|
|
|
|
|
The characterization of [Ex]Popperian says that the collections of functions that Popperian scientists can identify are precisely the collections of functions that are subsets of r.e. indexable collections of functions. |
|
|
|
|
|
|
|
|
5.65 Proposition (Barzdins
* and Freivalds [15], Case and Smith [35]) . |
|
|
|
|
|
|
|
|
Proof: Suppose . Fix a Popperian scientist M that identifies each function in . Let . It is easy to see that and that is r.e. indexable. |
|
|
|
|
|
|
|
|
Suppose is an r.e. indexable class of total functions. Choose a such that . Let M be a scientist that, for each , |
|
|
|
|
|
|
|
|
. |
|
|
|
|
|
|
|
|
It is easy to verify that M is Popperian and that . |
|
|
|
|
|
|
|
|
Reliable scientists were first studied by Minicozzi [132]. Reliability in the context of languages guarantees that if a scientist converges on a text for a language, then the scientist converges to an index for that language. The following formalizes reliability in the context of functions. |
|
|
|
|
|
|
|
|
5.66 Definition (Minicozzi [132]) |
|
|
|
|
|
|
|
|
(a) M is reliable on f just in case . |
|
|
|
|
|
|
|
|
(b) M is reliable on just in case M is reliable on each . |
|
|
|
|
|
|
|
|
(c) M is recursive-reliable just in case M is reliable on . |
|
|
|
|
|
|
|
|
(d) . |
|
|
|
|
|
|
|
|
The next proposition shows that recursire-reliable scientists pay a price for this useful property. |
|
|
|
|
|
|
|
|
5.67 Proposition . |
|
|
|
|
|
|
|
|
Proof: Clearly, . Below we argue that Our argument uses the following claim. |
|
|
|
|