprets , implying for some polynomial , and hence . Second, if fails to simulate for some true , then for all sufficiently large it also fails to simulate , where asserts the exact value of the -state Busy Beaver function. Thus any hard true extension yields a canonical Busy Beaver witness to nonsimulation. -certified simulation of a target yields , giving certification barriers rather than external lower bounds. The paper's central conjectural proposal is: for sound, finitely axiomatized sequential , if , then for every constant , . Under this proposal, hardness follows when is or a Kolmogorov-randomness axiom. The latter yields further conjectural consequences and extensions.

v2: KH and SETH-K-Finite are reformulated with the threshold EACon_SCon_{S+ϕ}, correcting the proof of HRC/KH; the no-mutual-help and density consequences are revised accordingly; figures and subsections on BB hardness and robustness are added

Topics & keywords

#proof complexity#bounded consistency#arithmetic theories#busy beaver#simulationS^1_2EACon_SBusy Beaver functionKolmogorov randomnesssequential theories