|
|
|
|
|
It is easy to verify that the collection of all finite languages can be identified by a consistent scientist. Consistency appears to be a rational strategy. However, the following proposition demonstrates that consistency restricts TxtEx. |
|
|
|
|
|
|
|
|
5.6 Proposition Suppose is consistently identifiable. Then is a collection of recursive languages. |
|
|
|
|
|
|
|
|
Proof: Suppose . By the locking sequence lemma (Theorem 3.22), there is a sequence s such that , WM( s ) = L, and for each with , we have that . |
|
|
|
|
|
|
|
|
Suppose . Then , since s is a locking sequence for L. Now suppose . Then, since M is consistent, is not an index for L, and hence . Thus if and only if . Since M is total, this constitutes an effective test for membership in L. |
|
|
|
|
|
|
|
|
There are certainly that include nonrecursive languages. One such collection is { K }! Hence Proposition 5.6 yields the following corollary. |
|
|
|
|
|
|
|
|
5.7 Corollary . |
|
|
|
|
|
|
|
|
Proposition 5.6 suggests the following question: If attention is limited to the recursive languages, does consistency still restrict TxtEx? The next proposition provides an affirmative answer. |
|
|
|
|
|
|
|
|
5.8 Proposition There is an such that . |
|
|
|
|
|
|
|
|
Our proof of the proposition uses the following lemma and definition. The lemma is a standard exercise. |
|
|
|
|
|
|
|
|
5.9 Lemma For each 0-1 valued recursive function h(·, ·), there is a recursive set S which differs from each of the recursive sets where i = 0, 1, · · ·. |
|
|
|
|
|
|
|
|
5.10 Definition is said to be r.e. indexable just in case there is such that ; in this case S is said to be an r.e. index set for . Similarly, is said to be r.e. indexable just in case there is such that and S is said to be an r.e. index set for C. |
|
|
|
|
|
|
|
|
Although the recursive sets are r.e. indexable (see Rogers [158, Page 73]), Lemma 5.9 says that the collection of characteristic functions of recursive sets is not r.e. indexable. |
|
|
|
|