Taking it to the limit: on infinite variants of NP-complete problems

Abstract

Infinite, recursive versions of NP optimization problems are defined. For example, MAX CLIQUE becomes the question of whether a recursive graph contains an infinite clique. The work was motivated by trying to understand what makes some NP problems highly undecidable in the infinite case, while others remain on low levels of the arithmetical hierarchy. Two results are proved; one enables using knowledge about the infinite case to yield implications to the finite case, and the other enables implications in the other direction. Taken together, the two results provide a method for proving (finitary) problems to be outside the syntactic class MAX NP, hence outside MAX SNP too. The technique is illustrated with many examples.<>