Liste des Groupes | Revenir à theory |
Not really. The term tuple has been accepted as the way to talk aboutOkay.
finite ordered lists (pairs, triples, etc.). The term sequence has
always been used for infinite indexed... well... sequences.
>
OK. So "over" is hiding something. M is not a map from U_M* to U_M*.My reply for this is that I started thinking of M as a skin (or a mask?)
That's messy and seems to be unnecessary. What is the problem in saying
that strings M maps? Why not simply allow M's domain to contain all the
string that it might "operate on"?
>
>Could you explain what you mean by clumsy? To be fair, I'd probably also
I was talking about the details. They look clumsy to me. What you say
below is straightforward and obvious.
>
That's fine and not all surprising. This does not require the oddI could always rework this down the line.
detail that M's domain does include all the string it might be applied
to!
>
Again, I rewrote this as I would not have liked to reply this way, but IThen we can say that S.(M(x)) satisfies the property {q_accept} and>
becomes stationary at iteration i to capture the same halting idea.
Yes, I can see you might want to do that. I suspect there will be a
simpler way to describe the properties you are interested in, but that's
not important right now.
>
Fine. By the way, you don't need to name everything:
>
s in S iff S.(R(s)) satisfies {q_accept}
s not in S iff S.(R(s)) satisfies {q_reject}
>
though you really should make it clear that {q_accept} is not what it
appears -- a set containing a state -- it's a set containing a string of
length one.
>Okay, noted.
I really would have liked to leave this point out at this stage. This is>>
1) We need to specify that the property satisfaction takes place after
the computation has become stationary.
That is not part of your definition of "satisfies" but it will, of
course, follow from the fact that the result of your iterations are TM
configurations and they always become constant when that state is either
q_accept or q_reject.
>
If did not go off into your own notation, there would be no need to
worry about this being obvious.
>
Yup, I should have left this out until we started talking specifics, but2) Also, R need not be constructed from a TM or have q_accept or>
q_reject in U_R (i.e., in the set it "recurses" over), so P1 and P2 need
only be two properties that a computation cannot both satisfy at the
same time when it is stationary*.
OK. You've lost me. I thought R (and iteration) was always a TM
configuration transition function.
>
This must be cleared up, because non-computable maps are going to take
you into all kinds of murky waters.
>I mean, I certainly hope not. At the very least, you would have educated3) And R need not act on S directly, we only need a surjective function,>
f, to map strings that R can "recurse" over, that is from (U_R)* to some
superset of S [*]
Again you've lost me. I no longer know what it means for an iteration
(what was a recursion) to decided something.
>So, to decide L_P,>
Ah, this sudden stop made me go look to see of there was another post.
And there is, and it suggests you plan to offer a different argument. I
hope I have not wasted my time here.
Les messages affichés proviennent d'usenet.