Re: How to write a self-referencial TM?

Op 19.mei.2025 om 06:07 schreef olcott:

On 5/18/2025 10:21 PM, André G. Isaak wrote:

On 2025-05-18 16:08, olcott wrote:

On 5/18/2025 4:58 PM, André G. Isaak wrote:

>

In English, both 'description' and 'specification' can refer to something which is either complete or only partial.
>

>
Description typically means partial and
specification typically means complete.
>

>
I don't think you'll find that most people will agree with this. That might be your usage.
>
The problem is that 'specification' has already been used in much of this discussion to mean something else. A TM's specification outlines what it is that that TM is supposed to do without going into the details of how it actually does it.
>

 When you refer to the spec of an algorithm you
are correct. When you refer to the every single
step of the exact behavior that a finite string
specified you are wrong.
 

For example, the specification of a Parity Decider would be a TM takes a representation of a natural number as its initial tape content and accepts it only if it is even.
>
The description of that machine, on the other hand, would describe what the alphabet of this machine is, what it's state transitions are, etc. i.e. it would give all the information necessary to actually construct the machine.
>
André
>

 I already know how TM machine descriptions actually work.
 DDD simulated by HHH1 has the exact same
sequence of steps as the directly executed DDD().
 People here think that when DDD is simulated by
the same simulator that it calls (thus causing
recursive simulation) that DDD must have the same
behavior as DDD simulated by HHH1 that DDD does
not call.
 _DDD()
[00002172] 55         push ebp      ; housekeeping
[00002173] 8bec       mov ebp,esp   ; housekeeping
[00002175] 6872210000 push 00002172 ; push DDD
[0000217a] e853f4ffff call 000015d2 ; call HHH(DDD)
[0000217f] 83c404     add esp,+04
[00002182] 5d         pop ebp
[00002183] c3         ret
Size in bytes:(0018) [00002183]
 There is no way that DDD simulated by any HHH can
possibly reach its "ret" instruction and halt. People
have consistently ignored this basic fact for 2.5 years.
 

No it can't reach the 'ret' instruction. The only thing being ignored is that we agree that HHH fails to follow the rules of the x86 language and reach the 'ret' instruction. Open your eyes! Face the facts! The input specifies a reachable 'ret' instruction, but HHH is unable to reach it. That failure of HHH does not says anything about what the input specifies.
Olcott seems to think that it is correct for HHH to violate the x86 language by halting the simulation and ignoring what the input specifies in the part of the code that contains the abort.