2026-08-28·by Sijie Wang#idea#math

recursively-enumerable-languages

Type 0 — recursively enumerable (unbounded tape = Turing)

Parent: chomsky-hierarchy

The top tier: unrestricted grammars = Turing machines, with unbounded tape. This is everything computable — and where undecidability lives.

  • Grammar: unrestricted (α → β, any strings).
  • Automaton: the Turing machine.
  • Class: recursively enumerable (r.e. / semi-decidable) — a machine that halts-and-accepts exactly on the members, but may run forever on non-members.
  • Examples: the halting set { ⟨M,x⟩ : M halts on x }; the valid first-order sentences (enumerate proofs); the theorems of any formal system; { ⟨p⟩ : the Diophantine equation p has a solution } (MRDP); { ⟨M⟩ : M ever prints 0 }. All are confirm-yes-only (semi-decidable) — you list members by dovetailing, but can't decide non-members.

r.e. vs recursive (the crucial split)

Recursive vs recursively enumerable

A set is recursive (decidable) if some machine halts on all inputs with the right yes/no. It is r.e. if some machine halts exactly on the yes-instances. A set is recursive ⟺ both it and its complement are r.e.

The halting set is r.e. but not recursive — you can confirm halting (wait and see) but not non-halting. That asymmetry is undecidability (history).

Closure: r.e. is closed under union, intersection, concatenation, star — but not complement (else it'd be recursive).

Rice's theorem

Every non-trivial semantic property of r.e. languages (of what a program computes, not how) is undecidable.

So at Type 0 almost nothing about behaviour is decidable — the reason you relax (certificates, tolerance, robustness) to get anything useful back.

about this entry

One of sijie's wiki entries. The AI on this site is grounded in the same corpus and answers in sijie's voice, with citations back to entries like this one — answering costs sijie money, so it waits behind a code: enter an access code →