Some strongly undecidable natural arithmetical problems, with an application to intuitionistic theories

Abstract

Although Church and Turing presented their path-breaking undecidability results immediately after their explication of effective decidability in 1936, it has been generally felt that these results do not have any direct bearing on ordinary mathematics but only contribute to logic, metamathematics and the theory of computability. Therefore it was such a celebrated achievement when Yuri Matiyasevich in 1970 demonstrated that the problem of the solvability of Diophantine equations is undecidable. His work was building essentially on the earlier work by Julia Robinson, Martin Davis and Hilary Putnam (1961), who had showed that the problem of solvability of exponential Diophantine equations is undecidable. One should note, however, that although it was only Matiyasevich's result which finally solved Hilbert's tenth problem, already the earlier result had provided a perfectly natural problem of ordinary number theory which is undecidable.Nevertheless, both the set of Diophantine equations with solutions and the set of exponential Diophantine equations with solutions are still semi-decidable, that is, recursively enumerable (i.e., Σ10); if an equation in fact has a solution, this can be eventually verified. More generally, they are — as are their complements, the sets of equations with no solutions, which are Π10, — also Trial and Error decidable (Putnam [1965]), or decidable in the limit (Shoenfield [1959]), for every Δ20 set is (and conversely). This last-mentioned natural “liberalized” notion of decidability has begun more recently to play an essential role e.g., in so-called Formal Learning Theory (see e.g., Osherson, Stob, and Weinstein [1986], Kelly [1996]).