Re: Philosophy of Computation: Three seem to agree how emulating termination analyzers are supposed to work

On 11/10/2024 3:02 PM, Richard Damon wrote:

On 11/10/24 2:28 PM, olcott wrote:

*The best selling author of theory of computation textbooks*
<MIT Professor Sipser agreed to ONLY these verbatim words 10/13/2022>
If simulating halt decider H correctly simulates its input D
until H correctly determines that its simulated D would never
stop running unless aborted then

 Right, if the correct (and thus complete) emulation of this precise input would not halt.
 

That is what I have been saying for years.
(even though there cannot be such a thing
as the complete emulation of a non-terminating input).

>
H can abort its simulation of D and correctly report that D
specifies a non-halting sequence of configurations.
</MIT Professor Sipser agreed to ONLY these verbatim words 10/13/2022>

 Which your H doesn't do.
 

It is a matter of objective fact H does abort its
emulation and it does reject its input D as non-halting.
I just ran the code and it does do this.
https://github.com/plolcott/x86utm/blob/master/Halt7.c

>
Correct simulation is defined as D is emulated by H according to
the semantics of the x86 language thus includes H emulating itself
emulating D.

 And also means that it can not be aborted, as "stopping" in the middle is not to the semantics of the x86 language.
 

Every H, HH, HHH, H1, HH1, and HHH1
(a) Predicts that its input would not stop running unless aborted.
(b) Lets its input continue to run until completion.

An thus, your H fails to determine that the CORRECT emulation by H will not terminate, since it doesn't do one.
 

>
I made D simpler so that the key essence of recursive simulation
could be analyzed separately. ChatGPT totally understood this.

 Nope, your broke the rules of the field, and thus invalidates your proof.
 Either by passing the address of DDD to HHH implies passing the FULL MEMORY that DDD is in (or at least every part accessed in the emulation of DDD) and thus changed in your
 

>
void DDD()
{
HHH(DDD);
return;
}
>
ChatGPT
Simplified Analogy:
Think of HHH as a "watchdog" that steps in during real execution
to stop DDD() from running forever. But when HHH simulates DDD(),
it's analyzing an "idealized" version of DDD() where nothing stops the
recursion. In the simulation, DDD() is seen as endlessly recursive, so
HHH concludes that it would not halt without external intervention.

 But DDD doesn't call an "ideaized" verision of HHH,

     If simulating halt decider H correctly simulates its input D
     until H correctly determines that its simulated D would never
     stop running unless aborted then
     *simulated D would never stop running unless aborted*
     has ALWAYS been this idealized input.

it calls the exact function defined as HHH, s0 your arguemet is based on false premises, and thus is just a :OE/
 

>
https://chatgpt.com/share/67158ec6-3398-8011-98d1-41198baa29f2
This link is live so you can try to convince ChatGPT that its wrong.
>
On 11/3/2024 12:20 PM, Richard Damon wrote:
 > On 11/3/24 9:39 AM, olcott wrote:
 >>
 >> The finite string input to HHH specifies that HHH
 >> MUST EMULATE ITSELF emulating DDD.
 >
 > Right, and it must CORRECTLY determine what an unbounded
 > emulation of that input would do, even if its own programming
 > only lets it emulate a part of that.
 >
>
*Breaking that down into its key element*
 > [This bounded HHH] must CORRECTLY determine what
 > an unbounded emulation of that input would do...
>
When that input is unbounded that means it is never
aborted at any level, otherwise it is bounded at some
level thus not unbounded.
>

 No, because there aren't "levels" of emulation under consideration here.

There sure the Hell are.
     *simulated D would never stop running unless aborted*
     *simulated D would never stop running unless aborted*
     *simulated D would never stop running unless aborted*
Has always involved levels of simulation when
H emulates itself emulating D

Only does the emulation that the top level HHH is doing, since everything else is just fixed by the problem.
 

     *simulated D would never stop running unless aborted*
     *simulated D would never stop running unless aborted*
     *simulated D would never stop running unless aborted*
has always meant reject D
--
Copyright 2024 Olcott "Talent hits a target no one else can hit; Genius
hits a target no one else can see." Arthur Schopenhauer