Re: [PO] Can a regular Turing Machine provide Protected Memory?

From: Jym (moyen_at_loria.fr)
Date: 08/31/04 

Date: Tue, 31 Aug 2004 00:55:15 +0200

On Mon, 30 Aug 2004, Peter Olcott wrote:

> Likewise with what you are saying. Why don't you join in
> the refutation developed by newstome@comcast.net?

Because he already said almost everything that has to be said for refuting
your screen TMs. Did you finally read the message you've printed sunday?

> You only have to derive an example of data that one
> model of my methods can not possibly process. Read

No, we only have to provide a way of building an unprocessable input from
any suposedly correct halt analyser. As you don't really provide a halt
analyser (for obvious reason), the unprocessable input is not necessarly
unique. And the same inout can very well be correctly processed on one
halt analyser and not on another. (and if one halt analyser does exist,
then infinitely many of them do exist, just add any number of useles
states in the beginning (going from u0 to u1 without moving and without
changing the tape and finally to uk to h0 the "true" initial state)).

Anyway, newstome has briantly provided such a construction and maybe it
would be wiser for us not to respond to your messages until you have
responded to his.

> the prior postings between him and me for the details.

Hypocoristiquement,
Jym.

Relevant Pages