Recursively inseparable-set proof of Tennenbaum theorem Created 2026-09-28 Updated 2026-11-05
Let be disjoint computably enumerable recursively inseparable sets, represented by existential arithmetic formulas with bounded computation witnesses and . In a non-standard model of arithmetic , the assertionholds at every standard bound . The overspill lemma gives a nonstandard bound where it still holds. This no-clash condition matters: an arbitrarily chosen nonstandard computation stage need not preserve the external disjointness of the two enumerations.
Inside , form the internally finite set and code it by the product of its corresponding prime numbers. Its standard trace contains and avoids , since the finite standard witnesses for either enumeration lie below . If the arithmetic operations of were computable, division by each standard prime would decide membership in this trace: search for and a standard remainder satisfying , where is the fixed code. The division algorithm in Peano arithmetic guarantees termination. This would give a computable set separating from , a contradiction.[ref]
Tennenbaum’s Theorem for Models of Arithmetic by Richard Kaye (2006). Internally coded separating set preceding Theorem2.8
Richard Kaye 2026-11-05
Richard Kaye works on mathematical logic and arithmetic model theory. His exposition of Tennenbaum's theorem studies why countable nonstandard arithmetic cannot have an ordinary computable presentation with computable operations.[ref]
Tennenbaum’s Theorem for Models of Arithmetic by Richard Kaye (2006). Author byline and theorem exposition
The proof encodes a standard nonrecursive set inside a nonstandard model, using recursively inseparable sets and overspill. If the model's arithmetic were computable in the required way, the coded set could be recovered algorithmically, giving a contradiction. Kaye's account makes the interaction between model theory and computability theory explicit.[ref]
Tennenbaum’s Theorem for Models of Arithmetic by Richard Kaye (2006). Construction preceding Theorem 2.8