|
|
|
|
|
5.25 Proposition (Scháfer-Richter [167], Fulk [69]) |
|
|
|
|
|
|
|
|
Proof: For each j, define and, for each j and n, define . Let M0, M1, M2, . . . denote an enumeration of all scientists. For each j, let |
|
|
|
|
|
|
|
|
Let . It is easy to see that . |
|
|
|
|
|
|
|
|
Suppose by way of contradiction that Mj is a set-driven scientist that identifies . Mj thus identifies the text . So there must be an and an index i for Lj such that Mj( s j,n) = i. In particular, there must be a least such that Mj( s j,n) = i and . But then Mj does not identify content( s j,n), since on the text must conjecture i in the limit, since Mj is set driven. Thus Mj fails to identify , a contradiction. |
|
|
|
|
|
|
|
|
§5.3.2 Rearrangement-independent Scientists |
|
|
|
|
|
|
|
|
Set-driven scientists ignore the order in which data arrive and base their conjectures only on the content of the data seen at any given time. The following definition introduces a less stringent variation on set-drivenness. |
|
|
|
|
|
|
|
|
5.26 Definition (Schäfer-Richter [167], Fulk [69, 71]) |
|
|
|
|
|
|
|
|
(a) M is rearrangement-independent just in case for every s , with and , we have . |
|
|
|
|
|
|
|
|
(b) . |
|
|
|
|
|
|
|
|
Unlike set-drivenness, rearrangement independence turns out not to be restrictive; we delay the proof of this fact to the next section. |
|
|
|
|
|
|
|
|
§5.4 Constraint on Convergence |
|
|
|
|
|
|
|
|
Suppose that a scientist M identifies . Then for each and each text T for L, M must converge to some index for L. This M does not necessarily converge to the same |
|
|
|
|