Re: Linz's proofs and other undecidable decision problems [LP as basis] [Mike Terry]

On 2024-05-12 14:18:05 +0000, olcott said:

On 5/12/2024 2:47 AM, Mikko wrote:

On 2024-05-11 16:06:29 +0000, olcott said:
 

On 5/11/2024 3:00 AM, Mikko wrote:

On 2024-05-10 18:16:37 +0000, olcott said:
 

On 3/1/2024 12:41 PM, Mike Terry wrote:

 

Obviously a simulator has access to the internal state (tape contents etc.) of the simulated machine.  No problem there.
 What isn't allowed is the simulated machine altering its own behaviour by accessing data outside of its own state.  (I.e. accessing data from its parent simulators state.)
 While an "active-simulator" [my own term] is at liberty to combine
straight simulation with add-on "enhancements" that extend the
functionality of the simulated machine, in doing so it would no
longer be a simulator in the sense you need it to be.  So you
mustn't do this!

 In principle an incorrect simulation is permissible. However, to prove
that the result inferred from an incorrect simulation is correct may
be impossible.
 

 Within the conventional terms-of-the-art of {termination analyzer}
and {simulator} an incorrect simulation is forbidden.

 The conventional meaning of "termination analyzer" does not prohibit
incorrect simulation.

 If it does not correctly determine termination then it is not
a termination analyzer.

That is not always required. IT is often considered sufficent that
the analyzer does not determine incorrectly. To not determine at
all is often considered acceptable.
An incorrect simulation as a part of the algorithm is acceptable
as long as the result about termination is correct.
--
Mikko